阅读说明：本技术 图像处理装置、图像变换方法 (Image processing apparatus, image conversion method, and program ) 是由 樋口晴彦 田中和彦 于 2018-06-07 设计创作，主要内容包括：本发明的图像处理装置具备：图像获取部,其获取具有第1空间分辨率的拍摄图像；距离获取部,其获取具有分辨率比第1空间分辨率低的第2空间分辨率的、为纵深信息的距离图像；图像识别部,其在拍摄图像中提取包括与立体物相对应的立体物区域在内的区域；距离算出部,其根据距离图像来算出立体物区域的纵深信息；修正部,其根据距离算出部算出的立体物区域的纵深信息来修正对拍摄图像进行坐标变换用的坐标变换信息；以及视点变换图像生成部,其使用经修正部修正后的坐标变换信息来生成对拍摄图像进行了坐标变换而得的视点变换图像。(An image processing apparatus of the present invention includes: an image acquisition section that acquires a captured image having a 1 st spatial resolution; a distance acquisition unit that acquires a distance image that is depth information and has a 2 nd spatial resolution lower than the 1 st spatial resolution; an image recognition unit that extracts a region including a three-dimensional object region corresponding to a three-dimensional object from a captured image; a distance calculation unit that calculates depth information of the three-dimensional object region from the distance image; a correction unit that corrects coordinate conversion information for coordinate conversion of the captured image, based on the depth information of the three-dimensional object region calculated by the distance calculation unit; and a viewpoint conversion image generation unit that generates a viewpoint conversion image obtained by coordinate conversion of the captured image, using the coordinate conversion information corrected by the correction unit.)

1. An image processing apparatus is characterized by comprising:

an image acquisition section that acquires a captured image having a 1 st spatial resolution;

a distance acquisition unit that acquires a distance image that is depth information and has a 2 nd spatial resolution lower than the 1 st spatial resolution;

an image recognition unit that extracts a region including a three-dimensional object region corresponding to a three-dimensional object from the captured image;

a distance calculation unit that calculates depth information of the three-dimensional object region from the distance image;

a correction unit that corrects coordinate conversion information for coordinate conversion of the captured image, based on the depth information of the three-dimensional object region calculated by the distance calculation unit; and

and a viewpoint-converted image generating unit that generates a viewpoint-converted image obtained by coordinate-converting the captured image, using the coordinate-converted information corrected by the correcting unit.

2. The image processing apparatus according to claim 1,

the image recognition unit extracts contours of a plurality of regions including the three-dimensional region according to at least 1 of brightness, lightness, saturation, and lightness of the captured image.

3. The image processing apparatus according to claim 1,

the coordinate transformation information includes a combination of transformation source coordinates on a plurality of the photographed images and transformation destination coordinates on the viewpoint transformation image,

the correction unit corrects the conversion destination coordinate based on depth information of the three-dimensional object region.

4. The image processing apparatus according to claim 3,

the coordinate conversion information is previously created on the premise that the object in the captured image is an area on the road surface,

the image recognizing section divides the captured image into a plurality of regions for each object contained in the captured image,

the correction unit determines whether or not each of the plurality of regions is the three-dimensional object region based on the distance image, and corrects the conversion destination coordinates corresponding to the conversion source coordinates in the region determined to be the three-dimensional object region.

5. The image processing apparatus according to claim 1,

the coordinate transformation information includes a combination of transformation source coordinates on a plurality of the photographed images and transformation destination coordinates on the viewpoint transformation image,

the correction unit corrects the conversion source coordinates in the vicinity of the outline of the three-dimensional object region.

6. The image processing apparatus according to claim 5,

the correction unit arranges the conversion source coordinates in the vicinity of the outline of the three-dimensional object region and outside and inside the three-dimensional object region.

7. The image processing apparatus according to claim 1,

the coordinate conversion information is calculated on the premise that the object of the captured image is present in a reference height plane,

the correction unit calculates a height of the object from the reference height plane and a position of the object on the reference height plane using the depth information of the three-dimensional object region calculated by the distance calculation unit,

the correction unit corrects the coordinate conversion information using a reference position of viewpoint conversion, the calculated height of the object from the reference height plane, and the calculated position of the object on the reference height plane.

8. An image conversion method comprising the steps of:

acquiring a photographed image having a 1 st spatial resolution;

acquiring a distance image which is depth information and has a 2 nd spatial resolution lower than the 1 st spatial resolution;

extracting a region including a three-dimensional object region corresponding to a three-dimensional object in the captured image;

calculating depth information of the three-dimensional object region from the distance image;

correcting coordinate conversion information for coordinate conversion of the captured image based on the calculated depth information of the three-dimensional object region; and

generating a viewpoint conversion image obtained by performing coordinate conversion on the captured image using the coordinate conversion information corrected by the correction.

9. The image transformation method according to claim 8,

extraction of a region including the three-dimensional object region is performed according to at least 1 of brightness, lightness, chroma, and lightness of the captured image.

10. The image transformation method according to claim 8,

the coordinate transformation information includes a combination of transformation source coordinates on a plurality of the photographed images and transformation destination coordinates on the viewpoint transformation image,

and correcting the transformation destination coordinate according to depth information in the three-dimensional object region.

11. The image conversion method according to claim 10,

the coordinate conversion information is previously created on the premise that the object in the captured image is an area on the road surface,

dividing the captured image into a plurality of regions for each object contained in the captured image,

and determining whether or not each of the plurality of regions is the three-dimensional object region based on the distance image, and correcting the conversion destination coordinates corresponding to the conversion source coordinates in the region determined as the three-dimensional object region.

12. The image transformation method according to claim 8,

the coordinate transformation information includes a combination of transformation source coordinates on a plurality of the photographed images and transformation destination coordinates on the viewpoint transformation image,

the correction based on the depth information of the three-dimensional object region includes a step of correcting the transformation source coordinates in the vicinity of the outline of the three-dimensional object region.

13. The image conversion method according to claim 12,

the correction based on the depth information of the three-dimensional object region includes a step of arranging the conversion source coordinates in the vicinity of the outline of the three-dimensional object region and outside and inside the three-dimensional object region.

14. The image transformation method according to claim 8,

the coordinate conversion information is calculated on the premise that the object of the captured image is present in a reference height plane,

and calculating a height of the subject from the reference height plane and a position of the subject on the reference height plane using the calculated depth information of the three-dimensional object region, and correcting the coordinate conversion information using a reference position of viewpoint conversion, the calculated height of the subject from the reference height plane, and the calculated position of the subject on the reference height plane.

Technical Field

The present invention relates to an image processing apparatus and an image conversion method.

Background

There is known an image display system that images the surroundings of a vehicle with a camera provided in the vehicle and displays the captured image in the vehicle. By using such an image display system, the driver can confirm the situation around the vehicle with high visibility. Patent document 1 discloses a display device for a vehicle, which includes: a captured image acquisition unit that acquires data of a captured image captured by a camera mounted on a vehicle; a viewpoint setting unit that sets a virtual viewpoint of an image to be displayed; a coordinate system transformation unit that transforms at least a part of a reference curved surface coordinate system having a curved surface, which is defined in advance for projecting the captured image, in accordance with a position of the virtual viewpoint with respect to the reference curved surface coordinate system to obtain a transformed coordinate system; a projection unit configured to project data of the captured image onto the deformed coordinate system obtained by the deformation, and generate an image for observing the vehicle and the surroundings of the vehicle from the set virtual viewpoint; and a display control unit that displays the generated video on a display screen.

Disclosure of Invention

Problems to be solved by the invention

In the invention described in patent document 1, when a three-dimensional object is included in an image obtained by imaging, image distortion occurs in the three-dimensional object in the viewpoint conversion image. A method of displaying an image in which a distance sensor is used to reduce distortion is widely known, but the spatial resolution of the distance sensor is generally closely related to image distortion caused by a three-dimensional object in a viewpoint-converted image. That is, it is difficult to generate a highly accurate viewpoint conversion image using distance information having a low spatial resolution.

Means for solving the problems

An image processing apparatus according to aspect 1 of the present invention includes: an image acquisition section that acquires a captured image having a 1 st spatial resolution; a distance acquisition unit that acquires a distance image that is depth information and has a 2 nd spatial resolution lower than the 1 st spatial resolution; an image recognition unit that extracts a region including a three-dimensional object region corresponding to a three-dimensional object from the captured image; a distance calculation unit that calculates depth information of the three-dimensional object region from the distance image; a correction unit that corrects coordinate conversion information for coordinate conversion of the captured image, based on the depth information of the three-dimensional object region calculated by the distance calculation unit; and a viewpoint conversion image generation unit that generates a viewpoint conversion image obtained by performing coordinate conversion on the captured image, using the coordinate conversion information corrected by the correction unit.

An image conversion method according to claim 2 of the present invention includes: acquiring a photographed image having a 1 st spatial resolution; acquiring a distance image which is depth information and has a 2 nd spatial resolution lower than the 1 st spatial resolution; extracting a region including a three-dimensional object region corresponding to a three-dimensional object in the captured image; calculating depth information of the three-dimensional object region from the distance image; correcting coordinate conversion information for coordinate conversion of the captured image based on the calculated depth information of the three-dimensional object region; and generating a viewpoint conversion image obtained by performing coordinate conversion on the captured image using the coordinate conversion information corrected by the correction.

ADVANTAGEOUS EFFECTS OF INVENTION

According to the present invention, a highly accurate viewpoint conversion image can be generated using distance information with low spatial resolution.

Drawings

Fig. 1 is a hardware configuration diagram of an image processing apparatus 100.

Fig. 2 (a) is a conceptual diagram of a distance image obtained by the distance detection unit 14, fig. 2 (b) is a conceptual diagram of a distance image obtained using a distance sensor having an ideal high spatial resolution, and fig. 2 (c) is a diagram showing a relationship between shading and distance.

Fig. 3 is a functional block diagram of the image processing apparatus 100 according to embodiment 1.

Fig. 4 is a top view of vehicle 20 on which image processing apparatus 100 is mounted.

Fig. 5 is a side view of the vehicle 20.

Fig. 6 is a diagram showing definitions of a camera coordinate system, a virtual viewpoint coordinate system, and a world coordinate system.

Fig. 7 is a diagram showing an example of table data 121.

Fig. 8 is a diagram illustrating an example of correction of table data 121 according to embodiment 1.

Fig. 9 (a) is a diagram showing a captured image 900 of the front camera 10A, fig. 9 (b) is a diagram showing a distance image 910 acquired by the distance detection unit 14, and fig. 9 (c) is a diagram showing a distance image 920 acquired by an ideal distance sensor.

Fig. 10 (a) is an enlarged view of the vicinity of symbol 911 in fig. 9 (b) and superimposed with the outline of tree 21, fig. 10 (b) is a view showing the boundary of regions of different distances in fig. 10 (a), and fig. 10 (c) is a view of the coordinate-corresponding pixel superimposed with table data 121 in fig. 10 (b).

Fig. 11 (a) is a diagram showing an example of segment division, and fig. 11 (b) is a diagram in which segments are denoted by reference numerals.

Fig. 12 is a flowchart showing the operation of the image processing apparatus 100 according to embodiment 1.

Fig. 13 is a diagram showing an example of table data 121A in modification 1.

Fig. 14 (a) is a diagram showing a case where coordinate corresponding points are set sparsely in the outline, fig. 14 (b) is a diagram showing ideal depth information between PQs in fig. 14 (a), and fig. 14 (c) is a diagram showing depth information between PQs in fig. 14 (a).

Fig. 15 (a) is a diagram showing coordinate corresponding points corrected by the table correction unit 113 according to embodiment 2, and fig. 15 (b) is a diagram showing depth information between PQs in fig. 15 (a).

Fig. 16 is a functional block diagram of the image processing apparatus 100 according to embodiment 2.

Fig. 17 is a diagram illustrating an example of correction of table data 121 according to embodiment 2.

Fig. 18 is a flowchart showing the operation of the image processing apparatus 100 according to embodiment 2.

Detailed Description

(embodiment 1)

Next, embodiment 1 of the image processing apparatus will be described with reference to fig. 1 to 12.

Fig. 1 is a hardware configuration diagram of an image processing apparatus 100 mounted on a vehicle 20. The vehicle 20 includes an image processing device 100, a front camera 10A, a left camera 10B, a right camera 10C, a rear camera 10D, a display unit 13, and a distance detection unit 14. Hereinafter, the front camera 10A, the left camera 10B, the right camera 10C, and the rear camera 10D are collectively referred to as the cameras 10. The image captured by the camera 10 is referred to as a captured image. The resolution of the captured image is determined by the number of image pickup devices incorporated in the camera 10, and has a resolution of 1920 × 1080, for example. Hereinafter, this resolution is also referred to as a spatial resolution.

The camera 10 operates in synchronization with the distance detection unit 14, and acquires surrounding information at the same time. The image processing apparatus 100 uses the image captured by the camera 10 to output an image (hereinafter referred to as a viewpoint conversion image) obtained when a virtual camera is placed at a virtual position (hereinafter referred to as a virtual viewpoint) different from the position where the camera 10 is placed to capture the surroundings of the vehicle 20, to the display unit 13. The mounting position and mounting posture of the camera 10 on the vehicle 20 are known and stored in the storage section 104.

The display unit 13 is, for example, an LCD display, a projector, or a display unit of a car navigation device mounted on the vehicle 20. The display unit 13 displays information output from the interface 105.

The distance detection unit 14 detects depth information of an object around the vehicle as information having a two-dimensional resolution in the horizontal direction and the vertical direction. The distance Detection unit 14 is, for example, lidar (light Detection and ranging). Hereinafter, the distance information having a two-dimensional resolution acquired by the distance detection unit 14 is referred to as a "distance image". The field of view of the distance detection unit 14 overlaps with at least a part of the camera 10. In the present embodiment, the field of view of the front camera 10A and the field of view of the distance detection unit 14 are identical. However, the resolution of the distance image acquired by the distance detection unit 14 is lower than the resolution of the camera 10. For example, when the camera 10 has a resolution of 1920 × 1080 in the horizontal direction × the vertical direction, the distance detection unit 14 has a resolution of 25 × 25. That is, when the spatial resolution of the captured image is referred to as the 1 st spatial resolution and the spatial resolution of the range image is referred to as the 2 nd spatial resolution, the 2 nd spatial resolution is lower than the 1 st spatial resolution. Hereinafter, the resolution in the depth direction in the range image is referred to as "range resolution", and is distinguished from the spatial resolution. The mounting position and mounting posture of the distance detection unit 14 on the vehicle 20 are known and stored in the storage unit 104.

Fig. 2 (a) is a conceptual diagram of a distance image obtained by the distance detection unit 14, fig. 2 (b) is a conceptual diagram of a distance image obtained using a distance sensor having an ideal high spatial resolution, and fig. 2 (c) is a diagram showing a relationship between shading and distance in fig. 2 (a) and 2 (b). The high-resolution distance sensor for capturing an image of fig. 2 (b) is shown for illustrative purposes and is not mounted on the vehicle 20 described above. Fig. 2 (a) and 2 (b) are conceptual diagrams of distance images obtained from the distance detection unit 14 and an ideal distance sensor provided on the ground without an obstacle and directed toward the horizon. That is, the lower part of the figure shows a case where the distance is close because of being near, and the upper part of the figure shows a case where the distance is far for the horizon or sky.

Fig. 2 (a) and 2 (b) are black at the upper end and white at the lower end. But the concentration varies differently. In fig. 2 (a), the concentration changes stepwise as shown in the right graph, and in fig. 2 (b), the concentration changes continuously. This is because the spatial resolution of the device that acquires the information in fig. 2 (b) is high. As described above, although the sensor having a low spatial resolution is used in the present embodiment, the result similar to the case of using the sensor having a high spatial resolution shown in fig. 2 (b) is obtained by the drilling described later. The description is continued returning to fig. 1.

The image processing apparatus 100 includes a CPU101, a ROM102, a RAM103, a storage unit 104, and an interface 105. The CPU101 is a central processing unit, and develops functions to be described later by expanding and executing programs stored in the ROM102 in the RAM 103. The storage unit 104 is a nonvolatile storage device, such as a flash memory or a hard disk drive. The interface 105 is an entrance/exit for information with other devices in the image processing apparatus 100, and the information input to the interface 105 is output to the CPU 101. The image processing apparatus 100 acquires the captured image obtained by the camera 10 and the distance image obtained by the distance detection unit 14 using the interface 105, and inputs the images to the CPU 101. The interface 105 is a serial port or the like, and may include an AD converter or the like.

Fig. 3 is a functional block diagram showing functions provided in the image processing apparatus 100 in the form of functional blocks. The image processing apparatus 100 includes an image recognition unit 111, a distance correction unit 112, a table correction unit 113, a correlation unit 114, an image conversion unit 115, and a display control unit 117. Table data 121 shown in fig. 2 is information stored in the storage unit 104. The table data 121 is a lookup table used when generating a viewpoint conversion image such as a bird's-eye view image or a bird's-eye view image. The table data 121 is created in advance in accordance with the position, angle, and imaging conditions of the camera 10 and the virtual viewpoint provided in the vehicle 20 at the stage of designing the system including the image processing apparatus 100. The table data 121 exists for each viewpoint for which the cameras 10 perform viewpoint conversion, but the following description will be made only for one of 1 virtual viewpoints of one of the cameras 10A.

The image recognition unit 111 performs a segmentation process of extracting a contour of each object included in the captured image and dividing the captured image into a plurality of regions, with the captured image of the camera 10 as a processing target. In the following description, each region set in the captured image by the segmentation process performed by the image recognition unit 111 is referred to as a "segment". The contour of the subject in the present processing may be extracted by a method based on known contour detection or a method of analyzing color information of a captured image, dividing the captured image into a plurality of regions according to similarity of brightness, hue, chroma, and lightness, and extracting the contour of each region. When a section corresponding to the three-dimensional object, that is, a region in which the object is a three-dimensional object, exists in the recognition result of the image recognition unit 111, the distance correction unit 112 corrects the distance, which is the measurement value of the distance detection unit 14, for each coordinate corresponding point described later in relation to the section. The distance corrected by the distance correcting unit 112 will be referred to as a "three-dimensional object distance" hereinafter. The table correction unit 113 overwrites the table data 121 with the three-dimensional object distance calculated by the distance correction unit 112.

The association unit 114 associates the captured image with the distance image based on the mounting position and mounting posture of the camera 10 and the distance detection unit 14 stored in the storage unit 104 on the vehicle 20. For example, the association unit 114 calculates a region in the distance image corresponding to a certain region in the captured image. In the present embodiment, as described above, the camera 10A and the distance detection unit 14 have the same field of view. Therefore, the correlation between the captured image of the camera 10A and the distance image can be easily performed. The image conversion unit 115 converts the captured images of the cameras 10 using the table data 121 rewritten by the table correction unit 113, and combines the captured images of the cameras 10 to generate a viewpoint conversion image. The display control unit 117 outputs the viewpoint-converted image generated by the image conversion unit 115 to the display unit 13 and displays the image.

(action Environment)

Fig. 4 and 5 are diagrams illustrating an environment in which the image processing apparatus 100 operates. Fig. 4 is a view of vehicle 20 with image processing apparatus 100 mounted thereon, as viewed from above, and fig. 5 is a view of vehicle 20 as viewed from the side. The vehicle 20 is present in a region between a straight line LL and a straight line LR which are white lines indicating lanes. A mark 22 is provided on the ground in front of the vehicle 20, and trees, i.e., a solid object 21, are provided in front of the vehicle 20 on the left.

The front camera 10A is disposed in the front of the vehicle 20 with its optical axis directed toward the road surface in front of the vehicle 20, and captures the tree 21, which is a solid object, and the mark 22 on the road surface within its shooting range. Similarly, a left camera 10B, a right camera 10C, and a rear camera 10D are provided on the left, right, and rear portions of the vehicle 20, respectively, with their optical axes directed to the left, right, and rear road surfaces of the vehicle 20, respectively. The cameras 10 include wide-angle lenses and have respective angles of view of about 180 degrees. The installation positions and installation angles of the camera 10 and the distance detection unit 14 are known and predetermined in the design stage of the vehicle 20.

The virtual viewpoint 25 shown in fig. 5 is a viewpoint that is directly below the vehicle 20 and overlooks the front of the vehicle 20. Next, a method of creating an image obtained from the virtual viewpoint 25 in the situation shown in fig. 5 will be described. The image processing apparatus 100 includes the cameras 10A to 10D, but the description will be made by taking a case where the captured image of the camera 10A before conversion is typically performed.

A method of creating an image obtained from the virtual viewpoint 25 will be described.

(coordinate transformation)

Fig. 6 is a diagram showing definitions of a camera coordinate system, a virtual viewpoint coordinate system, and a world coordinate system. The camera coordinate system is a coordinate system based on a camera that captures an image. FIG. 6 shows X as the 3-axis of the camera coordinate system R with reference to the front camera 10A_r、Y_r、Z_r. The virtual viewpoint coordinate system is a coordinate system based on a predetermined virtual viewpoint. FIG. 6 shows X as the 3-axis of a virtual viewpoint coordinate system V with reference to the virtual viewpoint 25_v、Y_v、Z_v. The world coordinate system is a coordinate system set with reference to a road surface on which the vehicle 20 travels. FIG. 6 shows the 3-axis X of the world coordinate system W_w、Y_w、Z_w。X_wAnd Y_wParallel to the road surface, the road surface being comprised in Z _w0 in the base height plane 230.

Z as the 1-axis of the camera coordinate system R_rIn line with the optical axis of the front camera 10A, i.e., in line withThe imaging elements being orthogonal, the other 2-axis being X_rAnd Y_rParallel to the long and short sides of the image pickup element of the front camera 10A. If the focal length z of the camera is used_rTo indicate that the position of each pixel constituting the captured image 301 is located at Z_r＝z_rX of (2)_rY_rCoordinate data on a plane. That is, the camera coordinate system R is equal to the coordinate system of the captured image 301.

Z as 1 axis of the virtual viewpoint coordinate system_vCoincides with the optical axis of a virtual camera placed at the virtual viewpoint 25, i.e., is orthogonal to the virtual image pickup element, and has 2 other axes, i.e., X_vAnd Y_vParallel to the long and short sides of the virtual image sensor. If the focal length z of the camera placed at the imaginary viewpoint 25 is used_vThen, the position of each pixel constituting the viewpoint conversion image 311 is located at Z_v＝z_vX of (2)_vY_vCoordinate data on a plane. That is, the virtual viewpoint coordinate system V is equal to the coordinate system of the viewpoint conversion image 311.

In the world coordinate system W, a certain point P is referred to as P_wIts coordinates are expressed as (x)_w,y_w,z_w). Will take a picture of the point P with the front camera 10A_wThe point P in the captured image at that time is called P_rWill point P_rIs expressed as (x)_r,y_r,z_r). The point P in the image obtained from the virtual viewpoint 25 is referred to as P_vA 1 is to P_vIs expressed as (x)_v,y_v,z_v)。

To put a point P in the world coordinate system W_wCoordinate (x) of_w,y_w,z_w) Transforming to a point P of the camera coordinate system R_rCoordinate (x) of_r,y_r,z_r) For example, affine transformation as shown in formula (1) is used.

[ numerical formula 1]

Here, Mr is a 4 × 4 perspective projection transformation matrix as shown in equation (2).

[ numerical formula 2]

In the formula (2), R_rIs a 3 × 3 rotation matrix, T_rIs a 1 × 3 translation matrix and 0 is a 3 × 1 zero matrix. Rotation matrix R_rAnd translation matrix T_rThe position and angle of installation of the camera 10A on the world coordinate system, the focal length, which is an internal parameter of the camera 10A, and the effective pixel size of the imaging device are calculated by a known method.

In addition, point P of world coordinate system W is to be identified_wCoordinate (x) of_w,y_w,z_w) Transformed into a point P of a virtual viewpoint coordinate system V_vCoordinate (x) of_v,y_v,z_v) For example, affine transformation as shown in formula (3) is used.

[ numerical formula 3]

Here, M_vIs a 4 × 4 perspective projection transformation matrix as shown in equation (4).

[ numerical formula 4]

In the formula (4), R_vIs a 3 × 3 rotation matrix, T_vIs a 1 × 3 translation matrix and 0 is a 3 × 1 zero matrix. Rotation matrix R_vAnd translation matrix T_vThe position and angle of the virtual viewpoint 25 on the world coordinate system, the virtual focal length of the virtual viewpoint 25, the effective pixel size of the imaging device, and the like are calculated by a known method.

When the above-described equations (1) and (3) are combined, a point P for converting the camera coordinate system R is obtained_rIs transformed into a point P of a virtual viewpoint coordinate system V_vEquation (5) of the coordinates of (a).

[ numerical formula 5]

In the formula (5), the matrix M is transformed by perspective projection_rThe inverse matrix of (a) will be the point P of the camera coordinate system R_rIs transformed into a point P of the world coordinate system_wBy transforming the matrix M by perspective projection_vWill be the point P_wIs transformed into a point P of a virtual viewpoint coordinate system V_vCoordinate (x) of_v,y_v,z_v). Using the coordinate transformation result of equation (5), it is possible to obtain the point P from the corresponding captured image 301_rCalculates a point P of the viewpoint converted image 311 from the pixel values of_vThe pixel value of (2).

However, the distance information from the object is not obtained from the captured image of the camera 10, and therefore, for example, the point P can be assumed_wOn the road surface, i.e. z_wCalculating a point P of the viewpoint conversion image 311 on a plane of 0_vThe pixel value of (2). Then, the point P of the viewpoint conversion image 311 is recalculated only for the region where the subject not on the road surface is captured_vThe pixel value of (2). For example, the point Pr in the captured image 301 is not the point P on the road surface_wBut is present in Z_w＝z_w1Point P on plane of (≠ 0)_w1In the case of (1), the corresponding point in the viewpoint conversion image 311 is not the point P_vBut point P_v1. Whether or not an object of a captured image of the camera 10 is present on the road surface can be determined from the distance to the object. Further, in the case where the subject is not present on the road surface, the corresponding point in the viewpoint conversion image 311 can be calculated by using the distance information thereof.

To describe an example of calculation of the corresponding point with reference to fig. 6, first, the point P is calculated using the geometric relationship between the origin position of the camera coordinate system R and the coordinates of the point Pr in the world coordinate system and the distance information to the subject calculated by the distance correction unit 112_w1The coordinates of (a). That is, the point P is calculated_w1Position on the reference height plane, i.e. point P_w1And the X and Y coordinates of the subject from the reference height plane 230Height, i.e. point P_w1Z coordinate of (2)_w1The value of (c). Then, based on the position of the camera 25 and the calculated point P_w1Calculates the corresponding point P in the viewpoint conversion image 311_v1To fix the table data 121.

(Table data 121)

Each table data 121 stored in the storage unit 104 describes a plurality of sets of points P on the captured image calculated in advance on the assumption that all the objects are present on the road surface_rTransforming point P on image with viewpoint_vThe corresponding relationship of (1). In other words, the table data 121 is calculated on the premise that the object exists in the reference height plane 230. That is, the predetermined point P of the camera coordinate system R is set_r1Coordinate (x) of_r1,y_r1)、P_r2Coordinate (x) of_r2,y_r2) And (c) are obtained by transforming the coordinates of the corresponding points in the virtual viewpoint coordinate system V by the above equation (5). Here, the correspondence relationship between corresponding points, that is, pixels in the 2 coordinate systems is referred to as coordinate correspondence information, and the coordinate correspondence information is created in the form of table data 121. In the table data 121, the focal length and the like of the camera 10A are regarded as fixed, and Z is omitted_rInformation of the coordinates.

In the following description, a pixel in which coordinate correspondence information is stored in the table data 121, out of the pixels of the captured image 301 and the viewpoint conversion image 311, is referred to as a coordinate-corresponding pixel or a coordinate-corresponding point. That is, a plurality of coordinate corresponding points are set in advance in the captured image 301 and the viewpoint conversion image 311. By storing the table data 121 in the storage unit 104 in advance and referring to the viewpoint conversion image 311 when it is created, the number of calculations of the above equation (5) can be reduced and the processing time for coordinate conversion can be shortened. The amount of data of the table data 121 increases as the coordinate correspondence information stored in advance in the table data 121 increases. To reduce the amount of data in the table data 121, coordinate-corresponding information is stored in advance only for some pixels of the captured image 301, and other pixels are interpolated to calculate the point P_vThe pixel value of (2). The table data 121 may be added in consideration of distortion of the lens of the camera 10 and the likeTo produce.

As described above, the table data 121 is calculated in advance on the assumption that all the subjects exist on the road surface. Therefore, when the subject is not present on the road surface, that is, a three-dimensional object having a height, it is necessary to calculate from the distance information and rewrite the table data 121. Hereinafter, rewriting of the table data 121 is also referred to as correction of the table data 121. In the present embodiment, the table data 121 is corrected by the table correcting unit 113. That is, the table correction unit 113 corrects the coordinate corresponding points included in the region of the three-dimensional object 21 in the table data 121. Hereinafter, the coordinate corresponding point to be corrected is referred to as a "correction target coordinate corresponding point".

Fig. 7 is a diagram showing an example of table data 121. The table data 121 is a coordinate correspondence table defining a correspondence relationship between coordinates of discrete pixels of the captured image 301 and coordinates of pixels of the viewpoint conversion image 311 corresponding to the coordinates. Fig. 7 shows coordinate correspondence information of the pixels having correspondence numbers 1, 2, …, and n. The image processing apparatus 100 calculates the coordinates of the corresponding pixel of the viewpoint conversion image 311 with reference to the table data 121 for each coordinate of each pixel of the captured image 301.

Fig. 8 is a diagram showing coordinates before and after correction by extracting a correction target coordinate corresponding point from the table data 121. The three-dimensional object distance required for correction of the table data 121 is also described. That is, the table shown in fig. 8 is merely an example for explanation, and the table itself is not necessarily stored in the RAM103 or the storage unit 104. In the present embodiment, the table correction unit 113 corrects the table data 121 by correcting the coordinate correspondence information on the correction target coordinate correspondence point among the coordinate correspondence information on each coordinate correspondence point of the table data 121. Specifically, the coordinates of the correction object coordinate corresponding points o1, o2, o3, ·, oi on the viewpoint conversion image, that is, the coordinates in the virtual viewpoint coordinate system V, i.e., (xv _ o1, yv _ o1), (xv _ o2, yv _ o2), (xv _ o3, yv _ o3), ·, (xv _ oi, yv _ oi) are corrected to (xv _ o1', yv _ o1'), (xv _ o2', yv _ o2'), (xv _ o3', yv _ o3'), ·, (xv _ oi ', yv _ oi') respectively. As described above, the table correction unit 113 performs the correction using the three-dimensional object distance.

If a three-dimensional object is captured in the entire region of the captured image of the camera 10A, the coordinates on the viewpoint conversion image are rewritten for all the coordinate corresponding points of the table data 121. However, in this case, the coordinates on the captured image in the table data 121 are not rewritten.

(necessity of distance image and correction)

Fig. 9 (a) is a diagram showing a captured image 900 of the front camera 10A, fig. 9 (b) is a diagram showing a distance image 910 acquired by the distance detection unit 14, and fig. 9 (c) is a diagram showing a distance image 920 acquired by an ideal distance sensor. As shown in fig. 9 (a), in the captured image 900, the tree 21 having a three-dimensional object on the left side is shown, and the white line LL and the white line LR extend from the lower side to the upper side of the center of the drawing. The lower part of the figure is shown with a mark 22 and the upper part with a horizon 35. As shown in fig. 2, the distance image 910 acquired by the distance detection unit 14 changes its distance to the ground in a stepwise manner, and the trees 21 are shown in a cross shape as indicated by a reference numeral 911. This is because the spatial resolution of the distance detection unit 14 is low. Assuming that an ideal distance sensor is used in the same environment, a distance image 920 shown in (c) of fig. 9 is obtained. In the distance image 920, the tree 21 is represented so as to match the shape of the tree 21 as indicated by a symbol 921.

Fig. 10 (a) is an enlarged view of the vicinity of symbol 911 of fig. 9 (b) and superimposed with the outline of the tree 21 indicated by symbol 21A. Fig. 10 (b) is a diagram showing the boundaries of the regions having different distances in fig. 10 (a). Fig. 10 (c) is a diagram in which the coordinate-corresponding pixels of the table data 121 are superimposed on fig. 10 (b). Here, the coordinate-corresponding pixels in (c) of fig. 10 are shown in 3 of a cross mark, a circle mark, and a triangle mark. The cross mark indicates a point at which appropriate distance information is obtained, the circular mark indicates a point farther than an appropriate distance, and the triangular mark indicates a point closer than an appropriate distance. For example, the circular mark shown by reference numeral 912 is a part of the tree 21, and thus should be recognized as being approximately the same distance as the other parts of the tree 21.

However, as is clear with reference to fig. 10 (a), the circular mark 912 is recognized to be at a distance farther than the tree 21. Further, referring to fig. 10 (a) and 8, the point of the triangular mark shown by symbol 913 is not the tree 21 but a point in the ground surface, and should be recognized as being farther than the tree 21. However, the triangular marker 913 is recognized as being at the same distance as the tree 21. Therefore, the distance correction unit 112 corrects the distance information as shown in fig. 10 (d) by the processing described later. That is, since the distance correction unit 112 calculates the three-dimensional object distance shown in fig. 8, the table correction unit 113 corrects the table data 121 using the three-dimensional object distance. This causes appropriate conversion to be performed on all the coordinate-corresponding pixels of the table data 121.

(example of operation of image recognition unit)

An operation example of the image recognition unit 111 will be described with reference to fig. 9 (a) and 11. The image recognition unit 111 divides the captured image of the camera 10A shown in fig. 9 (a) into a plurality of regions, that is, segments, by the segment division processing based on the method using color information and the like described above, with the captured image as a processing target. In fig. 11 (a), tree 21 is recognized as 1 area as indicated by reference numeral 1101, and mark 22 is recognized as 4 areas as indicated by reference numeral 1102. Fig. 11 (b) is a diagram in which each segment shown in fig. 11 (a) is labeled with a reference numeral in a convenient manner. As shown in fig. 11 (b), the captured image is divided into 9 segments a to I by the image recognition unit 111. However, in the segment division, the three-dimensional object is not distinguished from the road surface.

(operation of distance correcting unit)

The distance correction unit 112 corrects the distance of the coordinate corresponding point in each segment by any one of the following 3 methods. The 1 st method is simple averaging. The distance correction unit 112 calculates an average value of the distance information of all the coordinate-corresponding points in the block, and sets the average value as the three-dimensional object distance of all the coordinate-corresponding points in the block to be processed. That is, according to the method 1, all the coordinate corresponding points in the segment have the same three-dimensional object distance.

The 2 nd method is a 1 st approximation. The distance correction unit 112 approximates the correlation between the coordinate values and the distance information by a linear function for the coordinate corresponding points in the segment. Then, the distance correction unit 112 determines the three-dimensional object distance of each coordinate corresponding point based on the approximate expression. That is, according to the method 2, for example, the distance to the wall of the vehicle 20 in an inclined manner can be accurately calculated. The 3 rd method is a multidimensional approximation. The 3 rd method is a multidimensional function in which the approximation in the 2 nd method is performed 2 or more times. According to the method 3, the distance of an object having a complicated shape can be accurately calculated.

(operation of image processing apparatus)

The operation of the image processing apparatus 100 when displaying the viewpoint converted image on the display unit 13 will be described with reference to fig. 12. The operation of the image processing apparatus 100 described below is started at predetermined time intervals, for example, at 16ms intervals. The main execution unit of each step of the processing described below is the CPU 101.

First, in step S501, the CPU101 acquires a distance image from the distance detection unit 14. In the next step S502, the CPU101 acquires a captured image from the camera 10. In the next step S503, the CPU101 causes the image recognition unit 111 to process the distance image acquired in step S502 and execute segment division. The execution example of this step is the same as that explained in fig. 11. In the next step S504, the CPU101 executes steps S505 to S508 described below for each segment calculated in S503. Although the processing in S505 to S508 can be executed in parallel for all the sections, the description is given here in a form in which the processing objects are changed and the processing is performed sequentially one by one. The segment to be processed is referred to as a processing target segment.

In step S505, the CPU101 determines whether the section is an area corresponding to a solid object or an area corresponding to a road surface based on the distance information in the section to be processed. This determination can be made, for example, as follows. That is, since the mounting position and mounting posture of the camera 10A are known, the relationship between the position and the distance in the captured image can be calculated in advance by the association unit 114 assuming that the captured object is a road surface. Then, by comparing the difference between the distance information in the section and the distance calculated from the position of the section in the captured image, it is possible to determine whether or not the object in the section is a road surface. The CPU101 proceeds to step S506 when determining that the object in the section is a three-dimensional object, and proceeds to step S508 when determining that the object in the section is a road surface. If it is determined that the object in the processing target section is sky due to the infinite distance or the inability to measure the distance, the process proceeds to step S508.

In step S506, the CPU101 corrects the distance information of all the coordinate corresponding points in the processing target segment, that is, calculates the three-dimensional object distance, by using the distance correction unit 112. In the next step S507, the CPU101 causes the table correction unit 113 to rewrite all the coordinate corresponding points in the processing target block in the table data 121, and advances the process to step S509. However, the table data 121 corrected in this step is discarded after the execution of step S510 described later is completed, and it can be said that the processing in this step is correction of a temporary copy of the table data 121. In step S508 executed when it is determined in step S505 that the section to be processed is a road surface, the CPU101 proceeds to step S509 without causing the table correction unit 113 to correct the table data 121. That is, since no special processing is performed in S508, the process may be directly advanced to S509 when it is determined that the road surface is a road surface in step S505.

In step S509 executed after step S507 and step S508, the CPU101 determines whether all the extents have been targeted for processing. When determining that there is a section that has not yet been a processing target, the CPU101 sets the section as the processing target and returns to step S505. The CPU101 proceeds to step S510 if it is determined that all the sections have been targeted for processing. In step S510, the image conversion unit 115 converts the captured image of the camera 10 using the table data 121 corrected in step S507. Then, the display control unit 117 outputs the converted image to the display unit 13, and ends the processing shown in fig. 12.

According to embodiment 1 described above, the following operational effects are obtained.

(1) The image processing apparatus 100 includes: an interface 105 that acquires a captured image having a 1 st spatial resolution; an interface 105 that acquires a distance image as depth information having a 2 nd spatial resolution lower than the 1 st spatial resolution; an image recognition unit 111 that extracts a region including a three-dimensional object region corresponding to a three-dimensional object from a captured image; a distance correction unit 112 that calculates depth information of the three-dimensional object region from the distance image; a table correction unit 113 that corrects table data 121, which is coordinate conversion information for coordinate conversion of the captured image, based on the depth information of the three-dimensional object region calculated by the distance correction unit 112; and an image conversion unit 115 that generates a viewpoint conversion image obtained by coordinate-converting the captured image using the table data 121 corrected by the table correction unit 113. Therefore, a highly accurate viewpoint conversion image can be generated using a distance image that is distance information having a low spatial resolution.

(2) The image recognition unit 111 extracts the outlines of a plurality of segments including a three-dimensional object region from at least 1 of the brightness, lightness, chroma, and lightness of the captured image. Therefore, the image recognition unit 111 can easily divide the captured image into a plurality of sections including the three-dimensional object region.

(3) The table data 121 contains a combination of transformation source coordinates on a plurality of captured images and transformation destination coordinates on a viewpoint transformation image. The table correction unit 113 corrects the coordinates of the conversion destination, that is, the coordinates on the viewpoint conversion image, as shown in fig. 8, based on the distance information of the three-dimensional object region in the distance image. Therefore, the resolution of the range image can be improved in a pseudo manner.

(4) The table data 121 is created in advance on the assumption that the object in the captured image is an area on the road surface. The image recognition unit 111 divides the captured image into a plurality of sections for each object included in the captured image. The table correction unit 113 determines whether each of the plurality of zones is a three-dimensional object zone based on the distance image (S505 in fig. 12), and corrects the conversion destination coordinates corresponding to the conversion source coordinates in the zone determined to be a three-dimensional object zone in the table data 121 (S507). Therefore, the correction of the table data 121 is not necessary for the region that is not a three-dimensional object, and the processing time can be shortened.

(5) The table data 121 is calculated on the premise that the object of the captured image is present in the reference height plane. The table correction unit 113 calculates the height of the object from the reference height plane 230 and the position of the object on the reference height plane using the depth information of the three-dimensional object region calculated by the distance correction unit 112. Further, the table correction unit 113 corrects the table data 121 using the reference position of the viewpoint conversion, the calculated height of the object from the reference height plane, and the calculated position of the object on the reference height plane.

(modification 1)

Table data 121 in embodiment 1 represents a point P on a captured image represented by equation (5)_rTransforming point P on image with viewpoint_vThe corresponding relationship of (1). However, the table data 121 may also indicate the point P on the captured image represented by the equation (1)_rWith a point P in three-dimensional space_wThe corresponding relationship of (1). In order to distinguish the table data in modification 1 from the table data 121 in embodiment 1, the table data is hereinafter referred to as table data 121A. In the present modification, the captured image projected onto the three-dimensional section is distorted by correcting the table data 121A. The image conversion unit 115 creates a viewpoint conversion image, which is an image obtained by capturing a captured image projected onto a three-dimensional space from a virtual viewpoint.

Fig. 13 is a diagram showing an example of table data 121A. The table data 121A indicates a point P on the captured image 301 represented by equation (1)_rWith a point P in three-dimensional space_wThe corresponding relationship of (1). In the present embodiment, the table data 121A before correction includes a point P in the three-dimensional space_wZ of (A)_wThe values of the coordinates are all zero. That is, the table data 121A before correction indicates the point P on the captured image 301_rWith Z in three-dimensional space_wPoint P on plane 0_wThe corresponding relationship of (1).

According to the present modification, even when a virtual viewpoint that is not assumed in advance is set by the user, a viewpoint conversion image can be created using the table data 121A.

(modification 2)

In the above-described embodiment 1, the table data 121 is previously created on the assumption that all the objects exist on the road surface. However, the table data 121 may be created at any time instead of being created in advance. In this case, table data 121 is created in the processing shown in fig. 12. That is, the corresponding portion of the table data 121 is created in step S507 and step S508.

(modification 3)

In the above-described embodiment 1, the vehicle 20 is equipped with 4 cameras, that is, the front camera 10A, the left camera 10B, the right camera 10C, and the rear camera 10D. However, the vehicle 20 may be equipped with at least 1 camera. In addition, the vehicle 20 may be equipped with more than 5 cameras.

(modification 4)

In the above-described embodiment 1, the image recognition unit 111 divides the captured image into a plurality of regions by performing the segment division process, and the table correction unit 113 corrects the table data 121 to process the coordinate corresponding points of the region corresponding to the solid object. However, the image recognition unit 111 may specify only the region corresponding to the three-dimensional object in the captured image. For example, a region corresponding to a three-dimensional object can be specified within a captured image by extracting a portion in which depth information does not change in a stepwise manner in a range image and specifying a region of the same subject corresponding to the portion in the captured image.

(embodiment 2)

Embodiment 2 of the image processing apparatus 100 will be described with reference to fig. 14 to 18. In the following description, the same components as those in embodiment 1 are denoted by the same reference numerals, and different points are mainly described. The contents not particularly described are the same as those of embodiment 1. The present embodiment is different from embodiment 1 mainly in that the value of the coordinate on the captured image of the table data 121 is also corrected.

(outline of embodiment 2)

In embodiment 1, coordinate corresponding points are set densely in the extracted contour from the beginning. However, when the coordinate correspondence points are set sparsely in the extracted contour, the effect of embodiment 1 is limited. Therefore, in the present embodiment, the coordinate corresponding point is newly set in accordance with the extracted contour of the three-dimensional object.

Fig. 14 is a diagram illustrating a problem that may occur when embodiment 1 is applied. Fig. 14 (a) shows an example of a case where coordinate corresponding points are set sparsely in the outline. Fig. 14 (b) is a schematic diagram showing ideal depth information between PQs in fig. 14 (a), and fig. 14 (c) is a schematic diagram showing depth information between PQs in fig. 14 (a). In fig. 14 (b) and 14 (c), the upward direction in the figure indicates a long distance, and the white circles indicate outlines. As shown in fig. 14 (b), the distance information ideally clearly changes with the contour as a boundary. However, since the distance information is set only for the coordinate corresponding point and the interpolation processing is performed between the coordinate corresponding point and the coordinate corresponding point, it can be said that the distance information is different from the ideal as shown in fig. 14 (c).

Fig. 15 (a) is a diagram showing coordinate corresponding points corrected by the table correction unit 113 according to embodiment 2, and fig. 15 (b) is a diagram showing depth information between PQs in fig. 15 (a). In the present embodiment, as shown in fig. 15 (a), the coordinate corresponding points are set near the outline of the three-dimensional object region and on both sides of the outline, that is, outside and inside the three-dimensional object region. Therefore, as shown in fig. 15 (b), the distance changes sharply near the contour, and substantially coincides with the ideal distance shown in fig. 14 (b). In fig. 15, the space between the contour and the coordinate corresponding point is empty for convenience of illustration, but it is preferable that the contour and the coordinate corresponding point are as close to each other as possible.

(constitution)

The configuration of the image processing apparatus 100 and the hardware configuration of the vehicle 20 in embodiment 2 are the same as those in embodiment 1. However, as described later, the programs stored in the ROM102 of the image processing apparatus 100 operate differently.

Fig. 16 is a functional block diagram showing functions provided in the image processing apparatus 100 in the form of functional blocks. The image processing apparatus 100 includes a coordinate corresponding point setting unit 118 in addition to the functions of embodiment 1. The coordinate corresponding point setting unit 118 sets a coordinate corresponding point near the contour of the segment divided by the image recognition unit 111. The coordinate corresponding point setting unit 118 may correct the coordinates of the existing coordinate corresponding points, or may add new coordinate corresponding points.

(correction of Table data)

Fig. 17 is a diagram showing coordinates before and after correction by extracting a correction target coordinate corresponding point and a newly set coordinate corresponding point from the table data 121. Fig. 17 corresponds to fig. 8 in embodiment 1. Next, points different from fig. 8 will be mainly described. In embodiment 2, the coordinates on the captured image may also be corrected. For example, both of the Xr coordinate and the Yr coordinate of the coordinates on the captured image are corrected for the correspondence number "o 2" in fig. 17, and the Yr coordinate of the coordinates on the captured image is corrected for the correspondence number "o 3". The correspondence relation number "o 4" does not exist before correction, and thus it is known that the coordinate correspondence point is newly set.

(operation of image processing apparatus)

The operation of the image processing apparatus 100 according to embodiment 2 will be described with reference to fig. 18. Note that the operation common to embodiment 1 is not described. The processing up to step S505 is the same as that of embodiment 1, and thus, the description and illustration thereof are omitted.

If it is determined in step S505 that the object is a three-dimensional object, the process proceeds to step S521, and if it is determined that the object is a road surface or sky, the process proceeds to step S508. In step S521, the CPU101 causes the coordinate correspondence point setting unit 118 to set coordinate correspondence points in the vicinity of the contour of the processing target block and inside and outside the block.

In the next step S522, the CPU101 causes the distance correction unit 112 to set distance information of the coordinate corresponding points existing outside the segment among the coordinate corresponding points set in step S521. The distance of the coordinate corresponding point may be determined based on the distance of the coordinate corresponding point originally present on the outer side of the contour, or may be determined assuming that the coordinate corresponding point is present on the road surface. In the next step S506A, the CPU101 corrects the distance information of all the coordinate corresponding points in the processing target block, that is, calculates the three-dimensional object distance, by using the distance correction unit 112. The coordinate corresponding points newly set in step S521 are also included in all the coordinate corresponding points in the processing target segment. The processing after step S507 is the same as that in embodiment 1, and therefore, description thereof is omitted.

According to embodiment 2 described above, the following operational effects are obtained.

(6) The distance correction unit 112 includes a combination of the conversion source coordinates on the plurality of captured images and the conversion destination coordinates on the viewpoint conversion image. The table correction unit 113 corrects the transformation source coordinates in the vicinity of the outline of the three-dimensional object region. Therefore, even when the coordinate corresponding points are not set densely in the distance correction unit 112, the coordinate corresponding points can be corrected in the vicinity of the contour of the object, thereby making it possible to improve the reproduction accuracy of the three-dimensional shape in the viewpoint conversion image.

(7) The table correction unit 113 arranges the conversion source coordinates near the outline of the three-dimensional object region and outside and inside the three-dimensional object region. Therefore, the reproduction of the three-dimensional shape can be further highly accurately performed as shown in fig. 15.

The present invention includes various modifications, and is not limited to the embodiments described above. For example, the above-described embodiments have been described in detail with respect to the entire system in order to explain the present invention in a manner that is easy to understand, but the present invention is not necessarily limited to all the configurations described above. Note that a part of the configuration of one embodiment may be replaced with the configuration of another embodiment, and the configuration of one embodiment may be added to the configuration of another embodiment. Further, addition, deletion, and replacement of another configuration may be performed on a part of the configuration of each embodiment. Other embodiments contemplated within the scope of the technical idea of the present invention are also included in the scope of the present invention.

Further, each of the above-described configurations, functions, processing units, processing methods, and the like may be partially or entirely realized in hardware by designing them with an integrated circuit, for example. The above-described configurations, functions, and the like may be realized by software by a processor interpreting and executing a program for realizing the functions. Information such as programs, tables, and files for realizing the respective functions may be stored in a recording device such as a memory, a hard disk, an ssd (solid State drive), or a recording medium such as an IC card, an SD card, or a DVD.

The disclosure of the following priority base application is incorporated into this application by reference.

Japanese patent application 2017-128280 (application for 30.6.2017)

Description of the symbols

10 … video camera

14 … distance detecting unit

100 … image processing device

101…CPU

104 … storage part

105 … interface

111 … image recognition unit

112 … distance correction unit

113 … form correction unit

114 … related part

115 … image conversion unit

118 … coordinate corresponding point setting unit

121 … form data.