A corrugated cardboard visual counting device and method

Abstract

The present invention discloses a corrugated cardboard visual counting device and method, the device comprises an image acquisition module, a distance measurement module, a power supply module, a control module, a display module and a lifting module; the image acquisition module, the distance measurement module, the power supply module, the control module and the display module are all integrated in a shell, and the shell is rotatably mounted on the lifting module; the image acquisition module is connected to the control module, and is used to acquire the stacked image of the corrugated cardboard and send it to the control module; the distance measurement module is connected to the control module, and is used to detect the shooting distance and send it to the control module; the control module is connected to the display module, and is used to obtain the stacked number of the corrugated cardboard according to the stacked image and the shooting distance and send it to the display module for display. The present invention has the advantages of high integration, easy to carry, high counting efficiency, high accuracy, convenient operation, wide applicability, etc.

Description

Technical Field

[0001] The invention mainly relates to the technical field of corrugated cardboard production, and in particular to a corrugated cardboard visual counting device and method. Background Art

[0002] Corrugated cardboard is a multi-layer adhesive body composed of wavy core paper interlayer and cardboard. It has good cushioning and compression and shock resistance. It can be used to prevent damage to goods during transportation. Corrugated cardboard is low-priced and light-weight, so it is widely used in the packaging and transportation industry.

[0003] When corrugated cardboard is mass-produced, it is necessary to count the stacks of corrugated cardboard. Common counting methods include manual counting, photoelectric induction physical trigger counting, and machine vision-based image recognition counting. Among them, manual counting is inefficient and costly; physical trigger counting equipment is generally large and difficult to move; although existing machine vision counting devices can achieve non-contact measurement of corrugated cardboard stacks, they generally have the following problems: (1) The core computing module mostly uses a large industrial computer, which is inconvenient to move and has high energy consumption; (2) Due to cost constraints, most of them use a small number of cameras for long-distance, large-field non-parallel imaging, so that some stacks with uneven end faces are easily blocked due to tilted vision; (3) The image processing methods of existing visual counting devices are mostly to transform the grayscale features of the corrugated cardboard stack image and then use morphological processing to extract straight lines or holes in the image and then count the results. This type of image recognition algorithm is mostly only suitable for counting regular cardboard stack images in standard environments such as assembly lines. It is often less effective for counting cardboard under non-ideal conditions such as uneven stacking, low light brightness, and unclear imaging in loading and transportation. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in view of the problems existing in the prior art, the present invention provides a highly integrated and easy-to-carry corrugated cardboard visual counting device, and correspondingly provides a visual counting method with high counting efficiency, high accuracy and convenient operation.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0006] A corrugated cardboard visual counting device comprises an image acquisition module, a distance measurement module, a power supply module, a control module, a display module and a lifting module; the image acquisition module, the distance measurement module, the power supply module, the control module and the display module are all integrated in a shell, and the shell is rotatably mounted on the lifting module; the image acquisition module is connected to the control module, and is used to acquire a stacked image of the corrugated cardboard and send it to the control module; the distance measurement module is connected to the control module, and is used to detect a shooting distance and send it to the control module; the control module is connected to the display module, and is used to obtain the stacked number of the corrugated cardboard according to the stacked image and the shooting distance, and send it to the display module for display.

[0007] Preferably, the control module is a Raspberry Pi microcomputer.

[0008] Preferably, the image acquisition module is an area array camera.

[0009] The present invention also discloses a visual counting method based on the above-mentioned corrugated cardboard visual counting device, comprising the steps of:

[0010] 1) The images of the cardboard stack at different positions are collected by the image collection module, and a complete cardboard stack image is obtained by splicing the images of the cardboard stack at different positions;

[0011] 2) Cut out the middle image of a number of columns of pixels from the complete cardboard stack image, and then extract the cardboard area image;

[0012] 3) Enhance the cardboard holes in the cardboard area image to obtain a hole enhancement image;

[0013] 4) Obtain the center position of the cardboard hole in the hole enhancement image, obtain the number of local peaks based on the center position of the cardboard hole, and finally obtain the total number of cardboards.

[0014] Preferably, the specific process of step 1) is:

[0015] 1.1) The image acquisition module collects images of cardboard stacks in different directions to form an image sequence;

[0016] 1.2) Extract and match feature points of image sequences;

[0017] 1.3) Construct the transformation matrix between the image sequences through matching points to complete the stitching of the panoramic image;

[0018] 1.4) The corresponding images are transformed according to the transformation matrix of the image sequence to determine the overlapping area between the images, and the image to be fused is mapped to a new blank image to form a mosaic.

[0019] Preferably, in step 2), the specific process of extracting the cardboard area image is:

[0020] 2.1) Firstly, the image is processed by adaptive histogram equalization to enhance the image contrast, and then a convolution operation is performed with a specific result operator to obtain a cardboard bevel edge detection image, and finally a cardboard bevel edge enhancement image is obtained;

[0021] 2.2) Slide up and down in the cardboard bevel enhancement image to calculate the sum of the pixel values ​​in each area to determine the upper and lower boundary y coordinates of the cardboard area upper and lower ;

[0022] 2.3) The cardboard gap is covered by generating a corresponding cardboard gap mask image to obtain the final cardboard area image.

[0023] Preferably, the specific process of step 2.2) is as follows: assuming that the sum of the pixel values ​​of the i-th row in the cardboard bevel enhancement image is S i , select threshold ξ1, start traversing in positive order from the first row, when S i >ξ1 and S i+5 >ξ1, if and This line is the upper boundary of the cardboard area; similarly, the lower boundary of the cardboard area is determined by traversing in reverse order.

[0024] Preferably, the specific process of step 2.3) is: intercepting the row range from y to upper to y lower The cardboard area is traversed, and the sum of the pixel values ​​of each row is traversed. The gap area and the non-gap area are distinguished according to the threshold ξ2. All the pixels belonging to the gap position and the pixel points of the non-gap position are set to 0 and 255 respectively, and the cardboard area mask map is obtained. The row range from y upper to y lower The image is logically ANDed with the cardboard area mask image to obtain the final cardboard area image.

[0025] Preferably, the specific process of step 3) is:

[0026] For each pixel point p in the cardboard area map, the Hessian matrix at multiple different scales is calculated, and three different Gaussian function variances σ are determined as the scale of the Hessian matrix according to the main size of the cardboard hole;

[0027] Use the following filter function to enhance hole features:

[0028]

[0029] where λ ρ is the regularized λ2:

[0030]

[0031] Finally, the maximum filter response in scale space is F(p)=max σ V[eigH(p,σ)], at this time, the hole enhancement map is obtained.

[0032] Preferably, the specific process of step 4) is:

[0033] After using the LoG operator to suppress the noise of the hole enhancement image, the center position of each hole is extracted to obtain the hole center map;

[0034] After all the pixels in the hole center map are projected toward the Y axis, mean fuzzy processing is performed to obtain the hole center projection map;

[0035] The hole center projection image is processed by the second-order difference idea to count the number of local peaks, and the number of local peaks in the projection image is quickly counted through multiple rounds of iterative calculations; the number of local peaks is the number of flutes in the cardboard stack;

[0036] The total number of cardboards is obtained based on the number of local peaks and the cardboard prism shape.

[0037] Compared with the prior art, the advantages of the present invention are:

[0038] The overall hardware structure of the present invention is simple, low-cost, and easy to use. Since the embedded device is small in size, low in energy consumption, easy to carry, and high in work efficiency, it is suitable for moving the counting device at any time on the factory production line according to usage needs to detect and calculate the number of cardboards in real time.

[0039] In the image acquisition method proposed by the present invention, a counting device is moved on a portable lifting bracket to collect cardboard images in different directions, and then these images are spliced ​​to solve the problem of individual cardboard occlusion in imaging, and at the same time solve the problem that a single camera cannot capture all higher cardboard stacks in the past. Therefore, it is also well applicable to cardboard stacks of various heights.

[0040] The counting method proposed in the present invention effectively extracts hole information based on the characteristics of the hole shape in the cardboard stacking image and the difference between the second-order grayscale gradient of the hole area and the edge. It also has good detection effect on images with blurred cardboard edges and images with low brightness, solving the problem that the detection results of the prior art are greatly affected by imaging clarity and lighting.

[0041] The cardboard detection method of the present invention has high precision and wide adaptability, and has high accuracy in detecting different types of cardboard stacking patterns; and the counting process has small amount of calculation, short time consumption, stable performance, and can meet the high real-time requirements of lightweight embedded systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the structure of the device of the present invention in an embodiment.

[0043] Figure 2 It is an embodiment diagram of the device of the present invention in a specific application.

[0044] Figure 3 It is a structural schematic diagram of the cross section of the corrugated paperboard in the present invention.

[0045] Figure 4 It is a structural operator graph used to detect the bevel edge features of the cardboard area in the counting method of the present invention.

[0046] Figure 5 It is a schematic diagram of detecting the bevel edge feature in a poorly imaged cardboard image by the counting method of the present invention.

[0047] Figure 6 It is a schematic diagram of detecting cardboard holes in a poorly imaged cardboard image using the counting method of the present invention.

[0048] Figure 7 It is a schematic diagram of the projection of the center of the hole in the counting method of the present invention.

[0049] Figure 8 It is a schematic diagram of the detection effect of the present invention on different types of cardboard pictures.

[0050] Fig. 9 It is a flow chart of the counting method of the present invention in an embodiment.

[0051] Legend: 1. Image acquisition module; 2. Distance measurement module; 3. Power supply module; 4. Control module; 5. Display module; 6. Lifting module; 7. Fill light; 8. Shell. DETAILED DESCRIPTION

[0052] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0053] like Figure 1 and Figure 2As shown, the corrugated cardboard visual counting device of the embodiment of the present invention includes an image acquisition module 1, a distance measurement module 2, a power supply module 3, a control module 4, a display module 5 and a lifting module 6; the image acquisition module 1, the distance measurement module 2, the power supply module 3, the control module 4 and the display module 5 are all integrated in a shell 8, and the shell 8 is rotatably installed on the lifting module 6; the image acquisition module 1 is connected to the control module 4, and is used to acquire the stacked image of the corrugated cardboard and send it to the control module 4; the distance measurement module 2 is connected to the control module 4, and is used to detect the shooting distance and send it to the control module 4; the control module 4 is connected to the display module 5, and is used to obtain the stacked number of corrugated cardboard according to the stacked image and the shooting distance and send it to the display module 5 for display.

[0054] Specifically, the overall hardware of the counting device is implemented based on a mobile embedded platform, wherein the control module 4 is a Raspberry Pi microcomputer, which is the core computing module of the device and is a lightweight embedded development board with the characteristics of small size and low price; the distance measuring module 2 is a distance measuring sensor, which is used to detect the shooting distance and transmit the detection data to the Raspberry Pi in real time; the image acquisition module 1 is used for imaging the cardboard stack, including an array camera, a lens and an aperture, and can transmit the image data to the Raspberry Pi in real time; the Raspberry Pi calculates the number of cardboards in the currently collected image in real time; the display module 5 is a touch display screen, which displays a human-computer interaction interface, can be used to control the camera and call the counting function, and display the collected picture, shooting distance and counting results in real time.

[0055] In addition, the lifting module 6 includes a portable lifting bracket, which is small in size, light in material, and easy to carry. The shell 8 of the device is fixed to the top of the portable lifting bracket. The cardboard images of the cardboard stack at different viewing angles can be collected through the lifting counting device, thereby avoiding the problem of individual cardboards being blocked during imaging.

[0056] The overall hardware structure of the present invention is simple, low-cost, and easy to use. Since the embedded device is small in size, low in energy consumption, easy to carry, and high in work efficiency, it is suitable for moving the counting device at any time on the factory production line according to usage needs to detect and calculate the number of cardboards in real time.

[0057] like Fig. 9 As shown, the embodiment of the present invention also provides a visual counting method based on the corrugated cardboard visual counting device as described above, comprising the steps of:

[0058] 1) The image acquisition module 1 acquires images of the cardboard stack at different positions, and splices the images of the cardboard stack at different positions to obtain a complete image of the cardboard stack;

[0059] 2) Cut out the middle image of a number of columns of pixels from the complete cardboard stack image, and then extract the cardboard area image;

[0060] 3) Enhance the cardboard holes in the cardboard area image to obtain a hole enhancement image;

[0061] 4) Obtain the center position of the cardboard hole in the hole enhancement image, obtain the number of local peaks based on the center position of the cardboard hole, and finally obtain the total number of cardboards.

[0062] In a specific embodiment, in step 1), the shooting distance, lens focus and aperture, and camera parameters are adjusted according to the actual scene, and according to different cardboard stack heights, the counting device is moved on the portable lifting bracket to collect images of the cardboard stack at different positions. The camera field of view within the moving range covers the entire cardboard stack, and a complete cardboard stack image is obtained by splicing the image sequence of different positions of the cardboard stack and sent to counting. The specific steps are:

[0063] 1.1) The image acquisition module 1 acquires images of the cardboard stack at different positions to form an image sequence;

[0064] 1.2) Extract and match feature points of image sequences based on SURF algorithm;

[0065] 1.3) Construct the transformation matrix between image sequences by matching point pairs, thus completing the stitching of panoramic images;

[0066] 1.4) According to the inter-image transformation matrix, the corresponding images can be transformed to determine the overlapping area between the images, and the image to be fused can be mapped to a new blank image to form a mosaic image.

[0067] In the image acquisition method proposed by the present invention, a counting device is moved on a portable lifting bracket to collect cardboard images in different directions, and then these images are spliced ​​to solve the problem of individual cardboard occlusion in imaging, and at the same time solve the problem that a single camera cannot capture all higher cardboard stacks in the past. Therefore, it is also well applicable to cardboard stacks of various heights.

[0068] In a specific embodiment, in step 2), image data of a certain number of pixel widths in the middle are intercepted from the complete cardboard stack image, and after preprocessing such as bilateral filtering, the image of the cardboard region is extracted as the input image. This step can reduce a large amount of unnecessary calculations and greatly improve the real-time counting performance.

[0069] In a specific embodiment, in step 2), the process of extracting the cardboard area is as follows:

[0070] 2.1) The cross-sectional structure of the cardboard is as follows Figure 3As shown in the figure, the pre-processed image is first processed by adaptive histogram equalization to enhance the image contrast. Since the cardboard section has sinusoidal corrugation characteristics, the texture characteristics of the cardboard area and the non-cardboard area are quite different. The corrugated texture in the cardboard section area can be approximated by several curved oblique lines with an angle of approximately 45 degrees to the horizontal direction. Therefore, Figure 4 The specific structure operator shown is used to detect the corrugated bevel edge of the cardboard cut surface in the equalized image. Figure 5 As shown in the figure, for poorly imaged images, even if the corrugated features of the cardboard cross-section in the image are not obvious to the human eye, a cardboard bevel detection map is obtained by convolution operation with a specific result operator, and then the cardboard bevel area can be effectively enhanced through denoising, normalization, morphological processing and other operations, and a cardboard bevel enhancement map is obtained;

[0071] 2.2) Because the cardboard stack has a certain height, and the light intensity of the fill light 7 at the upper and lower ends of the cardboard stack has a certain attenuation, resulting in uneven image brightness. In the cardboard bevel enhancement image obtained in the previous step, the sum of the pixel values ​​in each area is calculated to determine the upper and lower boundary ordinates y of the cardboard area. upper and lower : Let the sum of the pixel values ​​in the ith row of the cardboard bevel enhancement image be S i , select threshold ξ1, start traversing in positive order from the first row, when S i >ξ1 and S i+5 >ξ1, if and This line is the upper boundary of the cardboard area. Similar to the method of determining the upper boundary, the lower boundary of the cardboard area is determined by traversing in reverse order.

[0072] 2.3) In order to solve the problem that the gap between cardboards may affect the final detection results, this step generates a corresponding cardboard gap mask map to cover the cardboard gap to eliminate the subsequent impact: intercept the row range from y upper to y lower The cardboard area is traversed, and the sum of the pixel values ​​of each row is traversed. The gap area and the non-gap area are distinguished according to the threshold ξ2. All the pixels belonging to the gap position and the non-gap position are set to 0 and 255 respectively to obtain the cardboard area mask map. The row range from y upper to y lower The image is logically ANDed with the mask image to obtain the final cardboard area image.

[0073] In a specific embodiment, in step 3), the specific process of hole enhancement is:

[0074] For each pixel p in the cardboard area map, the Hessian matrix at multiple different scales is calculated, and three different Gaussian function variances σ are determined as the scale of the Hessian matrix according to the main size of the cardboard hole. Since the cardboard hole area is a spot structure and the foreground is dark, the second-order gradient in this area is positive. Therefore, the eigenvalue of the Hessian matrix H(p, σ) in the cardboard hole area satisfies 0<λ1≤λ2, and the following filter function is used to enhance the hole features:

[0075]

[0076] where λ ρ is the regularized λ2:

[0077]

[0078] Finally, the maximum filter response in scale space is F(p)=max σ V[eigH(p,σ)], then the hole enhancement map can be obtained.

[0079] like Figure 6 As shown in the figure, for images with blurred hole edges and low overall brightness, this method calculates the second-order gradient at each pixel, smoothes and suppresses the image signal in the non-hole area through a filter in the scale space, and sharpens the image in the hole area to enhance holes of different sizes. This method also has a good hole enhancement effect in cardboard hole images with blurred holes and low brightness.

[0080] The counting method proposed in the present invention effectively extracts hole information based on the characteristics of the hole shape in the cardboard stacking image and the difference between the second-order grayscale gradient of the hole area and the edge. It also has good detection effect on images with blurred cardboard edges and images with low brightness, solving the problem that the detection results of the prior art are greatly affected by imaging clarity and lighting.

[0081] In a specific embodiment, in step 4), the center position of each hole is extracted after noise suppression is performed on the hole enhancement image obtained in the previous step using the LoG operator. Figure 7 As shown in the figure, at the center of each hole, a 2×2 structure element is used to replace the original hole, where the value of each point at the structure element position is 255, and the remaining points are 0. At this time, the hole center map is obtained, and then all the pixels in the hole center map are mapped and projected in the Y-axis direction and then mean blurred to obtain the hole center projection map.

[0082] The second-order difference idea is used to count the number of local peaks on the hole center projection map. Two rounds of iterative calculations are used to quickly count the number of local peaks in the projection map. The specific method is: the value of the yth row of the hole center projection map is N y ; From Ny Count all N′ in positive order starting from the first value y :

[0083]

[0084] Get all N′ y Then calculate all N" in reverse order starting from the last value y :

[0085]

[0086] For all N″ y , when there exists y that satisfies N″ y+1 -N″ y =2, y is the local peak ordinate; the number of local peaks is the number of flutes in the paperboard stack. For any prismatic paperboard stack, the total number of paperboards can be obtained based on the total number of local peaks and the prismatic shape of the paperboard.

[0087] Each peak position recorded above can be used for later manual verification, and the processed calculation results are directly displayed on the human-computer interface. The user can interactively choose to continue collecting or shut down the system.

[0088] like Figure 8 As shown, the cardboard detection method of the present invention has high precision and wide adaptability, and has high accuracy in detecting different types of cardboard stacking images; and the counting process has small calculation amount, short time consumption, stable performance, and can meet the high real-time requirements of lightweight embedded systems.

[0089] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A corrugated cardboard visual counting method, characterized in that: Includes steps: 1) Collecting images of cardboard stacking at different positions, and splicing the images of cardboard stacking at different positions to obtain a complete cardboard stacking image; 2) Cut out the middle image of a number of columns of pixels from the complete cardboard stack image, and then extract the cardboard area image; 3) Enhance the cardboard holes in the cardboard area image to obtain a hole enhancement image; 4) Obtain the center position of the cardboard hole in the hole enhancement image, obtain the number of local peaks according to the center position of the cardboard hole, and finally obtain the total number of cardboards; In step 2), the specific process of extracting the cardboard area image is as follows: 2.1) Firstly, the image is processed by adaptive histogram equalization to enhance the image contrast, and then a convolution operation is performed with the preset result operator to obtain the cardboard bevel edge detection image, and finally the cardboard bevel edge enhancement image is obtained; 2.2) Slide up and down in the cardboard bevel enhancement image to calculate the sum of the pixel values ​​in each area to determine the upper and lower boundary y coordinates of the cardboard area upper and lower ; 2.3) The cardboard gap is covered by generating a corresponding cardboard gap mask image to obtain the final cardboard area image.

2. The visual counting method according to claim 1, characterized in that: The specific process of step 1) is: 1.1) Collect images of cardboard stacks at different positions to form an image sequence; 1.2) Extract and match feature points of image sequences; 1.3) Construct the transformation matrix between the image sequences through matching points to complete the stitching of the panoramic image; 1.4) Transform the corresponding images according to the transformation matrix of the image sequence to determine the overlapping area between the images, and map the image to be fused into a new blank image to form a mosaic.

3. The visual counting method according to claim 1, characterized in that: The specific process of step 2.2) is as follows: Suppose the sum of the pixel values ​​in the ith row of the cardboard bevel enhancement image is S i , select the threshold ξ1, start traversing in positive order from the first row, when S i >ξ1 and S i+5 >ξ1, if and This line is the upper boundary of the cardboard area; similarly, the lower boundary of the cardboard area is determined by traversing in reverse order.

4. The visual counting method according to claim 3, characterized in that: The specific process of step 2.3) is as follows: on the cardboard bevel enhancement diagram, the row range is intercepted from y upper to y lower The cardboard area is traversed, and the sum of the pixel values ​​of each row is traversed. The gap area and the non-gap area are distinguished according to the threshold ξ2. All the pixels belonging to the gap position and the pixel points of the non-gap position are set to 0 and 255 respectively, and the cardboard area mask map is obtained. The row range from y upper to y lower The image is logically ANDed with the cardboard area mask image to obtain the final cardboard area image.

5. The visual counting method according to claim 4, characterized in that: The specific process of step 3) is: For each pixel point p in the cardboard area map, the Hessian matrix at multiple different scales is calculated, and three different Gaussian function variances σ are determined as the scale of the Hessian matrix according to the main size of the cardboard hole; Use the following filter function to enhance hole features: where λ ρ is the regularized λ2: Finally, the maximum filter response in scale space is F(p)=max σ V[eigH(p,σ)], at this time, the hole enhancement map is obtained.

6. The visual counting method according to claim 5, characterized in that: The specific process of step 4) is: After using the LoG operator to suppress the noise of the hole enhancement image, the center position of each hole is extracted to obtain the hole center map; After all the pixels in the hole center map are projected toward the Y axis, mean fuzzy processing is performed to obtain the hole center projection map; The hole center projection image is processed by the second-order difference idea to count the number of local peaks, and the number of local peaks in the projection image is quickly counted through multiple rounds of iterative calculations; the number of local peaks is the number of flutes in the cardboard stack; The total number of cardboards is obtained based on the number of local peaks and the cardboard prism shape.

Images