Method for detecting abnormal display of screens of multiple devices

Abstract

The invention discloses a method for detecting screen display abnormity of multiple devices. The method comprises the following steps: S1, preparing multiple devices to be detected, a high-definition camera and a data processing device; by combining the edge detection algorithm, the image gray processing and the mean processing, whether the screen of the equipment to be detected is lightened or not can be automatically identified, multiple pieces of equipment can be detected at the same time, the manual operation time is shortened, the labor intensity is reduced, the detection efficiency is improved, whether the display of the equipment is abnormal or not can be accurately judged, and the detection efficiency is improved. The device screen display abnormity detection method has the advantages of simple detection environment construction, simple and clear operation steps, easy implementation and popularization, suitableness for detection of device screen display abnormity of enterprises and institutions of various scales, and reduction of the detection cost.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormality detection of screen displays of multiple devices, and in particular to a method for detecting abnormality of screen displays of multiple devices. Background Art

[0002] As an important component of electronic devices, the display quality directly affects the user experience. However, the display may have abnormal display phenomena such as distorted screen, blank screen, striped screen, etc. when the device is turned on or when the screen is awakened. These abnormalities may disappear after the screen is restarted or re-displayed. Manual inspection is time-consuming and the test data may be inaccurate.

[0003] Although there are some display screen detection methods in the prior art, these methods often require complex environment construction and their accuracy needs to be improved. For example, some methods can only test one device at a time, and the testing efficiency is low. In addition, some methods may require physical contact with the display screen, which may cause damage to the display screen or leave marks.

[0004] Therefore, a method is needed that can quickly build an environment, has high detection efficiency and accurately detect display screen abnormalities. In view of this, the present application proposes a method for detecting screen display abnormalities of multiple devices. Summary of the invention

[0005] Based on the technical problems existing in the background technology, the present invention proposes a method for detecting display anomalies of screens of multiple devices.

[0006] The present invention provides a method for detecting display anomalies on screens of multiple devices, comprising the following steps:

[0007] S1: Prepare multiple devices to be tested, high-definition cameras and data processing equipment;

[0008] S2: The data processing device uses an image edge detection algorithm to accurately obtain screen images of multiple devices to be detected, providing basic data for subsequent anomaly detection;

[0009] S3: The device enters the test mode and sets a certain time interval to rotate the screen on and off. The rotation display simulates the actual usage scenario and does not require manual clicking, which improves the detection efficiency.

[0010] S4: The data processing equipment continuously obtains the image data of each device screen through a high-definition camera, uses digital image processing technology to determine whether the screen is lit, and takes a picture of the screen after a delay. It monitors the screen status of each device in real time and takes pictures after the screen is lit to ensure that a clear screen image is obtained, thereby improving the accuracy of detection;

[0011] S5: performing digital image processing on the captured image, and comparing the processed image with the normal image data to obtain a difference value;

[0012] S6: Set a certain difference threshold, and judge whether the display of the device is abnormal according to the comparison result. By setting the threshold, quantify the degree of difference and accurately judge the display abnormality;

[0013] S7: Output a detection report with abnormal pictures, timestamps, and abnormal number statistics based on the detection results to facilitate subsequent analysis and processing.

[0014] Preferably, in S2, the data processing device acquires image information within a certain range through a high-definition camera. Since the screen of the device to be detected is usually rectangular, an edge detection algorithm can be used to obtain the area to be displayed on the device, filter out irregular rectangles or rectangles whose size does not conform to the device screen, and screen out the device to be detected;

[0015] The edge detection algorithm is the Canny edge detection algorithm, and the implementation of the Canny edge detection algorithm includes the following five steps:

[0016] A: Gaussian filtering;

[0017] B: Gradient calculation;

[0018] C: non-maximum suppression;

[0019] D: dual threshold boundary tracking;

[0020] E: Edge connection.

[0021] Preferably, the Gaussian filter is an image denoising technology based on weighted averaging, the core of which is a Gaussian filter, which is a matrix with different weight values, wherein each pixel corresponds to a weight value, and these weight values ​​are calculated according to a Gaussian function, wherein the weight of the central pixel is the largest, and the weights of the surrounding pixels gradually decrease. During the filtering process, the value of each pixel is updated by weighted averaging the values ​​of the surrounding pixels to achieve image blurring and denoising. At the same time, since the pixel weights in the edge area are larger, the Gaussian filter can calculate the edge more accurately, thereby improving the effect of edge detection.

[0022] Preferably, the gradient calculation calculates the gradient of the filtered image, including the magnitude and direction of the gradient, the magnitude of the gradient represents the intensity of the grayscale change, and the direction of the gradient represents the direction of the grayscale change;

[0023] For a grayscale image f(x,y), its partial derivatives in the x and y directions can be expressed as:

[0024]

[0025] These partial derivatives describe the rate of change of the grayscale of the image in the x-direction and the y-direction;

[0026] According to the partial derivative, the magnitude and direction of the gradient can be calculated. The magnitude of the gradient represents the intensity of the grayscale change, while the direction of the gradient represents the direction of the grayscale change.

[0027] The magnitude of the gradient can be calculated using the following formula:

[0028]

[0029] Among them, G(x,y) represents the magnitude of the gradient;

[0030] The direction of the gradient can be calculated using the following formula:

[0031]

[0032] Among them, θ(x,y) represents the direction of the gradient, and its value range is [0,π);

[0033] Use the Sobel operator to approximate the partial derivatives of the image. Use the following template for the partial derivatives in the x and y directions:

[0034]

[0035] Preferably, the non-maximum suppression is used to perform a non-maximum suppression operation after calculating the amplitude and direction of the gradient. The purpose of the non-maximum suppression is to retain the local maximum and suppress the non-local maximum. For each pixel point, according to the direction of its gradient, the point with the largest gradient amplitude is found in the neighborhood. If the point is not a local maximum, its gradient amplitude is set to 0.

[0036] After non-maximum suppression, the edge pixels in the image are initially enhanced and refined, and can more accurately represent the actual edges. Due to the influence of factors such as noise and color changes, some non-edge pixels are still mistakenly retained. In order to remove these stray responses, the double threshold method is used to further filter the edge pixels, and only edge pixels with high gradient values ​​are retained.

[0037] Preferably, the dual-threshold boundary tracking is used to distinguish strong edges from weak edges, and the distinction is made by setting two thresholds: a higher threshold (called the high threshold) and a lower threshold (called the low threshold), and dividing edge pixels into three categories according to the relationship between their gradient values ​​and the two thresholds: strong edge pixels: if the gradient value of an edge pixel is higher than the high threshold, it is marked as a strong edge pixel, and these pixels represent obvious and reliable edges in the image; weak edge pixels: if the gradient value of an edge pixel is less than the high threshold but greater than the low threshold, it is marked as a weak edge pixel, and these pixels represent less obvious and possibly less reliable edges in the image; suppressed pixels: if the gradient value of an edge pixel is less than the low threshold, it is suppressed, that is, it is not regarded as an edge pixel.

[0038] Preferably, the edge connection is used to perform edge connection on the dual threshold image, connecting isolated edge points into continuous edge lines. After obtaining the edge image, each connected area is traversed to find the leftmost, rightmost, topmost and bottommost points on each connected area. These points will constitute the vertices of the minimum circumscribed rectangle. According to the found vertices, the width and height of the minimum circumscribed rectangle are calculated, and the minimum circumscribed rectangle of each connected area is judged. The connected areas that meet the width and height ratio and have an area greater than a certain value are saved to screen out the devices to be detected.

[0039] Preferably, in S4, the digital image processing determines whether the screen is lit by converting the acquired image data into a grayscale image, where each pixel has only one brightness value, so as to compare it with the grayscale image when the screen is off to determine whether the screen is lit;

[0040] The grayscale image conversion formula used to convert a color image to a grayscale image is:

[0041] Gray=0.299×R+0.587×G+0.114×B

[0042] Among them, R, G, B represent the values ​​of the red, green, and blue channels of the pixel respectively, and Gray represents the grayscale value after conversion;

[0043] When calculating the absolute difference between two grayscale images, the following formula is used:

[0044] Gray=|Gray 1 -Gray 2 |

[0045] Among them, Gray 1 and Gray 2 They represent the grayscale values ​​of the corresponding pixels in the two grayscale images, and Gray represents the calculated absolute difference;

[0046] When calculating the average of all pixels in the absolute difference image, the following formula is used:

[0047]

[0048] Where N represents the total number of pixels in the absolute difference image, Diff i represents the absolute difference of the i-th pixel, and AvgDiff represents the calculated average value;

[0049] Finally, the comparison threshold is set. The threshold is set according to the data of AvgDiff in different environments. If the average value is greater than a certain threshold, it means that the screen is on. If the average value is less than the threshold, it means that the screen is off.

[0050] Preferably, in S5, the normal display image data is obtained by shooting the first few groups of image data of the device to be detected with the screen on, obtaining the average of the grayscale image of the screen on each device to be detected, and then comparing the average of each device to be detected to exclude the device that displays abnormality at the beginning;

[0051] The mean of the bright screen grayscale image is achieved by displaying normal grayscale image data and the following formula:

[0052]

[0053] Among them, N represents the total number of pixels in the grayscale image. i represents the gray value of the i-th pixel, and Mean represents the calculated mean;

[0054] When comparing the means for each device under test, the following formula is used:

[0055] Difference=|Mean 1 -Mean 2 |

[0056] Among them, Mean 1 and Mean 2 They represent the mean values ​​of two devices respectively, and Difference represents the calculated difference value;

[0057] By comparing the difference values, it can be determined whether the mean difference between the two devices is obvious, thereby obtaining the screen lighting picture data that is confirmed to be displayed normally;

[0058] The captured image is compared with the obtained image data showing the normal screen lighting, and the average value of all pixels in the absolute difference image is obtained using the S4 comparison method.

[0059] Preferably, in S6, the difference threshold means that the actual test environment is affected by light and the angle of the high-definition camera, resulting in a certain difference between the obtained data and the truly displayed image, and a certain difference threshold needs to be set to dynamically determine whether the screen display of the device under test is abnormal.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] The present invention can automatically identify whether the screen of the device to be detected is lit by combining the edge detection algorithm, image grayscale processing and mean processing, and can detect multiple devices at the same time, reducing the time and labor intensity of manual operation, improving the detection efficiency, and can accurately determine whether the display of the device is abnormal, reducing misjudgment and missed judgment, and improving the detection accuracy. The detection environment is simple to set up, the operation steps are simple and clear, and it is easy to implement and promote. It is suitable for enterprises and institutions of various sizes to detect abnormal device screen displays, reducing the detection cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 The present invention provides a flow chart of a method for detecting display anomalies on screens of multiple devices. DETAILED DESCRIPTION

[0063] The present invention will be further explained below in conjunction with specific embodiments.

[0064] Example

[0065] Reference Figure 1 This embodiment proposes a method for detecting abnormal display of screens of multiple devices, comprising the following steps:

[0066] S1: Prepare multiple devices to be tested, high-definition cameras and data processing equipment (such as computers, etc.);

[0067] The data processing device obtains the image data of each frame through a high-definition camera, and converts the obtained image data from the BGR color space into a grayscale image, so as to facilitate the subsequent data processing by the computer vision algorithm;

[0068] S2: The data processing device uses an image edge detection algorithm to accurately obtain screen images of multiple devices to be detected, providing basic data for subsequent anomaly detection;

[0069] The edge detection algorithm is the Canny edge detection algorithm. The implementation of the Canny edge detection algorithm includes the following five steps:

[0070] A: Gaussian filtering; Gaussian filtering is an image denoising technology based on weighted averaging. Its core is the Gaussian filter, which is a matrix with different weight values. Each pixel corresponds to a weight value, which is calculated according to the Gaussian function. The weight of the central pixel is the largest, and the weight of the surrounding pixels gradually decreases. During the filtering process, the value of each pixel is updated by weighted averaging the values ​​of the surrounding pixels to achieve image blurring and denoising. At the same time, since the weight of the pixels in the edge area is large, Gaussian filtering can calculate the edge more accurately, thereby improving the effect of edge detection;

[0071] B: Gradient calculation: Gradient calculation calculates the gradient of the filtered image, including the magnitude and direction of the gradient. The magnitude of the gradient indicates the intensity of the grayscale change, and the direction of the gradient indicates the direction of the grayscale change.

[0072] For a grayscale image f(x,y), its partial derivatives in the x and y directions can be expressed as:

[0073]

[0074] These partial derivatives describe the rate of change of the grayscale of the image in the x-direction and the y-direction;

[0075] According to the partial derivative, the magnitude and direction of the gradient can be calculated. The magnitude of the gradient represents the intensity of the grayscale change, while the direction of the gradient represents the direction of the grayscale change.

[0076] The magnitude of the gradient can be calculated using the following formula:

[0077]

[0078] Among them, G(x,y) represents the magnitude of the gradient;

[0079] The direction of the gradient can be calculated using the following formula:

[0080]

[0081] Among them, θ(x,y) represents the direction of the gradient, and its value range is [0,π);

[0082] Use the Sobel operator to approximate the partial derivatives of the image. Use the following template for the partial derivatives in the x and y directions:

[0083]

[0084] C: Non-maximum suppression; Non-maximum suppression is used to perform non-maximum suppression after calculating the magnitude and direction of the gradient. The purpose of non-maximum suppression is to retain the local maximum and suppress the non-local maximum. For each pixel point, according to the direction of its gradient, find the point with the largest gradient magnitude in the neighborhood. If the point is not a local maximum, its gradient magnitude is set to 0.

[0085] After non-maximum suppression, the edge pixels in the image are initially enhanced and refined, which can more accurately represent the actual edge. Due to the influence of factors such as noise and color changes, some non-edge pixels are still mistakenly retained. In order to remove these stray responses, the double threshold method is used to further filter the edge pixels, and only the edge pixels with high gradient values ​​are retained;

[0086] D: Dual-threshold boundary tracking; Dual-threshold boundary tracking is used to distinguish strong edges from weak edges. The distinction is made by setting two thresholds: a higher threshold (called the high threshold) and a lower threshold (called the low threshold). According to the relationship between the gradient value of edge pixels and the two thresholds, they are divided into three categories: Strong edge pixels: If the gradient value of an edge pixel is higher than the high threshold, it is marked as a strong edge pixel. These pixels represent obvious and reliable edges in the image; Weak edge pixels: If the gradient value of an edge pixel is less than the high threshold but greater than the low threshold, it is marked as a weak edge pixel. These pixels represent less obvious and possibly less reliable edges in the image; Suppressed pixels: If the gradient value of an edge pixel is less than the low threshold, it is suppressed, that is, it is not considered an edge pixel;

[0087] E: Edge connection; Edge connection is used to connect the edges of the dual threshold image, connect the isolated edge points into continuous edge lines, and after obtaining the edge image, traverse each connected area to find the leftmost, rightmost, topmost and bottommost points on each connected area. These points will constitute the vertices of the minimum bounding rectangle. According to the found vertices, the width and height of the minimum bounding rectangle are calculated, and the minimum bounding rectangle of each connected area is judged. The connected areas that meet the width and height ratio and have an area greater than a certain value are saved to screen out the devices to be detected;

[0088] The acquired grayscale image data is used with an edge detection algorithm to enhance the edge information in the image and find the rectangular target in the image; first, all external contours are found in the edge image, and then the coordinates and size of the upper left corner of its minimum circumscribed rectangle are calculated; if the aspect ratio of the rectangle is between 0.9 and 1.1 (i.e., it is judged to be close to a square) and the area is greater than 150,000 pixels (the specific parameters are configured according to the size of the screen), it is considered to be a valid rectangular target, thereby obtaining the screen image data of multiple devices to be detected;

[0089] S3: The device enters the test mode and sets a certain time interval to rotate the screen on and off. The rotation display simulates the actual usage scenario and does not require manual clicking, which improves the detection efficiency.

[0090] The device enters the test mode with a certain time of 5 seconds. The device enters the test mode with a time interval of 5 seconds to switch between the screen off and on. The device's test mode needs to support the function of switching between the screen off and on at a certain interval. This function needs to be supported by the device. The purpose of this function is to avoid the need for the data processing device to control the screen on or off operation of the device to be tested through a wired or wireless connection, thereby reducing the difficulty of setting up the test environment.

[0091] S4: The data processing equipment continuously obtains the image data of each device screen through a high-definition camera, uses digital image processing technology to determine whether the screen is lit, and takes a picture of the screen after a delay. It monitors the screen status of each device in real time and takes pictures after the screen is lit to ensure that a clear screen image is obtained, thereby improving the accuracy of detection;

[0092] Digital image processing determines whether the screen is lit by converting the acquired image data into a grayscale image. A grayscale image is an image in which each pixel has only one brightness value. This is then compared with the grayscale image when the screen is off to determine whether the screen is lit.

[0093] The grayscale image conversion formula used to convert a color image to a grayscale image is:

[0094] Gray=0.299×R+0.587×G+0.114×B

[0095] Among them, R, G, B represent the values ​​of the red, green, and blue channels of the pixel respectively, and Gray represents the grayscale value after conversion;

[0096] When calculating the absolute difference between two grayscale images, the following formula is used:

[0097] Gray=|Gray 1 -Gray 2 |

[0098] Among them, Gray 1 and Gray 2 They represent the grayscale values ​​of the corresponding pixels in the two grayscale images, and Gray represents the calculated absolute difference;

[0099] When calculating the average of all pixels in the absolute difference image, the following formula is used:

[0100]

[0101] Where N represents the total number of pixels in the absolute difference image, Diff i represents the absolute difference of the i-th pixel, and AvgDiff represents the calculated average value;

[0102] Finally, set the comparison threshold. The threshold is set according to the data of AvgDiff in different environments. If the average value is greater than a certain threshold, it means that the screen is on. If the average value is less than the threshold, it means that the screen is off.

[0103] Compare the acquired screen grayscale image data with the image data when the display is off to determine whether the screen to be detected is in a lit state; if the device to be detected is in a lit state, take an image of the lit screen after a delay of 1 second to ensure that a clear image of the lit screen is obtained;

[0104] S5: performing digital image processing on the captured image, and comparing the processed image with the normal image data to obtain a difference value;

[0105] The image data showing normal display is obtained by taking the first few groups of image data of the device to be tested with the screen on, and obtaining the average value of the grayscale image of the screen on each device to be tested. The average values ​​of each device to be tested are then compared to exclude the device that shows abnormal display from the beginning;

[0106] The mean of the bright screen grayscale image is achieved by displaying normal grayscale image data and the following formula:

[0107]

[0108] Among them, N represents the total number of pixels in the grayscale image. i represents the gray value of the i-th pixel, and Mean represents the calculated mean;

[0109] When comparing the means for each device under test, the following formula is used:

[0110] Difference=|Mean 1 -Mean 2 |

[0111] Among them, Mean 1 and Mean 2 They represent the mean values ​​of two devices respectively, and Difference represents the calculated difference value;

[0112] By comparing the difference values, it can be determined whether the mean difference between the two devices is obvious, thereby obtaining the screen lighting picture data that is confirmed to be displayed normally;

[0113] The captured image is compared with the obtained image data showing the normal screen lighting, and the average value of all pixels in the absolute difference image is obtained by using the comparison method of S4;

[0114] The captured bright screen images are gray-processed, and the gray-scale data of the first 10 groups of bright screen images of each device to be tested are counted to obtain the average of the bright screen gray-scale images of each device to be tested; the average of each device to be tested is then compared to exclude devices that display abnormalities at the beginning; after obtaining the image data of the normal display of each device to be tested, it is continuously compared with the bright screen images captured later to obtain the difference value;

[0115] S6: Set a certain difference threshold, and judge whether the display of the device is abnormal according to the comparison result. By setting the threshold, quantify the degree of difference and accurately judge the display abnormality;

[0116] The difference threshold means that the actual test environment is affected by light and the angle of the high-definition camera, which causes a certain difference between the obtained data and the true displayed image. It is necessary to set a certain difference threshold to dynamically determine whether the screen display of the device under test is abnormal;

[0117] The difference threshold is set to 100, and whether the device to be tested shows abnormality is determined based on the obtained difference value;

[0118] S7: Output a detection report with abnormal pictures, timestamps, and abnormal number statistics based on the detection results to facilitate subsequent analysis and processing.

[0119] This embodiment only requires a high-definition camera and a computer to build a test environment, which greatly reduces the investment in manpower and material resources and reduces the detection cost. It uses edge detection algorithms, image grayscale processing, and mean processing to automatically identify whether the screen of the device to be tested is lit, and accurately detect multiple devices at the same time, thereby improving the accuracy and efficiency of detection.

[0120] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for detecting abnormal display of screens of multiple devices, characterized in that: The following steps are involved: S1: Prepare multiple devices to be tested, high-definition cameras and data processing equipment; S2: The data processing device accurately obtains screen images of multiple devices to be detected through an image edge detection algorithm; S3: The device enters the test mode and sets a certain time interval to switch between off and on screen display; S4: The data processing equipment continuously obtains the image data of each device screen through a high-definition camera, uses digital image processing technology to determine whether the screen is lit, and takes a picture of the screen after a delay. It monitors the screen status of each device in real time and takes pictures after the screen is lit to ensure that a clear screen image is obtained, thereby improving the accuracy of detection; S5: performing digital image processing on the captured image, and comparing the processed image with the normal image data to obtain a difference value; S6: Set a certain difference threshold, and judge whether the display of the device is abnormal according to the comparison result. By setting the threshold, quantify the degree of difference and accurately judge the display abnormality; S7: Output a detection report with abnormal pictures, timestamps, and abnormal number statistics based on the detection results to facilitate subsequent analysis and processing.

2. A method for detecting display anomalies on screens of multiple devices according to claim 1, characterized in that: In S2, the data processing device obtains image information within a certain range through a high-definition camera. Since the screen of the device to be detected is usually rectangular, an edge detection algorithm can be used to obtain the area to be displayed on the device, filter out irregular rectangles or rectangles whose size does not conform to the device screen, and screen out the device to be detected; The edge detection algorithm is the Canny edge detection algorithm, and the implementation of the Canny edge detection algorithm includes the following five steps: A: Gaussian filtering; B: Gradient calculation; C: non-maximum suppression; D: dual threshold boundary tracking; E: Edge connection.

3. A method for detecting abnormal display of screens of multiple devices according to claim 2, characterized in that: The Gaussian filter is an image denoising technology based on weighted averaging, the core of which is the Gaussian filter, which is a matrix with different weight values, in which each pixel corresponds to a weight value, and these weight values ​​are calculated according to the Gaussian function. The weight of the central pixel is the largest, and the weights of the surrounding pixels gradually decrease. During the filtering process, the value of each pixel is updated by weighted averaging the values ​​of the surrounding pixels to achieve image blurring and denoising. At the same time, since the weights of the pixels in the edge area are larger, the Gaussian filter can calculate the edge more accurately, thereby improving the effect of edge detection.

4. The method for detecting display anomalies of multiple devices according to claim 2, characterized in that: The gradient calculation calculates the gradient of the filtered image, including the magnitude and direction of the gradient, the magnitude of the gradient represents the intensity of the grayscale change, and the direction of the gradient represents the direction of the grayscale change; For a grayscale image f(x,y), its partial derivatives in the x and y directions can be expressed as: These partial derivatives describe the rate of change of the grayscale of the image in the x-direction and the y-direction; According to the partial derivative, the magnitude and direction of the gradient can be calculated. The magnitude of the gradient represents the intensity of the grayscale change, while the direction of the gradient represents the direction of the grayscale change. The magnitude of the gradient can be calculated using the following formula: Among them, G(x,y) represents the magnitude of the gradient; The direction of the gradient can be calculated using the following formula: Among them, θ(x,y) represents the direction of the gradient, and its value range is [0,π); Use the Sobel operator to approximate the partial derivatives of the image. Use the following template for the partial derivatives in the x and y directions:

5. The method for detecting abnormal display of screens of multiple devices according to claim 2, characterized in that: The non-maximum suppression is used to perform a non-maximum suppression operation after calculating the amplitude and direction of the gradient. The purpose of non-maximum suppression is to retain the local maximum and suppress the non-local maximum. For each pixel point, according to the direction of its gradient, the point with the largest gradient amplitude is found in the neighborhood. When the point is not a local maximum, its gradient amplitude is set to 0.

6. The method for detecting abnormal display of multiple device screens according to claim 2, characterized in that: The dual-threshold boundary tracking is used to distinguish strong edges from weak edges. The distinction is made by setting two thresholds: a higher threshold (called the high threshold) and a lower threshold (called the low threshold). According to the relationship between the gradient value of the edge pixels and the two thresholds, the edge pixels are divided into three categories: strong edge pixels: if the gradient value of the edge pixels is higher than the high threshold, they are marked as strong edge pixels, and these pixels represent obvious and reliable edges in the image; weak edge pixels: if the gradient value of the edge pixels is less than the high threshold but greater than the low threshold, they are marked as weak edge pixels, and these pixels represent less obvious and possibly less reliable edges in the image; suppressed pixels: if the gradient value of the edge pixels is less than the low threshold, they are suppressed, that is, they are not regarded as edge pixels.

7. The method for detecting display anomalies of multiple devices according to claim 2, characterized in that: The edge connection is used to perform edge connection on the dual threshold image, connect the isolated edge points into continuous edge lines, and after obtaining the edge image, traverse each connected area to find the leftmost, rightmost, topmost and bottommost points on each connected area. These points will constitute the vertices of the minimum circumscribed rectangle. According to the found vertices, the width and height of the minimum circumscribed rectangle are calculated, and the minimum circumscribed rectangle of each connected area is judged. The connected areas that meet the width and height ratio and have an area greater than a certain value are saved to screen out the devices to be detected.

8. The method for detecting display anomalies of multiple devices according to claim 1, characterized in that: In S4, the digital image processing determines whether the screen is lit, which is to convert the acquired image data into a grayscale image, where each pixel has only one brightness value, so as to compare it with the grayscale image when the screen is off to determine whether the screen is lit; The grayscale image conversion formula used to convert a color image to a grayscale image is: Gray=0.299×R+0.587×G+0.114×B Among them, R, G, B represent the values ​​of the red, green, and blue channels of the pixel respectively, and Gray represents the grayscale value after conversion; When calculating the absolute difference between two grayscale images, the following formula is used: Gray=|Gray1-Gray2| Among them, Gray1 and Gray2 represent the gray values ​​of corresponding pixels in two gray images, and Gray represents the calculated absolute difference; When calculating the average of all pixels in the absolute difference image, the following formula is used: Where N represents the total number of pixels in the absolute difference image, Diff i represents the absolute difference of the i-th pixel, and AvgDiff represents the calculated average value; Finally, the comparison threshold is set. The threshold is set according to the data of AvgDiff in different environments. If the average value is greater than a certain threshold, it means that the screen is on. If the average value is less than the threshold, it means that the screen is off.

9. The method for detecting display anomalies of multiple devices according to claim 1, characterized in that: In S5, the normal display image data is obtained by taking the first few groups of image data of the device to be detected with the screen on, and obtaining the average value of the grayscale image of each device to be detected with the screen on. The average values ​​of each device to be detected are then compared to exclude the device that displays abnormality at the beginning; The mean of the bright screen grayscale image is achieved by displaying normal grayscale image data and the following formula: Among them, N represents the total number of pixels in the grayscale image. i represents the gray value of the i-th pixel, and Mean represents the calculated mean; When comparing the means for each device under test, the following formula is used: Difference = |Mean1-Mean2| Among them, Mean1 and Mean2 represent the means of the two devices respectively, and Difference represents the calculated difference value; By comparing the difference values, it can be determined whether the mean difference between the two devices is obvious, thereby obtaining the screen lighting picture data that is confirmed to be displayed normally; The captured image is compared with the obtained image data showing the normal screen lighting, and the average value of all pixels in the absolute difference image is obtained using the S4 comparison method.

10. The method for detecting display anomalies on screens of multiple devices according to claim 1, characterized in that: In S6, the difference threshold means that the actual test environment is affected by light and the angle of the high-definition camera, which causes a certain difference between the obtained data and the truly displayed image. It is necessary to set a certain difference threshold to dynamically determine whether the screen display of the device under test is abnormal.

Images