Detection system and detection method for glass cracks on conveyor belt

Abstract

The invention relates to the technical field of image recognition, and discloses a system and method for detecting glass cracks on a conveyor belt, the detection system comprises a closed box used for shielding external light, the conveyor belt for conveying glass penetrates through the closed box, a light source and a camera are arranged in the closed box, a pneumatic blowing nozzle is arranged outside the closed box, and the pneumatic blowing nozzle is arranged in the closed box. The surface of the glass which is about to enter the closed box is purged; the camera is electrically connected with the server, the server establishes an ROI area according to the video data information and monitors pixel statistical characteristics of the ROI area, and when the pixel statistical characteristics of the ROI area meet glass crack conditions, it is judged that cracks are detected. According to the invention, video sampling is carried out on the glass in the closed box, interference of an ambient light source on imaging is eliminated, and the image processing complexity is reduced; whether cracks exist or not is judged by comparing the gray scale standard deviation of a normal glass image, the calculation amount is small, and glass crack judgment is completed under video image high-speed sampling.

Description

Technical Field

[0001] The present invention relates to the field of image recognition, and in particular to a glass crack detection system and method on a conveyor belt. Background Art

[0002] Glass is an important material in daily life, production and the field of science and technology. Due to its light transmittance, it is widely used in construction, furniture manufacturing, automobiles, electronics, instrumentation and other fields.

[0003] The glass produced from the kiln is usually transferred to other areas for storage via a conveyor belt. However, if the glass is cracked or even broken on the conveyor belt, it not only indicates that there is a quality problem with the glass, but also affects the normal transmission of the glass on the conveyor belt, and even causes blockage and subsequent glass breakage. Therefore, it is very necessary to detect glass cracks on the conveyor belt in a timely manner. Summary of the invention

[0004] In order to overcome or alleviate one or more of the above technical problems, the present invention aims to provide a glass crack detection system and method on a conveyor belt.

[0005] On one hand, the present invention provides a glass crack detection system on a conveyor belt, which comprises a closed box for shielding external light, a conveyor belt for transporting glass passes through the closed box, a light source for lighting and a camera for shooting glass video are arranged inside the closed box, and a pneumatic blowing nozzle is arranged outside the closed box to blow the surface of the glass that is about to enter the closed box;

[0006] The camera is electrically connected to the server, and the camera transmits the collected video data information to the server. The server establishes an ROI area based on the video data information and monitors the pixel statistical characteristics of the ROI area. When the pixel statistical characteristics of the ROI area meet the glass crack conditions, it is determined that a crack is detected, and a stop signal is sent to the control device of the conveyor belt.

[0007] According to some embodiments, the pneumatic blowing nozzle forms an angle of 45° with the horizontal plane.

[0008] On the other hand, the present invention also provides a detection method of the above-mentioned glass crack detection system on the conveyor belt, which comprises the following steps:

[0009] S1: Start data collection, place the glass to be tested on the conveyor belt, and after the pneumatic nozzle blows the glass surface, the glass to be tested enters the sealed box. Under the illumination of the light source, the camera arranged on the top of the sealed box collects video data of the glass surface and transmits it to the server;

[0010] S2: The server reads a frame of image data from the acquired video data;

[0011] S3: Detect whether there is a new mouse drawing frame. If so, update the ROI area in the read image data and extract the image data of the new ROI area; otherwise, directly extract the image data of the ROI area from the read image data;

[0012] S4: Perform grayscale processing on the image of the ROI area to obtain a grayscale picture;

[0013] S5: Perform pre-filtering processing on the grayscale picture;

[0014] S6: Calculate the grayscale mean mean and standard deviation std of the filtered image;

[0015] S7: Calculate the difference and product thr of mean and std respectively. The calculation formulas are as follows:

[0016] Δmean = mean - mean0; (1)

[0017] Δstd = std - std0; (2)

[0018] thr = Δmean * Δstd; (3)

[0019] Where mean0 and std0 are the grayscale mean and standard deviation of the previous picture respectively;

[0020] S8: If it is necessary to enter the calibration process, save and record the maximum values dMeanMax and dStdMax of the absolute values of Δmean and Δstd respectively, and the minimum value Thr of thr, and go to step S2; otherwise, go to step S9;

[0021] S9: Calculate A = (the first condition and the second condition and the third condition), where:

[0022] The first condition is: Is the absolute value of Δmean greater than dMeanMax?

[0023] The second condition is: Is the absolute value of Δstd greater than dStdMax?

[0024] The third condition is: Is thr < K * Thr? Where K is a positive number greater than 1, but too large a K value will cause crack undetected;

[0025] If the value of A is false, it is judged that there is no crack, and go to step S2; otherwise, it is judged that there is a crack and enter step S10;

[0026] S10: The server processes the alarm, sends an alarm message to the monitoring personnel, and sends a conveyor belt stop control signal, and then goes to step S2 to continue the cycle of detecting glass cracks.

[0027] According to some implementations, in step S9, the K value ranges from 2 to 5.

[0028] According to some embodiments, in step S3, the width of the ROI area is the maximum width of the glass on the conveyor belt; and the length of the ROI area is one quarter of the length of the video area of ​​the camera.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] (1) The present invention provides a glass crack detection system on a conveyor belt, which performs video sampling of glass in a closed box, eliminates the interference of ambient light sources on imaging, and reduces the complexity of image processing;

[0031] (2) The dust and debris on the glass surface are blown off by a nozzle in front of the sealed box of the present invention, thereby reducing the false detection rate;

[0032] (3) The detection method provided by the present invention determines whether there is a crack by comparing the grayscale standard deviation of a normal glass image. There is no need to perform processes such as edge feature extraction of the image. The calculation amount is small and the calculation speed is fast. It can complete the glass crack judgment under high-speed sampling of video images without affecting the normal transmission of glass on the production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of the structure of a glass crack detection system on a conveyor belt provided in an embodiment of the present invention.

[0034] Figure 2 A flowchart of a detection method of a glass crack detection system on a conveyor belt provided in an embodiment of the present invention.

[0035] Figure 3 This is a screenshot of the glass crack detection video provided in an embodiment of the present invention.

[0036] Figure 4 This is another screenshot of the glass crack detection video provided in an embodiment of the present invention.

[0037] Figure 5 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0038] Figure 6 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0039] Figure 7This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0040] Figure 8 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0041] Fig. 9 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0042] Fig.10 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0043] Fig.11 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0044] Fig.12 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0045] Fig.13 This is a screenshot of the ROI area at a certain time in the glass crack detection provided by an embodiment of the present invention.

[0046] In the figure:

[0047] Conveyor belt 1; sealed box 2; camera 3; light source 4; pneumatic blowing nozzle 5. DETAILED DESCRIPTION

[0048] The present invention is described in detail below in conjunction with the embodiments and drawings, but it should be understood that the embodiments and drawings are only used to exemplify the present invention and do not constitute any limitation on the protection scope of the present invention. All reasonable changes and combinations within the scope of the inventive concept of the present invention fall within the protection scope of the present invention.

[0049] The present invention will be further described below in conjunction with the accompanying drawings.

[0050] Example 1

[0051] like Figure 1 The present embodiment provides a glass crack detection system on a conveyor belt, the detection system comprises a sealed box 2 for shielding light from an external light source, a conveyor belt 1 for transporting glass passes horizontally through the middle of the sealed box 2, a light source 4 for illuminating the glass and a camera 3 for photographing the glass are provided on the top surface of the sealed box 2, a pneumatic blowing nozzle 5 is arranged at a certain angle outside the sealed box 2, and is used to blow the glass on the glass surface that is about to enter the sealed box 2, and is preferably arranged at an angle of 45° on the horizontal plane.

[0052] The camera 3 is electrically connected to the server and transmits the collected video data information to the server; the server monitors the pixel statistical characteristics of the ROI area (such as the mean and mean square error of the grayscale image) according to the video data information. When the pixel statistical characteristics of the ROI area exceed the set threshold, it is judged that a crack is detected, and a stop signal is sent to the control device of the conveyor belt 1.

[0053] Example 2

[0054] like Figure 2 Based on Example 1, this embodiment provides a method for detecting glass cracks on a conveyor belt, comprising the following steps:

[0055] S1: The glass to be tested is placed on the conveyor belt 1, and the pneumatic blowing nozzle 5 blows the glass surface and then enters the sealing box 2. Under the illumination of the light source 4, the camera 3 collects video data of the glass surface and transmits it to the server;

[0056] S2: Read a frame of image data from the acquired video data;

[0057] S3: further extracting image data of the ROI area (ROI stands for Region of Interest) from the read image data.

[0058] In step S3, the size of the ROI region is determined by the operator operating the mouse frame. Since the operator may update the ROI region at any time, it is necessary to detect whether a new mouse frame operation is generated in a timely manner in the flow chart. If the mouse frame is detected, the ROI region needs to be updated, otherwise the subsequent process is executed.

[0059] S4: grayscale the image in the ROI area to obtain a grayscale image;

[0060] S5: Perform filtering preprocessing on the grayscale image;

[0061] S6: Calculate the grayscale mean and mean square error std of the filtered image;

[0062] S7: Calculate the difference between mean and std and their product thr respectively. The calculation formulas are as follows:

[0063] Δmean=mean-mean0; (1)

[0064] Δstd= std - std 0; (2)

[0065] thr=Δmean*Δstd; (3)

[0066] Among them, mean0 and std 0 are the grayscale mean and mean square error of the previous image respectively.

[0067] S8: If it is necessary to enter the calibration process, save the maximum values dMeanMax and dStdMax of the absolute values of Δmean and Δstd, and the minimum value Thr of thr respectively, and go to execute step S2; otherwise, go to execute step S9;

[0068] S9: Calculate A = the first condition and the second condition and the third condition, where:

[0069] The first condition is: Is the absolute value of Δmean greater than dMeanMax?

[0070] The second condition is: Is the absolute value of Δstd greater than dStdMax?

[0071] The third condition is: Is thr < K * Thr? where K is a positive number greater than 1, but too large a K value will cause cracks to be missed; the value of K is taken as 2 - 5. If there are no cracks in the glass, but the system alarms for cracks, the value of K needs to be increased; if there are cracks in the glass, but the system fails to detect them, the value of K needs to be decreased.

[0072] If the value of A is false, it is judged that there are no cracks, no treatment is done, and go to execute step S2; otherwise, it is judged that there are cracks, and enter step S10;

[0073] S10: Alarm handling, send an alarm message to the monitoring personnel, and send a conveyor belt stop control signal, and then go to execute step S2 to continue to circularly detect the cracks in the glass.

[0074] In step S1, the conveyor belt 1 passes horizontally through the middle of the sealed box 2, and a pneumatic nozzle is configured at the entrance end of the sealed box to spray compressed filtered air flow at a 45° angle forward on the glass surface, reducing the dust entering the sealed box and blowing away the foreign objects on the glass at the same time.

[0075] In step S3, the width of the ROI area is the maximum width of the glass on the conveyor belt to ensure covering the width of the glass; in order to improve the detection speed and the area coverage of continuous sampling, the length of the ROI area is one - quarter of the length of the video area of the camera 3.

[0076] In step S1, the camera 3 adopts the video acquisition method, which does not affect the continuous transportation of the glass by the conveyor belt 1.

[0077] Embodiment 3

[0078] Based on the conveyor - belt glass crack detection system provided in Embodiment 1 and the conveyor - belt glass crack detection method provided in Embodiment 2, verification is carried out in this embodiment.

[0079] Figure 3This is a picture of a glass crack detection video. The white cylinder is the roller on the conveyor belt 1. The glass is on the roller (left side of the video screenshot). The glass is transparent, and only the white edge line on the right side can be seen in the picture. In the upper half of the picture, the front side of the white roller is a suspended part with a light white horizontal bar, which does not affect the system's crack detection, and the bottom is dark.

[0080] Figure 4 In the front area of ​​the white roller, a monitoring area ROI (yellow box) is drawn to cover the dark area in front of the roller.

[0081] The test results are explained as follows:

[0082] Attached Figures 3-4 The video screenshot shows the glass moving from top to bottom on the conveyor belt.

[0083] When the characteristic parameters of the video data meet the crack conditions, the system automatically saves the image in the ROI.

[0084] In a closed environment, when cracks appear on the glass, the refractive index at the crack changes, and irregular black textures (the dark background under the conveyor belt) appear after imaging.

[0085] When cracks appear, the system takes screenshots and saves images of the ROI area in chronological order.

[0086] The following are the results of an experiment:

[0087] During the calibration process, the results of dMeanMax, dStdMax, and Thr are:

[0088] dMeanMax=0.413527; dStdMax=0.23843; Thr=-0.435618.

[0089] The following table is attached Figures 5 to 13 The corresponding calibration result list:

[0090]

[0091] Figures 5 to 13 In the figure, the black lines are cracks.

[0092] The above embodiments are only preferred implementations 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, improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A glass crack detection system on a conveyor belt, characterized in that: It includes an airtight box for blocking external light. A conveyor belt for transporting glass passes through the airtight box. Inside the airtight box, there is a lighting light source and a camera for shooting videos of the glass. A pneumatic nozzle is arranged outside the airtight box to purge the surface of the glass about to enter the airtight box. The camera is electrically connected to a server. The camera transmits the collected video data information to the server. The server establishes a ROI region based on the video data information and monitors the pixel statistical features of the ROI region. When the pixel statistical features of the ROI region meet the glass crack condition, it is determined that a crack is detected, and then a stop signal is sent to the control device of the conveyor belt.

2. The glass crack detection system on a conveyor belt according to claim 1, characterized in that: The pneumatic nozzle forms an angle of 45° with the horizontal plane.

3. A method for detecting cracks in a glass conveyor belt according to any one of claims 1 to 2, characterized in that: It includes the following steps: S1: Start data collection. Place the glass to be tested on the conveyor belt. After the pneumatic nozzle purges the surface of the glass, the glass to be tested enters the sealed box. Under the illumination of the light source, the camera arranged at the top inside the sealed box collects video data of the glass surface and transmits it to the server. S2: The server reads a frame of image data from the acquired video data. S3: Detect whether there is a new mouse drawing frame. If so, update the ROI region in the read image data and extract the image data of the new ROI region; if not, directly extract the image data of the ROI region from the read image data. S4: Perform grayscale processing on the image of this ROI region to obtain a grayscale picture. S5: Perform filtering preprocessing on the grayscale picture. S6: Calculate the gray mean value mean and the mean square deviation std of the filtered image. S7: Calculate the difference and product thr of mean and std respectively. The calculation formulas are as follows: Δmean = mean - mean0; (1) Δstd = std - std0; (2) thr = Δmean * Δstd; (3) Where, mean0 and std0 are the gray mean value and the mean square deviation of the previous picture respectively. S8: If it is necessary to enter the calibration process, record the maximum values dMeanMax and dStdMax of the absolute values of Δmean and Δstd respectively, and the minimum value Thr of thr, and then go to execute step S2; otherwise, go to execute step S9. S9: Calculate A = (the first condition and the second condition and the third condition), where: The first condition is: Is the absolute value of Δmean greater than dMeanMax? The second condition is: Is the absolute value of Δstd greater than dStdMax? The third condition is: Is thr < K * Thr? Where K is a positive number greater than 1, but too large a K value will cause cracks to be missed. If the value of A is no, it is judged that there is no crack, and then go to execute step S2; otherwise, it is judged that there is a crack, and enter step S10. S10: The server performs alarm processing, sends an alarm message to the monitoring personnel, and sends a conveyor belt stop control signal, and then goes to execute step S2 to continue to cycle and detect glass cracks.

4. The detection method according to claim 3, characterized in that: In step S9, the value range of K is 2 to 5.

5. The detection method according to claim 3, characterized in that: In step S3, the width of the ROI area is the maximum width of the glass on the conveyor belt; the length of the ROI area is one quarter of the length of the video area of ​​the camera.

Images