阅读说明：本技术 基于tof相机的栈板位姿确定方法、装置及计算机可读存储介质 (Method, apparatus and computer readable storage medium are determined based on the pallet pose of TOF camera ) 是由 赵军 陈佳搏 胡继文 宋利伟 于 2019-08-15 设计创作，主要内容包括：基于TOF相机的栈板位姿确定方法、装置及计算机可读存储介质,所述基于TOF相机的栈板位姿确定方法包括：获取TOF相机拍摄的深度图、点云数据；基于深度图确定待检测区域；选取第一区域进行地面拟合,确定地面区域；根据所述地面区域和点云数据,基于第一预设阈值确定栈板前端面的位置；采用物体连通域检测方法,获取所述待检测区域内所有的物体连通域区块；根据第二预设阈值筛选出实际的所述栈板的连通域区块；对所述实际的而所述栈板的连通域区块进行搜索,搜索第一线段；对所述第一线段进行拟合,计算出所述第一线段的空间位置；根据所述第一线段的空间位置确定栈板位姿。利用TOF相机解决目前的自动识别的场景复杂问题,抗干扰性强,本发明方法计算简单,使用方便,准确率高。(Determine that method, apparatus and computer readable storage medium, the pallet pose based on TOF camera determine that method includes: the depth map for obtaining TOF camera shooting, point cloud data based on the pallet pose of TOF camera；Area to be tested is determined based on depth map；It chooses first area and carries out ground fitting, determine ground region；According to the ground region and point cloud data, the position of pallet front end face is determined based on the first preset threshold；Area detecting method is connected to using object, obtains object connected domain block all in the area to be tested；The connected domain block of the actual pallet is filtered out according to the second preset threshold；The connected domain block of the pallet is scanned for, the first line segment is searched for；First line segment is fitted, the spatial position of first line segment is calculated；Pallet pose is determined according to the spatial position of first line segment.The scene challenge of current automatic identification, strong interference immunity are solved using TOF camera, the method for the present invention calculates simple, easy to use, accuracy rate height.)

1. a kind of pallet pose based on TOF camera determines method characterized by comprising

Obtain depth map, the point cloud data of TOF camera shooting；

Area to be tested is determined based on depth map；

It chooses first area and carries out ground fitting, determine ground region；

According to the ground region and point cloud data, the position of pallet front end face is determined based on the first preset threshold；

Area detecting method is connected to using object, obtains object connected domain block all in the area to be tested；

The connected domain block of the actual pallet is filtered out according to the second preset threshold；

The connected domain block of the pallet is scanned for, the first line segment is searched for；

First line segment is fitted, the spatial position of first line segment is calculated；

Pallet pose is determined according to the spatial position of first line segment.

2. determine that method, feature are being based on the pallet pose of TOF camera as described in claim 1,

The area to be tested range includes the pallet areas imaging.

3. determine that method, feature are being based on the pallet pose of TOF camera as described in claim 1,

The selection first area carries out ground fitting, determines that ground region includes:

The specific region being located at below the area to be tested is chosen as the first area, and with first area progress The fitting on ground；

Second area is fitted using the first area, the second area is also a part of ground region.

4. determine that method, feature are being based on the pallet pose of TOF camera as described in claim 1, it is described according to institute Ground region and point cloud data are stated, the position for determining pallet front end face based on the first preset threshold includes:

The point cloud data on ground or more is obtained based on the area to be tested and the second area；

The point cloud data for meeting the first preset threshold is the point cloud data of the pallet, and determines the front end face position of the pallet It sets.

5. determining that method, feature are being based on the pallet pose of TOF camera as claimed in claim 4, described first is pre- If threshold value is set as the width of the pallet.

6. determine that method, feature are being based on the pallet pose of TOF camera as described in claim 1, it is described according to The connected domain block that two preset thresholds filter out the actual pallet includes:

The size of all object connected domain blocks and the described second default threshold in the area to be tested that will acquire Value is compared the connected domain for filtering out the actual pallet.

7. determine that method, feature are being based on the pallet pose of TOF camera as described in claim 1, it is described according to institute The spatial position for stating the first line segment determines that pallet pose includes:

Determine first line segment up to a little less space line, first line segment and the space line Angle be angle of the pallet relative to camera.

8. a kind of paths planning method, feature is being, comprising:

As the described in any item pallet poses based on TOF camera of claim 1-7 determine method；

Determine the coordinate of the pallet；

Carry out path planning.

9. a kind of pallet pose determining device based on TOF camera characterized by comprising

Module is obtained, the module that obtains obtains the depth map of TOF camera shooting, point cloud data；

Computing module, the computing module calculate the ground region, calculate the position of the pallet front end face, described in calculating All object connected domain blocks in area to be tested calculate that connected domain block of the actual accounting, calculate institute The first line segment is stated, the spatial position of first line segment is calculated, the pallet pose is calculated, is obtained described in the computing module acquisition The information of modulus block.

10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that when the computer journey When sequence is read out by the processor, execute as the described in any item pallet poses based on TOF camera of claim 1-7 determine method.

Technical field

The present invention relates to object detecting areas more particularly to a kind of pallet pose based on TOF camera to determine method, apparatus And computer readable storage medium.

Background technique

At present under industry scene, pallet or pallet is cooperated to carry out the carrying of cargo frequently with handling device；In order to improve The intelligence of whole efficiency, handling device is gradually popularized, and during being that pallet or pallet are cooperated, handling device is equal Need first to pallet identify can it is safe and accurate, smoothly with pallet or pallet into cooperation.

Generally monocular is provided on current handling device or binocular camera is identified, but this method is for environment ratio It is more sensitive；It is also the pallet that special design pallet front end face is plane, by identifying that pallet front plan carries out identification positioning； Carry out identification positioning using in the fixation position of pallet label is enclosed in addition there are some, it is done so that cost it is very high, and And the service life of pallet is also limited.

That is, the discrimination of pallet needs to ensure, but currently without general and mature skill under existing industry scene Art can be realized the identification of pallet, can determine the pose of pallet so as to the development of other subsequent work.

Summary of the invention

The purpose of the present invention is to provide a kind of pallet poses based on TOF camera to determine that method, apparatus and computer can Storage medium is read, and a kind of paths planning method is provided, needs to ensure to solve the discrimination of pallet in the prior art, but mesh The preceding identification that can be realized pallet without general and mature technology can determine the pose of pallet so as to other subsequent work The problem of development.

As above design, the technical scheme adopted by the invention is that:

A kind of pallet pose based on TOF camera determines method, comprising:

Obtain depth map, the point cloud data of TOF camera shooting；

Area to be tested is determined based on depth map；

It chooses first area and carries out ground fitting, determine ground region；

According to the ground region and point cloud data, the position of pallet front end face is determined based on the first preset threshold；

Area detecting method is connected to using object, obtains object connected domain block all in the area to be tested；

The connected domain block of the actual pallet is filtered out according to the second preset threshold；

The connected domain block of the actual pallet is scanned for, the first line segment is searched for；

First line segment is fitted, the spatial position of first line segment is calculated；

Pallet pose is determined according to the spatial position of first line segment.

Further, the area to be tested range is greater than the pallet areas imaging.

Further, the selection first area carries out ground fitting, determines that ground region includes:

The specific region being located at below the area to be tested is chosen as the first area, and with first area progress The fitting on ground；

Second area is fitted using the first area, the second area is also a part of ground region.

Further, described according to the ground region and point cloud data, pallet front end is determined based on the first preset threshold The position in face includes:

The point cloud data on ground or more is obtained based on the area to be tested and the second area；

The point cloud data for meeting the first preset threshold is the point cloud data of the pallet, and determines the front end face position of the pallet It sets.

Further, first preset threshold is set as the width of the pallet.

Further, the connected domain block for filtering out the actual pallet according to the second preset threshold includes:

The size of all object connected domain blocks and the described second default threshold in the area to be tested that will acquire Value is compared the connected domain for filtering out the actual pallet.

Further, the spatial position according to first line segment determines that pallet pose includes:

Determine first line segment up to a little less space line, first line segment and the space line Angle be angle of the pallet relative to camera.

The present invention also provides a kind of paths planning methods, comprising:

The pallet pose based on TOF camera as described in any of the above item determines method；

Determine the coordinate of the pallet；

Carry out path planning.

The pallet pose determining device based on TOF camera that the present invention also provides a kind of, comprising:

Module is obtained, the module that obtains obtains the depth map of TOF camera shooting, point cloud data；

Computing module, the computing module calculate the ground region, calculate the position of the pallet front end face, described in calculating All object connected domain blocks in area to be tested calculate that connected domain block of the actual accounting, calculate institute The first line segment is stated, the spatial position of first line segment is calculated, the pallet pose is calculated, is obtained described in the computing module acquisition The information of modulus block.

The present invention also provides a kind of computer readable storage mediums, are stored thereon with computer program, when the computer When program is read out by the processor, executes the pallet pose based on TOF camera as described in any of the above item and determine method.

Beneficial effects of the present invention:

Pallet pose of the present invention based on TOF camera determines that method solves the field of current automatic identification using TOF camera Scape challenge, since TOF camera can be obtained comprising the three dimensional point cloud under pallet scene, the three dimensional point cloud of acquisition It is not influenced by external environments such as light, anti-interference is stronger.It is analyzed by the three dimensional point cloud obtained, identifies stack Plate, to obtain position and the posture of pallet.The method of the present invention calculates simple, easy to use, accuracy rate height.

Paths planning method of the present invention is due to determining that method can be quasi- using the pallet pose based on TOF camera The true position for getting the pallet carries out path planning, helps to realize the control of handling device, realize handling device with Pallet is accurate, safety, smoothly cooperates.

Detailed description of the invention

Fig. 1 is the flow chart of the present invention that method is determined based on the pallet pose of TOF camera；

Fig. 2 is the flow chart of paths planning method of the present invention；

Fig. 3 is the structural block diagram of the pallet pose determining device of the present invention based on TOF camera.

Specific embodiment

To keep the technical problems solved, the adopted technical scheme and the technical effect achieved by the invention clearer, below In conjunction with attached drawing and pass through specific embodiment to further illustrate the technical scheme of the present invention.It is understood that this place is retouched The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just Only the parts related to the present invention are shown in description, attached drawing and it is not all.

TOF camera is mounted on handling device, can shoot the position of pallet, handling device can be AGV fork truck or Other have the handling device of the mechanism cooperated with pallet.

With reference to Fig. 1, the present invention provides the pallet poses based on TOF camera to determine method, comprising:

Obtain depth map, the point cloud data of TOF camera shooting；

Area to be tested is determined based on depth map；

It chooses first area and carries out ground fitting, determine ground region；

According to the ground region and point cloud data, the position of pallet front end face is determined based on the first preset threshold；

Area detecting method is connected to using object, obtains object connected domain block all in the area to be tested；

The connected domain block of the actual pallet is filtered out according to the second preset threshold；

The connected domain block of the actual pallet is scanned for, the first line segment is searched for；

First line segment is fitted, the spatial position of first line segment is calculated；

Pallet pose is determined according to the spatial position of first line segment.

The depth map for obtaining TOF camera shooting, point cloud data refer to, in the engineering of handling device work, TOF phase Machine can get depth map, three dimensional point cloud in coverage in real time.In the field for needing to be determined pallet pose Jing Zhong, handling device preferably rest at the position that can clap full pallet, it is further preferable that when being located at the position, pallet position In the central area block position of the viewing field of camera of TOF camera, in brief, preferably in the areas imaging of TOF camera, pallet imaging Data are located in the middle part of range.

Described to determine that area to be tested refers to based on depth map, a part of range in selection depth map range walks to be checked Survey region.Preferably, the area to be tested range includes the pallet areas imaging, avoids pallet areas imaging not to be checked It surveys in regional scope, detection is caused to fail.The selection of area to be tested and the size of pallet are related, also, also with carrying The distance of device to pallet is related, can be chosen according to the actual situation.

By above-mentioned steps, the range of the depth map of TOF camera imaging and three dimensional point cloud is further reduced, And depth map after diminution and three dimensional point cloud are closer to the depth map and three dimensional point cloud of pallet.

The selection first area carries out ground fitting, determines that ground region includes:

The specific region being located at below the area to be tested is chosen as the first area, and with first area progress The fitting on ground；

Second area is fitted using the first area, the second area is also a part of ground region.

The step for main function be to fit ground data in depth map and three-dimensional data that is, which determines Part is ground, and ground and pallet, other objects are distinguished.The general specific region chosen below area to be tested is as the One region, in one embodiment, handling device be AGV fork truck, TOF shooting depth map in, select two fork tooth among areas Domain is as first area, in a program can be to the area because this part of specific region known to us must be ground region Domain is preset, and when subsequent progress pallet pose determines, the specific region among two fork teeth is chosen in first area.In In other embodiments, naturally it is also possible to select other specific regions.It will be appreciated that lower section here is primarily referred to as depth map In a direction relations, and invent the scheme and be not limited to literal " lower section " meaning, as long as the first area is ground Region, because the three dimensional point cloud of ground region has the characteristics that certain, for example deep equality therefore can be accordingly The second area within the scope of area to be tested is fitted, the second area is also ground region.It will be appreciated that here Two regions are likely located at all directions up and down of area to be tested in depth map.

By above-mentioned movement, by the ground imaging region and pallet imaging region, the progress of other regions in area to be tested It divides.

It is described according to the ground region and point cloud data, determine that the position of pallet front end face is wrapped based on the first preset threshold It includes:

The point cloud data on ground or more is obtained based on the area to be tested and the second area；

The point cloud data for meeting the first preset threshold is the point cloud data of the pallet, and determines the front end face position of the pallet It sets.

It, can be with after being divided ground imaging region and pallet imaging region, other regions in preamble movement The point cloud data other than ground region is obtained from all point cloud datas in area to be tested, at this point, these point cloud data packets The point cloud data of pallet is included, or the point cloud data including pallet and cargo.In general, under industrial scene, picking or It placed pallet at discharge location, do not have other objects in addition to pallet, cargo around pallet, it such as if so, can also be with By carrying out processing screening except the three dimensional point cloud for belonging to pallet and cargo to three dimensional point cloud.

After the point cloud data for filtering out the ground region in area to be tested, in order to further obtain the point cloud number of pallet According to determining the front end face of pallet, can be set according to the actual situation with the first preset threshold of preset in advance, the first preset threshold Set, be preferably arranged to the width of pallet, that is, pallet front end face width, getting the point cloud number except ground region Accordingly in outer point cloud data, when relationship the first preset threshold of satisfaction between point cloud data, then front end face can be determined substantially Point cloud data, also determined that the front end face of pallet.

Described to be connected to area detecting method using object, obtaining object connected domain block all in the area to be tested is Refer to, the point cloud data within the scope of the area to be tested filtered out in addition to ground region is carried out using object connection area detecting method Processing, obtains object connected domain block all in area to be tested.

The connected domain block for filtering out the actual pallet according to the second preset threshold includes:

The size of all object connected domain blocks and the described second default threshold in the area to be tested that will acquire Value is compared the connected domain for filtering out the actual pallet.

Second preset threshold can be multiple threshold values, it is preferred to use the size of the front end face of pallet and including stack The length of plate, this makes it possible to the connected domain blocks of the actual pallet of further screening, in this way according to the size feelings of connected domain The available only pallet connected domain of condition, excludes other objects, further reduces the point cloud range of pallet after the completion, this Sample processing can also filter out the interference that cargo detects pallet.

The connected domain block of the actual pallet is scanned for, the first line segment is searched for, refers to, to above-mentioned steps The connected domain block of the actual pallet of middle determination is handled, and searches for the first line segment, the first line segment here can be according to reality Border situation is configured, main that reason is selected to be to facilitate detection and high with pallet pose correlation, it is preferable that the choosing of the first line segment It has selected on pallet front end face apart from the pin pitch of pallet from for 0 and the length line segment equal with pallet width, the lower section of the first line segment The as entry of pallet；The mode of search can be carries out line search from top to bottom, when next from top to bottom is no longer Determine that currently this is the first line segment when line segment but when entry；In embodiment, when search, connected domain block shows as two-value Figure, pallet numerical value shows as 111, and entry position numerical value shows as 000, in this way once occurring 000 it is known that upper one Line segment is the first line segment.

First line segment is fitted, the spatial position of first line segment is calculated, preferably according to the first line segment Go out the space line where depth by space line RANSAC the Fitting Calculation, that is, fitting is insinuated by depth map and determines the The space line of one line segment.So far the position for obtaining the first line segment of practical pallet is arranged according to preparatory coordinate system etc., can To know the pose of pallet.

The spatial position according to first line segment determines pallet pose further include:

Determine first line segment up to a little less space line, first line segment and the space line Angle be angle of the pallet relative to camera.

The space line of all the points on first line segment can be calculated using above-mentioned same method, that is, determined The spatial position of the straight line of all the points of first line segment, because at least needing two points that can determine straight line, excellent The same prevention of choosing calculate the mistake all the points on the first line segment and with perpendicular straight of the X-axis of the coordinate system of setting or Y axis The space line of line is pallet front end face relative to the appearance under camera coordinates system using the angle that two straight lines calculate State.

It will be appreciated that the pose of pallet is different according to different coordinate system settings or other artificial settings, but pallet sheet Position in space be it is identical, therefore, the different not shadows of definition setting of the different settings of coordinate system to the pose of pallet Ring the application of the method for the invention.

Pallet pose of the present invention based on TOF camera determines that method solves current automatic identification using ToF camera Scene challenge, since tof camera can be obtained comprising the three dimensional point cloud under pallet scene, the three-dimensional point cloud of acquisition Data are not influenced by external environments such as light, and anti-interference is stronger.It is analyzed, is identified by the three dimensional point cloud obtained Pallet, to obtain position and the posture of pallet.

Another aspect of the present invention additionally provides a kind of paths planning method, comprising:

Obtain depth map, the point cloud data of TOF camera shooting；

Area to be tested is determined based on depth map；

It chooses first area and carries out ground fitting, determine ground region；

According to the ground region and point cloud data, the position of pallet front end face is determined based on the first preset threshold；

Area detecting method is connected to using object, obtains object connected domain block all in the area to be tested；

The connected domain block of the actual pallet is filtered out according to the second preset threshold；

The connected domain block of the pallet is scanned for, the first line segment is searched for；

First line segment is fitted, the spatial position of first line segment is calculated；

Pallet pose is determined according to the spatial position of first line segment；

Determine the coordinate of the pallet；

Carry out path planning.

The pose of pallet has been determined through the above scheme, that is, the real space position of the first line segment of pallet has been determined Afterwards, second area to be tested can be chosen by the central region in the first line segment, pallet is found by way of search The position of crura intermedium, Left-right Searching sequence can get the regional scope of the crura intermedium of pallet；When the pallet of preferred definition Coordinate be pallet crura intermedium coordinate when, by calculating the coordinate average value of all the points on crura intermedium, can obtain To the coordinate of pallet.

Then path planning is carried out, navigation handling device goes to pallet to work.

Paths planning method of the present invention is due to determining that method can be quasi- using the pallet pose based on TOF camera The true position for getting the pallet carries out path planning, helps to realize the control of handling device, realize handling device with Pallet is accurate, safety, smoothly cooperates.

Another aspect of the present invention additionally provides a kind of pallet pose determining device based on TOF camera, comprising:

Module is obtained, the module that obtains obtains the depth map of TOF camera shooting, point cloud data；

Computing module, the computing module calculate the ground region, calculate the position of the pallet front end face, described in calculating All object connected domain blocks in area to be tested calculate that connected domain block of the actual accounting, calculate institute The first line segment is stated, the spatial position of first line segment is calculated, the pallet pose is calculated, is obtained described in the computing module acquisition The information of modulus block.

Another aspect of the present invention additionally provides a kind of computer readable storage medium, is stored thereon with computer program, when When the computer program is read out by the processor, executes the pallet pose as described above based on TOF camera and determine method.

Embodiment of above only elaborates that the basic principles and features of the invention, the present invention are not limited by above embodiment System, without departing from the spirit and scope of the present invention, the invention also has various changes and changes, and these changes and changes are all It falls into scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent circle It is fixed.