CN107869954A - A binocular vision volume weight measurement system and its realization method - Google Patents

Abstract

The invention discloses a kind of binocular vision volume weight measuring system, including Installation cabinet, and it is arranged on the first industrial camera, the second industrial camera, LASER Light Source, LED bar graph light source, signal processing system, measuring table and Bluetooth electronic scale in the Installation cabinet, wherein, the measuring table is arranged on the bottom of the Installation cabinet, and the Bluetooth electronic scale is arranged on the lower section of the measuring table；Second industrial camera is arranged on the top of the Installation cabinet, and the LASER Light Source is arranged side by side with second industrial camera, and the shooting direction keeping parallelism of the radiation direction of the LASER Light Source and second industrial camera；The present invention is compared to traditional artificial tape measure mode, and the advantage such as have quick, accurate, non-contact and cost cheap suitable for the automatic measurement of batch, greatly reduces cost of labor, improves production efficiency.

Description

A binocular vision volume weight measurement system and its realization method

technical field

The invention relates to the technical field of non-contact intelligent measurement and control, in particular to a binocular vision volume weight measurement system and an implementation method thereof.

Background technique

With the rapid development of the logistics industry in China, the rapid measurement and sorting of the size and weight of various logistics boxes to avoid traffic jams in the logistics industry has become an urgent problem to be solved. At present, the measurement of the logistics box is still in the process of manually measuring the volume and weight of the logistics box. In the measurement technology, the general laser ranging technology can only obtain the depth information of the object, but cannot give the object area. Although laser scanning technology can create a three-dimensional model by scanning the surface of the object, it has strict requirements on the surface of the object. For example, it is necessary to coat the surface of the object with a developer so that the laser can produce diffuse reflection on it. This method is cumbersome to operate and has limitations in application. sex.

Contents of the invention

The purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and provide a binocular vision volume weight measurement system.

Another object of the present invention is to provide a method for realizing a binocular vision volumetric weight measurement system.

The object of the present invention is achieved through the following technical solutions:

A binocular visual volumetric weight measurement system, including an installation cabinet, and a first industrial camera, a second industrial camera, a laser light source, an LED bar light source, a signal processing system, a measurement platform and a Bluetooth electronic device installed in the installation cabinet scale, wherein the measurement platform is arranged at the bottom of the installation cabinet, the bluetooth electronic scale is arranged below the measurement platform; the second industrial camera is arranged at the top of the installation cabinet, and the laser light source It is arranged side by side with the second industrial camera, and the light direction of the laser light source is kept parallel to the shooting direction of the second industrial camera; the second industrial camera and the laser light source are arranged on the front of the measurement platform above, and the shooting direction of the second industrial camera and the light direction of the laser light source are both perpendicular to the measurement platform; the first industrial camera is arranged on the top of the installation cabinet, and the first industrial camera Located obliquely above the measurement platform, and the first industrial camera is set at an angle of 45 degrees to the measurement platform; the first industrial camera and the second industrial camera have the same field of view of the measurement platform; There are four LED bar-shaped light sources, two of which are horizontally arranged on the front and rear sides of the second industrial camera, and the other two LED bar-shaped light sources are vertically arranged on the sides of the first industrial camera. front and rear sides;

The first industrial camera, the second industrial camera, the laser light source and the bluetooth electronic scale are respectively connected to the signal processing system; the signal processing system is provided with a display screen.

Preferably, the measurement platform is painted and polished to a black smooth surface; such setting can ensure that the logistics box enters and exits the measurement platform smoothly, and can reduce background interference during image processing.

A kind of realization method by above-mentioned binocular vision volumetric weight measurement system, comprises the following steps:

Step 1, start the binocular vision volume and weight measurement system, the first industrial camera, the second industrial camera, laser light source, Bluetooth electronic scale, LED bar light source and signal processing system start to work;

Step 2: Calibrate the internal parameters, external parameters and distortion coefficients of the first industrial camera and the second industrial camera, and transmit the calibration results to the signal processing system. The specific workflow is as follows:

(1) The first industrial camera and the second industrial camera satisfy the pinhole camera model, and the image coordinate vector is set as Where (u, v) is the pixel coordinates of the target point; the internal parameter matrix of the camera is Where f _x , f _y , c _x , c _y are the x-direction focal length, y-direction focal length and optical axis center coordinates respectively; the camera extrinsic parameter matrix is (R|T), where R is the camera optical center coordinate system relative to the world coordinate system The 3×3 rotation matrix, T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system; the world coordinate vector is Among them, X, Y, and Z are the world coordinates of the target point; the camera imaging model satisfies the relation:

where z _c is the scale factor; in addition, the camera distortion model satisfies the following relationship:

Among them, (x, y) are the physical coordinates of the image before distortion correction, (x _cor , y _cor ) are the physical coordinates of the image after distortion correction, r=x ² +y ² ,k ₁ ,k ₂ ,k ₃ ,p ₁ , p ₂ are three radial distortion coefficients and two tangential distortion coefficients of the camera;

The purpose of the calibration of the first industrial camera and the second industrial camera is to solve the internal parameters, external parameters and distortion coefficient of the camera. The measurement platform is set as the zero plane, and Zhang Zhengyou’s checkerboard plane calibration method is used to take 20 images by changing the position and angle of the checkerboard. , to calibrate the camera, and solve to obtain the camera internal parameters, external parameters and distortion coefficients;

Step 3, place the logistics box on the measurement platform, the laser light source illuminates the logistics box, and forms a laser spot on the upper surface of the logistics box, the Bluetooth electronic scale detects the weight of the logistics box, and when the data of the Bluetooth electronic scale is stable, record the logistics box Weight and transmit the data to the signal processing system through the Bluetooth module on the electronic scale, and then the signal processing system transmits the weight data to the first industrial camera and triggers it to collect the side view image C1 of the logistics box. The first industrial camera defaults to a small aperture mode, which can ensure that the brightness of the collected image is low, and it is easy to extract the laser point in the image in the later stage; firstly, the side-view image C1 is binarized to obtain the side-view image C2, and then the side-view image C2 is extracted from the laser point outline. After obtaining the outline of the laser point, take its smallest enclosing circle, and determine the pixel coordinates of the center of the smallest enclosing circle. The pixel coordinates of the center of the circle are the pixel coordinates (u1, v1) of the feature point R of the side-view image. After that, the first industrial camera The side-view image feature point R pixel coordinates (u1, v1) are transmitted to the signal processing system, and the signal processing system transmits the weight data and the side-view image feature point R pixel coordinates (u1, v1) to the second industrial camera;

Step 4: After the second industrial camera receives the weight data and the R pixel coordinates (u1, v1) of the feature point of the side-view image, it triggers the second industrial camera to collect the top-view image of the logistics box. The specific workflow is as follows:

(1) The second industrial camera defaults to the small aperture mode. The small aperture mode can ensure that the captured image has low brightness and is easy to obtain the pixel coordinates of the laser point in the later stage; the laser light source illuminates the logistics box and forms a laser point on the upper surface of the logistics box; The second industrial camera collects the top-view image F1 of the first logistics box, first binarizes the top-view image F1 to obtain the top-view image F2, then extracts the contour of the laser point from the top-view image F2, obtains the contour of the laser point and takes the minimum Enclose the circle, and determine the center pixel coordinates of the smallest encircling circle, the center pixel coordinates are the L pixel coordinates (u2, v2) of the feature point of the overlooking image;

(2) The second industrial camera automatically switches to the large aperture mode to ensure the high brightness of the captured image, and can clearly distinguish the logistics box from the background; the second industrial camera collects the second overhead image F2 of the logistics box, and performs an image analysis on the overhead image F2 The edge is extracted to obtain the top-view image F3, and then the contour is extracted from the top-view image F3, and the interference of background impurities is eliminated by judging the perimeter of the contour among the multiple extracted contours, and finally the top-view image F4 containing only the contour of the logistics box is obtained. , if the number of closed contours in the top-view image F4 is 0, an error message will be returned to the signal processing system, indicating that the logistics box is not completely within the field of view of the first camera and needs to be rearranged; finally, the correct contour of the top-view image F4 is extracted Its minimum circumscribed rectangle, and obtain the image length and width data l and w of the minimum circumscribed rectangle;

Step five, the second industrial camera performs the 3D reconstruction algorithm of laser feature point matching, the specific workflow is as follows:

(1) Through the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, the relationship between the pixel coordinates (u, v) of the laser feature point and the three-dimensional world coordinates (X, Y, Z) can be established. The corresponding relationship, that is:

where A is the internal parameter matrix (R|T) is the external parameter matrix, R is the 3×3 rotation matrix of the camera optical center coordinate system relative to the world coordinate system, and T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system;

(2) The second industrial camera obtains the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, and uses the R pixel coordinates (u1, v1) of the side-view image feature points obtained in step 3 and step 4 ) and the pixel coordinates (u2, v2) of the feature point L of the top-view image are respectively substituted into the relational formula in step 5 (1), and a pair of equations of the pixel coordinates of the laser feature point and the three-dimensional world coordinates of the point can be established, and the least square method Solve the world coordinates (X, Y, Z) to restore the real position of the laser feature point in the world coordinate system, and complete the three-dimensional reconstruction process of laser feature point matching, where Z is the height data of the logistics box;

Step six, the second industrial camera calculates the real length and width data of the logistics box, and the image length and width data l, w obtained in step four (2) and the pre-measured camera height H and camera focal length f, according to the similar triangle in the optical imaging model The formula for calculating:

Among them, pix represents the number of long or wide pixels in the image, Δ represents the pixel size, and the value of the image length and width data l or w is its corresponding pix*Δ, that is, l=pix*Δ or w=pix*Δ, f is the focal length, H is the height of the camera from the measurement platform, and h is the height of the logistics box, so that the actual length and width information L and W of the logistics box can be calculated, then the volume of the logistics box V is V=L*W*Z, the volume of the logistics box Information transmission to the signal processing system;

In step seven, the display screen of the signal processing system displays the volume and weight information of the logistics box, or displays a prompt message indicating that it is placed incorrectly.

Working principle of the present invention:

When working, step 1, start the binocular visual volume and weight measurement system, the first industrial camera, the second industrial camera, laser light source, Bluetooth electronic scale, LED bar light source and signal processing system start to work;

Step 2: Calibrate the internal parameters, external parameters and distortion coefficients of the first industrial camera and the second industrial camera, and transmit the calibration results to the signal processing system. The specific workflow is as follows:

(1) The first industrial camera and the second industrial camera satisfy the pinhole camera model, and the image coordinate vector is set as Where (u, v) is the pixel coordinates of the target point; the internal parameter matrix of the camera is Where f _x , f _y , c _x , c _y are the x-direction focal length, y-direction focal length and optical axis center coordinates respectively; the camera extrinsic parameter matrix is (R|T), where R is the camera optical center coordinate system relative to the world coordinate system The 3×3 rotation matrix, T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system; the world coordinate vector is Among them, X, Y, and Z are the world coordinates of the target point; the camera imaging model satisfies the relation:

where z _c is the scale factor; in addition, the camera distortion model satisfies the following relationship:

Among them, (x, y) are the physical coordinates of the image before distortion correction, (x _cor , y _cor ) are the physical coordinates of the image after distortion correction, r=x ² +y ² ,k ₁ ,k ₂ ,k ₃ ,p ₁ , p ₂ are three radial distortion coefficients and two tangential distortion coefficients of the camera;

The purpose of the calibration of the first industrial camera and the second industrial camera is to solve the internal parameters, external parameters and distortion coefficient of the camera. The measurement platform is set as the zero plane, and Zhang Zhengyou’s checkerboard plane calibration method is used to take 20 images by changing the position and angle of the checkerboard. , to calibrate the camera, and solve to obtain the camera internal parameters, external parameters and distortion coefficients;

Step 3, place the logistics box on the measurement platform, the laser light source illuminates the logistics box, and forms a laser spot on the upper surface of the logistics box, the Bluetooth electronic scale detects the weight of the logistics box, and when the data of the Bluetooth electronic scale is stable, record the logistics box Weight and transmit the data to the signal processing system through the Bluetooth module on the electronic scale, and then the signal processing system transmits the weight data to the first industrial camera and triggers it to collect the side view image C1 of the logistics box. The first industrial camera defaults to a small aperture mode, which can ensure that the brightness of the collected image is low, and it is easy to extract the laser point in the image in the later stage; firstly, the side-view image C1 is binarized to obtain the side-view image C2, and then the side-view image C2 is extracted from the laser point outline. After obtaining the outline of the laser point, take its smallest enclosing circle, and determine the pixel coordinates of the center of the smallest enclosing circle. The pixel coordinates of the center of the circle are the pixel coordinates (u1, v1) of the feature point R of the side-view image. After that, the first industrial camera The side-view image feature point R pixel coordinates (u1, v1) are transmitted to the signal processing system, and the signal processing system transmits the weight data and the side-view image feature point R pixel coordinates (u1, v1) to the second industrial camera;

Step 4: After the second industrial camera receives the weight data and the R pixel coordinates (u1, v1) of the feature point of the side-view image, it triggers the second industrial camera to collect the top-view image of the logistics box. The specific workflow is as follows:

(1) The second industrial camera defaults to the small aperture mode. The small aperture mode can ensure that the captured image has low brightness and is easy to obtain the pixel coordinates of the laser point in the later stage; the laser light source illuminates the logistics box and forms a laser point on the upper surface of the logistics box; The second industrial camera collects the top-view image F1 of the first logistics box, first binarizes the top-view image F1 to obtain the top-view image F2, then extracts the contour of the laser point from the top-view image F2, obtains the contour of the laser point and takes the minimum Enclose the circle, and determine the center pixel coordinates of the smallest encircling circle, the center pixel coordinates are the L pixel coordinates (u2, v2) of the feature point of the overlooking image;

(2) The second industrial camera automatically switches to the large aperture mode to ensure the high brightness of the captured image, and can clearly distinguish the logistics box from the background; the second industrial camera collects the second overhead image F2 of the logistics box, and performs an image analysis on the overhead image F2 The edge is extracted to obtain the top-view image F3, and then the contour is extracted from the top-view image F3, and the interference of background impurities is eliminated by judging the perimeter of the contour among the multiple extracted contours, and finally the top-view image F4 containing only the contour of the logistics box is obtained. , if the number of closed contours in the top-view image F4 is 0, an error message will be returned to the signal processing system, indicating that the logistics box is not completely within the field of view of the first camera and needs to be rearranged; finally, the correct contour of the top-view image F4 is extracted Its minimum circumscribed rectangle, and obtain the image length and width data l and w of the minimum circumscribed rectangle;

Step five, the second industrial camera performs the 3D reconstruction algorithm of laser feature point matching, the specific workflow is as follows:

(1) Through the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, the relationship between the pixel coordinates (u, v) of the laser feature point and the three-dimensional world coordinates (X, Y, Z) can be established. The corresponding relationship, that is:

where A is the internal parameter matrix (R|T) is the external parameter matrix, R is the 3×3 rotation matrix of the camera optical center coordinate system relative to the world coordinate system, and T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system;

(2) The second industrial camera obtains the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, and uses the R pixel coordinates (u1, v1) of the side-view image feature points obtained in step 3 and step 4 ) and the pixel coordinates (u2, v2) of the feature point L of the top-view image are respectively substituted into the relational formula in step 5 (1), and a pair of equations of the pixel coordinates of the laser feature point and the three-dimensional world coordinates of the point can be established, and the least square method Solve the world coordinates (X, Y, Z) to restore the real position of the laser feature point in the world coordinate system, and complete the three-dimensional reconstruction process of laser feature point matching, where Z is the height data of the logistics box;

Step six, the second industrial camera calculates the real length and width data of the logistics box, and the image length and width data l, w obtained in step four (2) and the pre-measured camera height H and camera focal length f, according to the similar triangle in the optical imaging model The formula for calculating:

Among them, pix represents the number of long or wide pixels in the image, Δ represents the pixel size, and the value of the image length and width data l or w is its corresponding pix*Δ, that is, l=pix*Δ or w=pix*Δ, f is the focal length, H is the height of the camera from the measurement platform, and h is the height of the logistics box, so that the actual length and width information L and W of the logistics box can be calculated, then the volume of the logistics box V is V=L*W*Z, the volume of the logistics box Information transmission to the signal processing system;

In step seven, the display screen of the signal processing system displays the volume and weight information of the logistics box, or displays a prompt message indicating that it is placed incorrectly.

Compared with the prior art, the present invention has the following beneficial effects:

(1) Compared with the traditional manual scale measurement method, the present invention has the advantages of fastness, accuracy, non-contact and low cost, and is suitable for automatic batch measurement, which greatly reduces labor costs and improves production efficiency;

(2) The present invention does not need to use the traditional laser ranging module, and can realize the measurement of the length, width and height of the object entirely by visual technology. Higher technical precision and lower cost;

(3) Compared with the problems affected by factors such as illumination, texture, optical distortion and noise involved in the general three-dimensional reconstruction feature point matching, the laser light source-guided stereo matching algorithm adopted in the present invention greatly reduces the problem of feature point matching. It reduces the difficulty of mismatching and missing matching, and improves the stability and reliability of the system.

Description of drawings

Fig. 1 is the overall structure schematic diagram of the present invention;

Fig. 2 is a schematic structural diagram of the system of the present invention;

Fig. 3 is a three-dimensional reconstruction algorithm flowchart of the present invention;

Fig. 4 is a schematic diagram of calculating the length and width of the logistics box of the present invention.

Reference signs in the figure are: 1. First industrial camera; 2. Signal processing system; 3. LED bar light source; 4. Bluetooth electronic scale; 5. Measurement platform; 6. Second industrial camera; 7. Laser light source; 8. Laser point; 9. Logistics box.

Detailed ways

The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

As shown in Figures 1 to 4, a binocular visual volumetric weight measurement system includes an installation cabinet, and a first industrial camera 1, a second industrial camera 6, a laser light source 7, and an LED strip installed in the installation cabinet. Light source 3, signal processing system 2, measuring platform 5 and bluetooth electronic scale 4, wherein, described measuring platform 5 is arranged on the bottom of described installation cabinet, and described measuring platform 5 is polished into black smooth surface through painting, can be arranged like this Ensure that the logistics box 9 enters and exits the measurement platform 5 smoothly and smoothly, and can reduce background interference in the image processing process; the Bluetooth electronic scale 4 is arranged below the measurement platform 5; the second industrial camera 6 is arranged on the The top of the installation cabinet, the laser light source 7 and the second industrial camera 6 are arranged side by side, and the light direction of the laser light source 7 is kept parallel to the shooting direction of the second industrial camera 6; the second industrial The camera 6 and the laser light source 7 are arranged directly above the measurement platform 5, and the shooting direction of the second industrial camera 6 and the light direction of the laser light source 7 are all perpendicular to the measurement platform 5 at the same time; The first industrial camera 1 is set on the top of the installation cabinet, the first industrial camera 1 is located obliquely above the measurement platform 5, and the first industrial camera 1 is at an angle of 45 degrees to the measurement platform 5 Inclined setting; the first industrial camera 1 and the second industrial camera 6 have the same field of view of the measurement platform 5; the LED bar light source 3 is provided with four, of which two LED bar light sources 3 are arranged horizontally On the front and rear sides of the second industrial camera 6, two other LED strip light sources 3 are vertically arranged on the front and rear sides of the first industrial camera 1, and the setting of these LED strip light sources 3 can realize uniform lighting For the first industrial camera 1 and the second industrial camera 6 to obtain high-quality images; the first industrial camera 1, the second industrial camera 6, the laser light source 7 and the Bluetooth electronic scale 4 are respectively connected to the signal processing system 2; The signal processing system 2 is provided with a display screen.

Binocular stereo vision is one of the key technologies of computer vision. It is based on the principle of parallax and uses imaging equipment to obtain two images of the measured object from different positions. By calculating the position deviation between the corresponding points of the image, the three-dimensional geometric information of the object is obtained. The method using computer vision is not affected by artificial factors, and the accuracy is not limited by the accuracy of reference objects such as measuring scales. Compared with traditional manual scale measurement, it is fast, accurate, non-contact and low cost. Advantages, suitable for batch automated measurement, greatly reducing labor costs and improving production efficiency. The binocular vision measurement technology proposed by the invention has the advantages of high efficiency, appropriate precision, simple system structure, low cost, etc., and is very suitable for on-line, non-contact product detection and quality control at the manufacturing site.

When working, step 1 starts the binocular visual volume and weight measurement system, and the first industrial camera 1, the second industrial camera 6, the laser light source 7, the Bluetooth electronic scale 4, the LED bar light source 3 and the signal processing system 2 start to work;

Step 2: Calibrate the internal parameters, external parameters and distortion coefficients of the first industrial camera 1 and the second industrial camera 6, and transmit the calibration results to the signal processing system 2. The specific workflow is as follows:

(1) The first industrial camera 1 and the second industrial camera 6 satisfy the pinhole camera model, assuming that the image coordinate vector is Where (u, v) is the pixel coordinates of the target point; the internal parameter matrix of the camera is Where f _x , f _y , c _x , c _y are the x-direction focal length, y-direction focal length and optical axis center coordinates respectively; the camera extrinsic parameter matrix is (R|T), where R is the camera optical center coordinate system relative to the world coordinate system The 3×3 rotation matrix, T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system; the world coordinate vector is Where X, Y, Z are the world coordinates of the target point;

The camera imaging model satisfies the relation:

where z _c is the scale factor; in addition, the camera distortion model satisfies the following relationship:

Among them, (x, y) are the physical coordinates of the image before distortion correction, (x _cor , y _cor ) are the physical coordinates of the image after distortion correction, r=x ² +y ² ,k ₁ ,k ₂ ,k ₃ ,p ₁ , p ₂ are three radial distortion coefficients and two tangential distortion coefficients of the camera;

The purpose of the calibration of the first industrial camera 1 and the second industrial camera 6 is to solve the internal parameters, external parameters and distortion coefficients of the cameras. The measurement platform 5 is set as the zero plane, and Zhang Zhengyou’s checkerboard plane calibration method is used to take pictures by changing the position and angle of the checkerboard. 20 images were used to calibrate the camera, and the internal parameters, external parameters and distortion coefficients of the camera were obtained by solving;

Step 3, the logistics box 9 is placed on the measurement platform 5, the laser light source 7 irradiates the logistics box 9, and a laser spot 8 is formed on the upper surface of the logistics box 9, and the Bluetooth electronic scale 4 detects the weight of the logistics box 9. When the Bluetooth electronic scale When the data of 4 is stable, record the weight of the logistics box 9 and transmit the data to the signal processing system 2 through the Bluetooth module on the electronic scale, and then the signal processing system 2 transmits the weight data to the first industrial camera 1 and triggers it to collect the logistics box 9 The side-view image C1 of the first industrial camera 1 defaults to the small aperture mode, which can ensure that the brightness of the collected image is low, and it is easy to extract the laser point 8 in the image in the later stage; first, binarize the side-view image C1 to obtain the side-view image View the image C2, and then extract the contour of the laser point 8 from the side view image C2. After obtaining the contour of the laser point 8, take the smallest enclosing circle, and determine the pixel coordinates of the center of the smallest enclosing circle. The pixel coordinates of the center of the circle are the side view The pixel coordinates (u1, v1) of the image feature point R, and then the first industrial camera 1 transmits the pixel coordinates (u1, v1) of the side view image feature point R to the signal processing system 2, and the signal processing system 2 combines the weight data and the side view Image feature point R pixel coordinates (u1, v1) are transmitted to the second industrial camera 6;

Step 4, after the second industrial camera 6 receives the weight data and the R pixel coordinates (u1, v1) of the feature point of the side-view image, it triggers the second industrial camera 6 to collect the overhead image of the logistics box 9, and the specific workflow is as follows:

(1) The second industrial camera 6 defaults to the small aperture mode. The small aperture mode can ensure that the brightness of the image collected is low, and it is easy to obtain the pixel coordinates of the laser point 8 in the later stage; A laser point 8 is formed on the surface; the second industrial camera 6 collects the first overhead image F1 of the logistics box 9, and first binarizes the overhead image F1 to obtain an overhead image F2, and then extracts the outline of the laser point 8 from the overhead image F2, After obtaining the outline of the laser point 8, get its minimum enclosing circle, and determine the center pixel coordinates of the minimum enclosing circle, and this circle center pixel coordinate is the L pixel coordinate (u2, v2) of the feature point L of the bird's-eye view image;

(2) The second industrial camera 6 is automatically converted into a large aperture mode to ensure that the image brightness collected is high, and the logistics box 9 and the background can be clearly distinguished; the second industrial camera 6 collects the second overhead image F2 of the logistics box 9, and the Perform edge extraction on the top-view image F2 to obtain the top-view image F3, then perform contour extraction on the top-view image F3 and eliminate the interference of background impurities by judging the perimeter of the contours among the multiple extracted contours, and finally obtain a top-view image that only contains the contour of the logistics box 9 Image F4, at this time, if the number of closed contours in the top view image F4 is 0, an error message will be returned to the signal processing system 2, indicating that the logistics box 9 is not completely within the field of view of the first camera, and needs to be rearranged; finally correct Extract its minimum circumscribed rectangle from the contour of the bird's-eye view image F4, and obtain the image length and width data l and w of the minimum circumscribed rectangle;

Step five, the second industrial camera 6 performs a three-dimensional reconstruction algorithm for laser feature point matching, and the specific workflow is as follows:

(1) Through the internal parameters and external parameters obtained by the calibration of the first industrial camera 1 and the second industrial camera 6 in step 2, the pixel coordinates (u, v) and three-dimensional world coordinates (X, Y, Z) of laser feature points can be established The corresponding relationship between, namely:

where A is the internal parameter matrix (R|T) is the external parameter matrix, R is the 3×3 rotation matrix of the camera optical center coordinate system relative to the world coordinate system, and T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system;

(2) the second industrial camera 6 acquires the internal parameters and external parameters that the first industrial camera 1 and the second industrial camera 6 calibrate in step 2, and utilizes the side-view image feature point R pixel coordinates obtained in step 3 and step 4 ( u1, v1) and L pixel coordinates (u2, v2) of the feature point of the top-view image are respectively substituted into the relational expression in step 5 (1), and a pair of equations of the pixel coordinates of the laser feature point and the three-dimensional world coordinates of the point can be established, through The least square method solves the world coordinates (X, Y, Z) to restore the real position of the laser feature point in the world coordinate system, and completes the three-dimensional reconstruction process of laser feature point matching, where Z is the height data of the logistics box 9;

Step six, the second industrial camera 6 calculates the real length and width data of the logistics box 9, the image length and width data l, w obtained in step four (2) and the pre-measured camera height H and camera focal length f, according to the optical imaging model The calculation formula of similar triangles:

Among them, pix represents the number of long or wide pixels in the image, Δ represents the pixel size, and the value of the image length and width data l or w is its corresponding pix*Δ, that is, l=pix*Δ or w=pix*Δ, f is the focal length, H is the height of the camera from the measurement platform 5, and h is the height of the logistics box 9, so that the actual length and width information L and W of the logistics box 9 can be calculated, and the volume V of the logistics box 9 is V=L*W*Z, Transmitting the volume information of the logistics box 9 to the signal processing system 2;

Step 7, the display screen of the signal processing system 2 displays the volume and weight information of the logistic box 9, or displays a prompt message indicating that it is placed incorrectly.

Compared with the traditional manual scale measurement method, the present invention has the advantages of fast, accurate, non-contact and low cost, and is suitable for batch automatic measurement, greatly reducing labor costs and improving production efficiency; no need to use traditional laser ranging modules , the length, width and height of an object can be measured entirely by visual technology. In a short-distance, small-volume measurement environment, the binocular vision 3D reconstruction technology is more accurate and less costly than the laser ranging technology; The problems involved in the three-dimensional reconstruction feature point matching are affected by factors such as illumination, texture, optical distortion, and noise. The laser light source-guided stereo matching algorithm adopted in the present invention greatly reduces the difficulty of feature point matching and reduces errors. Problems such as matching and missing matching improve the stability and reliability of the system.

The above is a preferred embodiment of the present invention, but the embodiment of the present invention is not limited by the above content, and any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention are all Replacement methods that should be equivalent are all included within the protection scope of the present invention.

Claims (3)

1. A binocular visual volumetric weight measurement system, characterized in that it comprises an installation cabinet, and the first industrial camera, the second industrial camera, a laser light source, an LED strip light source, and a signal processing system arranged in the installation cabinet , a measurement platform and a bluetooth electronic scale, wherein the measurement platform is arranged at the bottom of the installation cabinet, and the bluetooth electronic scale is arranged below the measurement platform; the second industrial camera is arranged at the bottom of the installation cabinet At the top, the laser light source and the second industrial camera are arranged side by side, and the light direction of the laser light source is kept parallel to the shooting direction of the second industrial camera; the second industrial camera and the laser light source are arranged on directly above the measurement platform, and the shooting direction of the second industrial camera and the light direction of the laser light source are both perpendicular to the measurement platform; the first industrial camera is arranged on the top of the installation cabinet, The first industrial camera is located obliquely above the measurement platform, and the first industrial camera is inclined at an angle of 45 degrees to the measurement platform; the first industrial camera and the second industrial camera photograph the measurement platform The field of view range is the same; there are four LED bar light sources, two of which are horizontally arranged on the front and rear sides of the second industrial camera, and the other two LED bar light sources are vertically arranged on the The front and rear sides of the first industrial camera; The first industrial camera, the second industrial camera, the laser light source and the bluetooth electronic scale are respectively connected to the signal processing system; the signal processing system is provided with a display screen. 2. The binocular vision volumetric weight measurement system according to claim 1, wherein the measurement platform is painted and polished into a black smooth surface. 3. A method for realizing the binocular vision volume weight measuring system according to any one of claims 1 to 2, characterized in that it comprises the following steps: Step 1, start the binocular vision volume and weight measurement system, the first industrial camera, the second industrial camera, laser light source, Bluetooth electronic scale, LED bar light source and signal processing system start to work; Step 2: Calibrate the internal parameters, external parameters and distortion coefficients of the first industrial camera and the second industrial camera, and transmit the calibration results to the signal processing system. The specific workflow is as follows: (1) The first industrial camera and the second industrial camera satisfy the pinhole camera model, and the image coordinate vector is set as Where (u, v) is the pixel coordinates of the target point; the internal parameter matrix of the camera is Where f _x , f _y , c _x , c _y are the x-direction focal length, y-direction focal length and optical axis center coordinates respectively; the camera extrinsic parameter matrix is (R|T), where R is the camera optical center coordinate system relative to the world coordinate system The 3×3 rotation matrix, T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system; the world coordinate vector is Among them, X, Y, and Z are the world coordinates of the target point; the camera imaging model satisfies the relation: <mrow><msup><mi>m</mi><mo>&amp;prime;</mo></msup><mo>=</mo><mfrac><mn>1</mn><msub><mi>z</mi><mi>c</mi></msub></mfrac><mi>A</mi><mrow><mo>(</mo><mi>R</mi><mo>|</mo><mi>T</mi><mo>)</mo></mrow><msup><mi>M</mi><mo>&amp;prime;</mo></msup><mo>,</mo></mrow> where z _c is the scale factor; in addition, the camera distortion model satisfies the following relationship: <mrow><mfenced open = "(" close = ")"><mtable><mtr><mtd><msub><mi>x</mi><mrow><mi>c</mi><mi>o</mi><mi>r</mi></mrow></msub></mtd></mtr><mtr><mtd><msub><mi>y</mi><mrow><mi>c</mi><mi>o</mi><mi>r</mi></mrow></msub></mtd></mtr></mtable></mfenced><mo>=</mo><mrow><mo>(</mo><mn>1</mn><mo>+</mo><msub><mi>k</mi><mn>1</mn></msub><msup><mi>r</mi><mn>2</mn></msup><mo>+</mo><msub><mi>k</mi><mn>2</mn></msub><msup><mi>r</mi><mn>4</mn></msup><mo>+</mo><msub><mi>k</mi><mn>3</mn></msub><msup><mi>r</mi><mn>6</mn></msup><mo>)</mo></mrow><mfenced open = "(" close = ")"><mtable><mtr><mtd><mi>x</mi></mtd></mtr><mtr><mtd><mi>y</mi></mtd></mtr></mtable></mfenced><mo>+</mo><mfenced open = "(" close = ")"><mtable><mtr><mtd><mn>2</mn><msub><mi>p</mi><mn>1</mn></msub><mi>x</mi><mi>y</mi><mo>+</mo><msub><mi>p</mi><mn>2</mn></msub><mo>(</mo><msup><mi>r</mi><mn>2</mn></msup><mo>+</mo><mn>2</mn><msup><mi>x</mi><mn>2</mn></msup><mo>)</mo></mtd></mtr><mtr><mtd><msub><mi>p</mi><mn>1</mn></msub><mo>(</mo><msup><mi>r</mi><mn>2</mn></msup><mo>+</mo><mn>2</mn><msup><mi>y</mi><mn>2</mn></msup><mo>)</mo><mo>+</mo><mn>2</mn><msub><mi>p</mi><mn>2</mn></msub><mi>x</mi><mi>y</mi></mtd></mtr></mtable></mfenced><mo>,</mo></mrow> Among them, (x, y) are the physical coordinates of the image before distortion correction, (x _cor , y _cor ) are the physical coordinates of the image after distortion correction, r=x ² +y ² ,k ₁ ,k ₂ ,k ₃ ,p ₁ , p ₂ are three radial distortion coefficients and two tangential distortion coefficients of the camera; The purpose of the calibration of the first industrial camera and the second industrial camera is to solve the internal parameters, external parameters and distortion coefficient of the camera. The measurement platform is set as the zero plane, and Zhang Zhengyou’s checkerboard plane calibration method is used to take 20 images by changing the position and angle of the checkerboard. , to calibrate the camera, and solve to obtain the camera internal parameters, external parameters and distortion coefficients; Step 3, place the logistics box on the measurement platform, the laser light source illuminates the logistics box, and forms a laser spot on the upper surface of the logistics box, the Bluetooth electronic scale detects the weight of the logistics box, and when the data of the Bluetooth electronic scale is stable, record the logistics box Weight and transmit the data to the signal processing system through the Bluetooth module on the electronic scale, and then the signal processing system transmits the weight data to the first industrial camera and triggers it to collect the side view image C1 of the logistics box. The first industrial camera defaults to a small aperture mode, which can ensure that the brightness of the collected image is low, and it is easy to extract the laser point in the image in the later stage; firstly, the side-view image C1 is binarized to obtain the side-view image C2, and then the side-view image C2 is extracted from the laser point outline. After obtaining the outline of the laser point, take its smallest enclosing circle, and determine the pixel coordinates of the center of the smallest enclosing circle. The pixel coordinates of the center of the circle are the pixel coordinates (u1, v1) of the feature point R of the side-view image. After that, the first industrial camera The side-view image feature point R pixel coordinates (u1, v1) are transmitted to the signal processing system, and the signal processing system transmits the weight data and the side-view image feature point R pixel coordinates (u1, v1) to the second industrial camera; Step 4: After the second industrial camera receives the weight data and the R pixel coordinates (u1, v1) of the feature point of the side-view image, it triggers the second industrial camera to collect the top-view image of the logistics box. The specific workflow is as follows: (1) The second industrial camera defaults to the small aperture mode. The small aperture mode can ensure that the captured image has low brightness and is easy to obtain the pixel coordinates of the laser point in the later stage; the laser light source illuminates the logistics box and forms a laser point on the upper surface of the logistics box; The second industrial camera collects the top-view image F1 of the first logistics box, first binarizes the top-view image F1 to obtain the top-view image F2, then extracts the contour of the laser point from the top-view image F2, obtains the contour of the laser point and takes the minimum Enclose the circle, and determine the center pixel coordinates of the smallest encircling circle, the center pixel coordinates are the L pixel coordinates (u2, v2) of the feature point of the overlooking image; (2) The second industrial camera automatically switches to the large aperture mode to ensure the high brightness of the captured image, and can clearly distinguish the logistics box from the background; the second industrial camera collects the second overhead image F2 of the logistics box, and performs an image analysis on the overhead image F2 The edge is extracted to obtain the top-view image F3, and then the contour is extracted from the top-view image F3, and the interference of background impurities is eliminated by judging the perimeter of the contour among the multiple extracted contours, and finally the top-view image F4 containing only the contour of the logistics box is obtained. , if the number of closed contours in the top-view image F4 is 0, an error message will be returned to the signal processing system, indicating that the logistics box is not completely within the field of view of the first camera and needs to be rearranged; finally, the correct contour of the top-view image F4 is extracted Its minimum circumscribed rectangle, and obtain the image length and width data l and w of the minimum circumscribed rectangle; Step five, the second industrial camera performs the 3D reconstruction algorithm of laser feature point matching, the specific workflow is as follows: (1) Through the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, the relationship between the pixel coordinates (u, v) of the laser feature point and the three-dimensional world coordinates (X, Y, Z) can be established. The corresponding relationship, that is: <mrow><mi>Z</mi><mi>c</mi><mfenced open = "[" close = "]"><mtable><mtr><mtd><mi>u</mi></mtd></mtr><mtr><mtd><mi>v</mi></mtd></mtr><mtr><mtd><mn>1</mn></mtd></mtr></mtable></mfenced><mo>=</mo><mi>A</mi><mrow><mo>(</mo><mi>R</mi><mo>|</mo>mo><mi>T</mi><mo>)</mo></mrow><mfenced open = "[" close = "]"><mtable><mtr><mtd><mi>X</mi></mtd></mtr><mtr><mtd><mi>Y</mi></mtd></mtr><mtr><mtd><mi>Z</mi></mtd></mtr><mtr><mtd><mn>1</mn></mtd></mtr></mtable></mfenced><mo>,</mo></mrow> where A is the internal parameter matrix (R|T) is the external parameter matrix, R is the 3×3 rotation matrix of the camera optical center coordinate system relative to the world coordinate system, and T is the 3×1 translation matrix of the camera optical center coordinate system relative to the world coordinate system; (2) The second industrial camera obtains the internal parameters and external parameters obtained by the calibration of the first industrial camera and the second industrial camera in step 2, and uses the R pixel coordinates (u1, v1) of the side-view image feature points obtained in step 3 and step 4 ) and the pixel coordinates (u2, v2) of the feature point L of the top-view image are respectively substituted into the relational formula in step 5 (1), and a pair of equations of the pixel coordinates of the laser feature point and the three-dimensional world coordinates of the point can be established, and the least square method Solve the world coordinates (X, Y, Z) to restore the real position of the laser feature point in the world coordinate system, and complete the three-dimensional reconstruction process of laser feature point matching, where Z is the height data of the logistics box; Step six, the second industrial camera calculates the real length and width data of the logistics box, and the image length and width data l, w obtained in step four (2) and the pre-measured camera height H and camera focal length f, according to the similar triangle in the optical imaging model The formula for calculating: <mrow><mfrac><mrow><mi>p</mi><mi>i</mi><mi>x</mi><mo>*</mo><mi>&amp;Delta;</mi></mrow><mrow><mi>L</mi><mrow><mo>(</mo><mi>W</mi><mo>)</mo></mrow></mrow></mfrac><mo>=</mo><mfrac><mi>f</mi><mrow><mi>H</mi><mo>-</mo><mi>h</mi></mrow></mfrac><mo>,</mo></mrow> Among them, pix represents the number of long or wide pixels in the image, Δ represents the pixel size, and the value of the image length and width data l or w is its corresponding pix*Δ, that is, l=pix*Δ or w=pix*Δ, f is the focal length, H is the height of the camera from the measurement platform, and h is the height of the logistics box, so that the actual length and width information L and W of the logistics box can be calculated, then the volume of the logistics box V is V=L*W*Z, the volume of the logistics box Information transmission to the signal processing system; In step seven, the display screen of the signal processing system displays the volume and weight information of the logistics box, or displays a prompt message indicating that it is placed incorrectly.