阅读说明：本技术 一种基于双相机视觉的熔融态悬浮椭球液滴图像处理算法 (Molten state suspension ellipsoid droplet image processing algorithm based on double-camera vision ) 是由 钟秋 杨莉萍 李会东 陶冶 雒彩云 徐子君 汪文兵 于 2020-04-10 设计创作，主要内容包括：本发明公开一种基于双相机视觉的熔融态悬浮椭球液滴图像处理算法,包括通过两台相机采集标准球的标准球图像,根据采集到的标准球图像计算出两个相机图像的单个像元各自对应的实际尺寸；利用两台相机以规定的夹角同步采集液滴的图像,获得液滴的两个图像；建立液滴的椭球体二次曲面方程,对两个图像分别检测边缘轮廓线,建立边缘轮廓线的椭圆轮廓线方程；引入规定的夹角,根据两个椭圆轮廓线方程与椭球体二次曲面方程之间各参数的关系构建方程组；求解方程组,计算出液滴的各个半轴的长度；根据各个半轴的长度求出液滴的体积。(The invention discloses a molten state suspension ellipsoid droplet image processing algorithm based on dual-camera vision, which comprises the steps of collecting standard sphere images of a standard sphere by two cameras, and calculating the actual sizes corresponding to single pixels of the two camera images according to the collected standard sphere images; synchronously acquiring images of the liquid drops by using two cameras at a specified included angle to obtain two images of the liquid drops; establishing an ellipsoid quadric surface equation of the liquid drop, respectively detecting edge contour lines of the two images, and establishing an ellipse contour line equation of the edge contour lines; introducing a specified included angle, and constructing an equation set according to the relation of each parameter between two ellipse contour line equations and an ellipsoid quadric surface equation; solving an equation set, and calculating the length of each half shaft of the liquid drop; the volume of the drop is determined from the length of each half-axis.)

1. A molten state suspension ellipsoid droplet image processing algorithm based on dual-camera vision is characterized by comprising the following steps:

1) the method comprises the steps that standard ball images of a standard ball are collected through two cameras, and the actual sizes corresponding to single pixels of the two camera images are calculated according to the collected standard ball images;

2) synchronously acquiring images of the liquid drops by utilizing the two cameras at a specified included angle to obtain two images of the liquid drops;

3) establishing an ellipsoid quadric surface equation of the liquid drop, respectively detecting edge contour lines of the two images, and establishing an ellipse contour line equation of the edge contour lines;

4) introducing the specified included angle, and constructing an equation set according to the relation of each parameter between the two ellipse contour line equations and the ellipsoid quadric surface equation;

5) solving the equation set, and calculating the length of each half shaft of the liquid drop;

6) and calculating the volume of the liquid drop according to the length of each half shaft.

2. The molten state suspended ellipsoid droplet image processing algorithm based on dual camera vision as claimed in claim 1,

the step 3) comprises the following steps:

constructing an ellipsoid quadric surface equation of the liquid drop by using a symmetric positive definite matrix;

performing binarization processing on each image by using a threshold method, detecting the edge contour line of each image by using an edge detection algorithm, and constructing an ellipse contour line equation of each edge contour line;

regarding each image as an image on a respective y-z plane, and extracting a coordinate set of each point on each edge contour line;

and obtaining the estimation value of each parameter in each elliptical contour line equation by adopting a least square method.

3. The molten state suspended ellipsoid droplet image processing algorithm based on dual-camera vision as claimed in claim 1 or 2,

in the step 3), after the edge contour lines of each image are detected, the center of each edge contour line is located at the center of the corresponding image.

4. The molten state ellipsoid droplet image processing algorithm based on dual camera vision as claimed in claim 2,

in the step 5), an iterative method is adopted to obtain an optimal solution of each parameter in the equation set, and the length of each half shaft of the liquid drop is calculated through the symmetrical positive definite matrix.

5. The molten state ellipsoid droplet image processing algorithm based on dual camera vision as claimed in claim 4,

the liquid drop is formed into a shape of a spheroid;

in the step 6), the semi-axes are classified into a first semi-axis and a second semi-axis by using a cluster analysis algorithm, and the average value of each of the first semi-axis and the second semi-axis is obtained.

6. The molten state suspended ellipsoid droplet image processing algorithm based on dual camera vision as claimed in claim 1,

measuring the liquid drops before and after the experiment respectively, calculating a mass average value, and taking the mass average value as the mass of the liquid drops;

and calculating the density of the liquid drop according to the volume and the mass of the liquid drop.

Technical Field

The invention relates to the field of liquid sample image processing under a container-free method, in particular to a molten state suspended ellipsoid droplet image processing algorithm based on double-camera vision.

Background

Space material science is used as an important branch in the fields of space science and application, is an extension of traditional material science to the space environment, and is one of the most active frontier cross disciplines in the new fields of developing a new theory of material science, exploring a material preparation process and expanding material application. The space has special effects of microgravity, ultra-vacuum, no container, strong radiation and the like, and is an ideal test condition for researching melting, solidification and the like of materials. However, the space resources available to people to date are still quite limited. Therefore, terrestrial methods that simulate various effects in a spatial environment are applied. One of them is the levitation technique, which can simulate a containerless state in a space environment. In the suspension technique, density measurement of suspended liquid drops is the most effective way for obtaining density data of liquid materials. In the density measurement, under the action of the surface tension of the liquid drop, the liquid sample in a container-free state shrinks into a sphere, so that the image of the sample is shot by a camera, the diameter of the sample is obtained, the volume of the sample can be further obtained, and finally, the density of the sample is obtained by combining the mass measurement result of the sample.

However, in the suspension process of the sample, the sample inevitably rotates due to the influence of uneven acting force such as suspension force, laser light pressure and the like. And under the rotation effect, the liquid sample is influenced by the centrifugal force, the appearance of the sample is changed from spherical to ellipsoidal, and the suspension attitude of the ellipsoid cannot be judged because the surface of the sample is a smooth surface, so that the three-dimensional information of the image is difficult to obtain by the traditional double-camera vision method so as to obtain the volume of the sample, and the accuracy of obtaining the density by the image of the ellipsoid sample is influenced.

Disclosure of Invention

The problems to be solved by the invention are as follows:

in view of the above problems, the present invention provides a molten suspension ellipsoid droplet image processing algorithm based on dual camera vision, which can improve the measurement accuracy in the container-less technology.

The technical means for solving the problems are as follows:

the invention provides a molten state suspended ellipsoid droplet image processing algorithm based on double-camera vision, which comprises the following steps of:

1) the method comprises the steps that standard ball images of a standard ball are collected through two cameras, and the actual sizes corresponding to single pixels of the two camera images are calculated according to the collected standard ball images;

2) synchronously acquiring images of the liquid drops by utilizing the two cameras at a specified included angle to obtain two images of the liquid drops;

3) establishing an ellipsoid quadric surface equation of the liquid drop, respectively detecting edge contour lines of the two images, and establishing an ellipse contour line equation of the edge contour lines;

4) introducing the specified included angle, and constructing an equation set according to the relation of each parameter between the two ellipse contour line equations and the ellipsoid quadric surface equation;

5) solving the equation set, and calculating the length of each half shaft of the liquid drop;

6) and calculating the volume of the liquid drop according to the length of each half shaft.

According to the invention, the two cameras are used for synchronously acquiring the images of the liquid drops in the shape of the ellipsoid of revolution, a contour line equation set is established according to the two images, and the equation set is solved to obtain the long axis and the short axis of the liquid drops so as to calculate the volume of the liquid drops, so that the three-dimensional information can be accurately obtained even if the liquid drops are molten state suspension liquid drops with smooth surfaces and rotation axes which are difficult to judge, and the volume of the liquid drops is calculated with high precision.

In the present invention, the step 3) may include: constructing an ellipsoid quadric surface equation of the liquid drop by using a symmetric positive definite matrix; performing binarization processing on each image by using a threshold method, detecting the edge contour line of each image by using an edge detection algorithm, and constructing an ellipse contour line equation of each edge contour line; regarding each image as an image on a respective y-z plane, and extracting a coordinate set of each point on each edge contour line; and obtaining the estimation value of each parameter in each elliptical contour line equation by adopting a least square method.

In the present invention, in the step 3), after the edge contour lines of each of the images are detected, the center of each edge contour line may be located at the center of the corresponding image. Therefore, the ellipse contour equation can be simplified, and the calculation amount can be reduced.

In the present invention, in the step 5), an iterative method is adopted to find an optimal solution of each parameter in the equation set, and the length of each half axis of the droplet is calculated through the symmetric positive definite matrix.

In the present invention, the liquid droplets may be formed in a shape of a spheroid; in the step 6), the semi-axes are classified into a first semi-axis and a second semi-axis by using a cluster analysis algorithm, and the average value of each of the first semi-axis and the second semi-axis is obtained.

In the present invention, the droplets may be measured before and after the experiment, and the mass average value may be determined as the mass of the droplets; and calculating the density of the liquid drop according to the volume and the mass of the liquid drop. Therefore, the influence of mass change caused by volatilization in the sample melting process on the density measurement result can be reduced.

The invention has the following effects:

the invention can improve the measurement precision of the molten suspension liquid drop in the container-free technology, improve the reliability and accuracy of the volume calculation of the ellipsoid sample, and provide accurate data support for the research of liquid materials.

Drawings

FIG. 1 is a schematic diagram of a containerless levitation and imaging system based on a dual camera vision based molten-state suspended ellipsoid droplet image processing algorithm according to one embodiment of the present invention;

FIG. 2 is a flow chart of a molten state suspended ellipsoid droplet image processing algorithm (hereinafter referred to as image processing algorithm) based on dual camera vision according to the present invention;

FIG. 3 is a diagram showing an image acquired by two cameras capturing a droplet with an included angle in an arbitrary plane and a contour line corresponding to an ellipsoid image by edge detection;

description of the symbols:

1. a droplet; 2. an upper electrode; 3. a lower electrode; 4. a first camera; 5. a second camera; 6. a first background light source; 7. a second background light source.

Detailed Description

The present invention is further described below in conjunction with the following embodiments and the accompanying drawings, it being understood that the drawings and the following embodiments are illustrative of the invention only and are not limiting thereof.

A molten suspended ellipsoid droplet image processing algorithm based on dual camera vision is disclosed herein that can improve measurement accuracy in containerless technology density measurements.

The invention shoots the molten suspension ellipsoid liquid drop under the container-free technology by two cameras with included angles, extracts the contour line in the camera image and constructs an equation set to calculate the volume of the ellipsoid liquid drop, and further obtains the density of the ellipsoid liquid drop.

FIG. 1 is a schematic diagram of a containerless levitation and imaging system based on a dual camera vision based molten-state ellipsoid droplet image processing algorithm according to one embodiment of the present invention. As shown in fig. 1, in the present embodiment, the levitation system is an electrostatic levitation capacitorless method, and is mainly composed of an upper electrode 2, a lower electrode 3, and a laser, a thermometer, and the like, which are not shown. The upper electrode 2 and the lower electrode 3 can be high-voltage parallel polar plates, and the generated high-voltage electric field enables the charged sample to counteract the gravity of the sample to suspend under the action of coulomb force. After the sample is stably suspended, the sample is heated using a laser not shown, and the temperature of the sample is taken by a thermometer not shown, whereby the droplet 1 in a molten state is formed. The liquid droplets 1 are rotated in an arbitrary posture by a levitation force, a laser light pressure, or the like, and are subjected to a centrifugal force by the rotation action to be changed from a spherical shape to an ellipsoidal shape, specifically, a rotational ellipsoidal shape.

The imaging system is mainly composed of two groups of imaging devices, namely a first camera 4 and a corresponding first background light source 6, and a second camera 5 and a corresponding second background light source 7. The first camera 4 and the second camera 5 may be, for example, CCD cameras, which are in an arbitrary plane and enclose an angle α with each other, mainly for capturing images of the droplet 1 from different angles. Since the liquid drop 1 emits light in a molten state, the first background light source 6 and the second background light source 7 are mainly used for reducing the influence of the light emission of the liquid drop 1, so that a corresponding camera can obtain a clear image.

Fig. 2 is a flow chart of the molten suspension ellipsoid droplet image processing algorithm based on dual camera vision according to the present invention. The main steps of the molten suspension ellipsoid droplet image processing algorithm based on dual-camera vision according to the present invention are described in detail below with reference to fig. 2.

1) Calibrating the camera imaging image, and determining the actual size corresponding to a single pixel of the camera image. Specifically, a standard sphere of known diameter size (which may be a SiC sphere, for example) is image-captured from any angle using a camera. And calculating to obtain the actual sizes corresponding to the single pixels of the two camera images according to the number of diameter pixels of the standard sphere image and the diameter of the standard sphere, which are acquired by the two cameras respectively.

Fig. 3 is a diagram showing an image acquired by capturing a droplet with two cameras having an included angle between the two cameras on an arbitrary plane and a contour line corresponding to an ellipsoid image obtained by edge detection. The left side of fig. 3 is an image of the ellipsoid droplets acquired by each of the two cameras, and the right side is a contour line of the image obtained by performing edge detection on each of the two images.

2) Two cameras with an included angle alpha are used for collecting images of the molten suspension ellipsoid droplets, and the positions of the cameras are shown in figure 1. Images IM1 and IM2 of ellipsoidal droplets as shown on the left side of fig. 3 were obtained. The angle α between the two cameras may be any angle, preferably 90 °.

In step 1) and step 2), only one camera may be used to capture the standard ball or the liquid drop from two positions on any plane that form the included angle α.

3) Edge contour lines are detected for the images IM1 and IM2, and elliptical contour line equations for the respective edge contour lines are established.

First, binarization processing is performed on the images IM1 and IM2 by using a threshold method, respectively, and edge contour lines of the images shown on the right side of fig. 3 are detected by an edge detection algorithm. In the present embodiment, it is preferable that the otsu method is used to calculate the threshold value, the image is binarized, and the sobel algorithm is used to perform edge detection on the binarized image, so as to obtain the edge contour line of the image.

Because the surface of the suspension drop sample can be considered to be completely smooth and the rotating shaft is uncertain, the traditional double-camera vision method cannot be used for carrying out three-dimensional reconstruction on the image in a characteristic point extraction mode, and the ellipsoid needs to be subjected to quadric surface modeling to obtain an ellipsoid quadric surface equation. The ellipsoid quadric surface is a three-dimensional positive quadratic form, so the ellipsoid quadric surface equation satisfies the formula (1):

(X-X₀)^TA(X-X₀)＝1

wherein A is a symmetric positive definite matrix, X₀As the coordinate center, X ═ X, y, z)^TIs an ellipsoidal space coordinate. Here, for convenience of processing, in the present embodiment, the center of the edge contour line of the elliptical shape is overlapped with the center position of each image, that is, the center of each edge contour line is set to the coordinate center of the coordinate system of each image. In this case, the formula (1) is simplified to X^TAX＝1。

Next, the image IM1 and the image IM2 are regarded as sample images obtained by capturing the droplets along different x-axes (the included angle between the two x-axes is α) with two cameras, that is, a graphThe images IM1 and IM2 are images in the respective y-z planes, with X being 0, i.e., X being (0, y, z) at each point on the edge profile^TThereby, a two-dimensional elliptical contour expression of the image can be obtained. Therefore, based on the simplified expression (1), the elliptical contour expressions of the image IM1 and the image IM2 can be expressed as expression (2):

(0,y,z)^TA(0,y,z)＝1。

since a is a symmetric positive definite matrix, it can be assumed that a is:

the symmetric positive definite matrix A is substituted into an expression (2), and the elliptic contour expression is further simplified into an expression (3):

for the image IM1, the coordinate set of points on the edge outline of the image IM1 is extracted { (Y)₁,Z₁)|Y₁＝(y₁,y₂,…y_n)^T，Z₁＝(z₁,z₂,…z_n)^TSubstituting it into equation (3) to obtain the ellipse contour equation:

the above formula can also be expressed as formula (4):

wherein I_nIs a column vector of length n, element 1.

Similarly, the coordinate set { (Y) of each point on the edge contour is also extracted for the image IM2₂,Z₂)|Y₂＝(y₁,y₂,…y_m)^T，Z₂＝(z₁,z₂,…z_m)^TSubstituting formula (3) to obtain formula (5):

wherein I_mIs a column vector of length m and element 1.

Then, the equations (4) and (5) are estimated by the least square method, respectively. Respective parameters d of the elliptical contour equation of the image IM1₁,e₁,f₁The estimated values of (a) are:

respective parameters d of the elliptical contour equation of the image IM2₂,e₂,f₂The estimated values of (a) are:

4) considering that the two cameras capture images of suspended ellipsoidal droplets at an angle α between them, the planes of the images IM1 and IM2 are actually two planes at an angle α between them, i.e., the elliptical contours of the images IM1 and IM2 are two contours of the same ellipsoid at an angle α between them, thus, the parameters a-f in the equation for ellipsoid quadric surface described in equation (1) and the parameter d in the equations for the two elliptical contours described in equations (4) and (5)₁,d₂,e₁,e₂,f₁,f₂There is the following relationship between:

wherein T is an intermediate parameter, α is an included angle between two cameras, d₁,d₂,e₁,e₂,f₁,f₂The elliptical profile equations for the images IM1 and IM2, respectively, as described in step 3) above, are estimated using a least squares method.

5) Using stacksSolving the nonlinear quadratic equation set by the substitution method can obtain the optimal solution of the parameters a-f of the quadric surface equation, thereby obtaining the symmetric positive definite matrix A. Whereas for the symmetric positive definite matrix a, there is a eigenvector (λ)₁,λ₂,λ₃) And a feature matrix R of 3 × 3, such that the following holds:

from this, the length of the three half-axes of the ellipsoid can be determined:

6) since the molten liquid drop rotates in suspension and is formed into a rotational ellipsoid shape rotating around its own minor axis under the action of centrifugal force, the determined ellipsoid semiaxis length L is determined_a、L_b、L_cDividing three half axes of the rotation ellipsoid into two types according to the length by using a K-Means clustering (K-Means) algorithm in a clustering analysis algorithm, wherein one type is the length of the long half axis of the rotation ellipsoid, and the other type is the length of the short half axis of the rotation ellipsoid, and respectively calculating the average values of the three half axes to obtain the long half axis L of the liquid drop of the rotation ellipsoid₁And a short semi-axis L₂Finally, obtaining an expression (6) of the volume of the suspension ellipsoid droplet:

in the present embodiment, the mass of the droplets is measured twice before and after the experiment to obtain the mass average value, the mass average value is used as the mass of the droplets, and the density of the droplets is determined from the mass of the droplets and the volume of the droplets obtained as described above. Therefore, the influence of mass change caused by volatilization in the sample melting process on the density measurement result can be reduced.

The present invention will be described in detail by way of examples. It is also to be understood that the following examples are illustrative of the present invention and are not to be construed as limiting the scope of the invention, and that certain insubstantial modifications and adaptations of the invention by those skilled in the art may be made in light of the above teachings. The specific process parameters and the like of the following examples are also only one example of suitable ranges, i.e., those skilled in the art can select the appropriate ranges through the description herein, and are not limited to the specific values exemplified below.