Automatic detection method and device for mold penetration and coplanar defects of visual system

Abstract

The invention relates to a method and a device for automatically detecting cross-mode and coplanar defects of a visual system. The method comprises the following steps of: generating an automatic test eyespot instruction sequence; the test eyespot instruction sequence is sequentially sent to visual system software, and a visual system automatically generates a test picture corresponding to an eyespot under the driving of the test eyespot instruction, and stores the test picture to obtain a video file; performing automatic analysis and screening on the video file to obtain a scene through-mode and coplanar region; and storing eye point information corresponding to the scene through-mode and coplanar areas. According to the method, the cross-model and coplanar conflict of the visual scene model can be efficiently recognized in real time, and software research and development personnel are guided to modify problems, so that the reality sense and reliability of visual simulation software are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of visual scene detection, and particularly relates to an automatic detection method and device for the problems of model penetration and coplanarity defects in a visual scene system. Background Art

[0002] A flight simulator is used to simulate a real aircraft, and the visual scene system is a key component of the flight simulator. Through computer simulation, it realistically simulates the natural and cultural landscapes outside the cockpit, thereby improving the authenticity of flight training. With the continuous improvement of the modeling complexity of visual scene software, it is inevitable that there will be problems of model penetration or coplanarity conflicts between objects during the modeling process of complex scene elements. These problems will cause abnormal flickering phenomena at the model penetration or coplanarity positions under specific viewing angles (the scene switches between the penetrated or coplanar models back and forth, resulting in high-frequency flickering in vision). These phenomena not only affect the visual effect but may also mislead the pilot's judgment, thereby reducing the quality and effectiveness of training. Due to the extremely large simulation range of the simulator scene, it is extremely laborious and inefficient to check for model penetration and coplanarity phenomena through manual methods, and it is very easy to miss problems. In existing visual scene systems, it mainly relies on manual roaming of the scene to check for model penetration and coplanarity conflicts. Specifically, the tester manually adjusts the viewing angle and controls the observation of the scene at different eye point positions, and observes the model penetration or coplanarity between objects from multiple viewing angles. This method relies on the tester's experience and intuition to identify possible abnormal phenomena in the visual scene; after discovering a problem, the tester needs to manually record the specific position information where the model penetration and coplanarity occur for subsequent processing and correction. Although this method can discover problems to a certain extent, in complex and large-scale scenes, the manual operation workload is huge and the efficiency is low. Moreover, since many coplanarity phenomena can only be seen from specific viewing angles, and it is almost impossible to achieve traversal inspection of the scene through manual roaming, many coplanarity and model penetration problems are often missed. Therefore, the manual inspection workload is huge, and due to the limitation of the viewing angle, it is easy to miss or misjudge, affecting the comprehensiveness of detection; after discovering a problem, it is very difficult to manually record the problem viewing angle and generate a problem report; software engineers need to reproduce the problem one by one in the scene according to the problem viewing angle recorded by the tester in order to confirm and modify the problem, and the problem reproduction and troubleshooting are difficult. Summary of the Invention

[0003] In order to overcome the problems existing in the prior art, the present invention provides an automatic detection method and device for the problems of model penetration and coplanarity defects in a visual scene system to overcome the current existing defects.

[0004] An automatic detection method for the problems of model penetration and coplanarity defects in a visual scene system, the method comprising the steps of: S1. Generating an automatic test eye point instruction sequence; S2. Send the test eye point instruction sequence to the visual system software in sequence. Under the drive of the test eye point instruction, the visual system automatically generates a test screen at the corresponding eye point, saves the test screen, and obtains a video file; S3. Analyze and screen the video file to obtain the scene penetration and coplanar regions; S4. Save the eye point information corresponding to the scene penetration and coplanar regions.

[0005] In the above-mentioned aspect and any possible implementation manner, a further implementation manner is provided. The eye point instruction sequence includes eye point position parameters, and each eye point position parameter includes longitude, latitude, heading angle, altitude, pitch angle, and roll angle parameters.

[0006] In the above-mentioned aspect and any possible implementation manner, a further implementation manner is provided. The S1 includes: S11. Set the eye point parameter range and the content of the eye point instruction; S12. Generate random eye point position parameters within the eye point parameter range; S13. Detect the flickering phenomenon caused by coplanarity and penetration. If no flickering phenomenon is found, randomly select a direction and move forward along this direction at a certain step length to detect the change of the viewing angle; S14. If flickering is found, keep the current direction and continue to move forward at the set step length to observe the change of the flickering area; S15. If the flickering area becomes larger, repeat S14 until the size of the flickering area exceeds the set threshold to confirm the existence of flickering caused by coplanarity and penetration; S16. If the flickering area decreases or disappears, repeat step S13, randomly select a new direction and continue to check; S17. In the same scene, repeat S13 to S16 until the specified number of iterations is reached, and end the check.

[0007] In the above-mentioned aspect and any possible implementation manner, a further implementation manner is provided. The S2 includes: S21. When receiving the test eye point instruction, record the test screen generated by the visual system at the corresponding eye point and keep it still for a certain period of time; S22. After the still time ends, start recording and start capturing and recording the test screen generated by the visual system; S23. Record for a preset period of time to obtain the required video segment; S24. After recording for the preset time, stop recording and generate a recording end flag, and automatically save the video file of this recording.

[0008] For the aspects and any possible implementation manners described above, a further implementation manner is provided. S3 includes: S31. Extract each frame of the video file to obtain each frame of color image; S32. Convert each frame of color image into a grayscale image; S33. Perform inter-frame difference on three consecutive frames of the grayscale images, calculate the pixel differences between adjacent frames, so as to obtain a first difference image and a second difference image; S34. Perform combined calculation on the first difference image and the second difference image to obtain a combined difference image; S35. After performing difference accumulation, normalization processing, binarization processing and superposition on the combined difference image, obtain the scene penetration and coplanar regions.

[0009] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The difference accumulation specifically includes: performing accumulation processing on the combined difference image to obtain an accumulated difference image; performing maximization calculation on all pixel values of the accumulated difference image to obtain the maximum pixel value of the accumulated difference image.

[0010] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The normalization processing includes: performing normalization processing on all pixels of the accumulated difference image, and mapping all pixel values to the range of [0, 1].

[0011] For the aspects and any possible implementation manners described above, a further implementation manner is provided. The binarization processing includes: comparing the pixel value of a certain point of the normalized accumulated difference image with a set threshold. When the pixel value of a certain point of the accumulated difference image is greater than or equal to the set threshold, the region where the point is located is used as a significantly changed region and the value of the point is set to 1; when the pixel value of a certain point of the accumulated difference image is less than the set threshold, the region where the point is located is used as an unchanged region and the value of the point is set to 0.

[0012] The present invention also provides an automatic detection device for scene penetration and coplanar defects of a visual scene system. The device is used to implement the method described above and includes the following modules: A generation module, used to generate an automatic test eye point instruction sequence; A recording module, used to sequentially send the test eye point instruction sequence to the visual scene system software. Under the drive of the test eye point instruction, the visual scene system automatically generates a test picture at the corresponding eye point, saves the test picture, and obtains a video file; A screening module, used to automatically analyze and screen the video file to obtain the scene penetration and coplanar regions; A saving module for saving the eye point information corresponding to the scene penetration and coplanar regions.

[0013] Advantages of the present invention The automatic detection method for the penetration and coplanar defects of the vision system of the present invention, the method comprising the steps of: generating an automatic test eye point instruction sequence; sequentially sending the test eye point instruction sequence to the vision system software, and under the drive of the test eye point instruction, the vision system automatically generates a test picture at the corresponding eye point, saves the test picture to obtain a video file; automatically analyzing and screening the video file to obtain the scene penetration and coplanar regions; saving the eye point information corresponding to the scene penetration and coplanar regions. The method of the present invention can identify the penetration and coplanar conflicts of the vision scene model in real time and efficiently, and guide software R & D personnel to modify problems, thereby improving the realism and reliability of the vision simulation software. The present invention has the following beneficial effects: 1). By automatically generating the test eye point instructions for perspective sampling, it can ensure comprehensive and multi-dimensional (including longitude and latitude, pitch angle, heading angle, etc.) coverage sampling of the scene, effectively eliminating the problem omissions caused by the limitations of manual inspection perspectives, and improving the comprehensiveness and accuracy of detection.

[0014] 2). The present invention adopts the automatic recording and picture change analysis technology, completely getting rid of the manual experience judgment, and significantly reducing the risk of human misjudgment and omission.

[0015] 3). Automatically record the eye point parameters of the problem area and generate a detailed problem report, which greatly accelerates the process of problem discovery, location and handling. This greatly reduces the operation difficulty and time cost when software engineers reproduce problems, and moreover, testers no longer need to manually record and explain the problem perspectives, reducing communication barriers and operation errors, and improving the efficiency of problem troubleshooting and correction.

[0016] 4). Based on the automated inter-frame difference analysis and threshold screening, the present invention can accurately extract the significant change regions in the scene caused by penetration or coplanarity. Combining the cumulative difference image and binary image processing enhances the stability and reliability of anomaly detection, providing an accurate basis for scene optimization. Description of the drawings

[0017] Figure 1 is the flowchart of the method of the present invention; Figure 2 is the flowchart of automatically analyzing and screening the scene penetration and coplanar regions of the present invention; Figure 3 is the binary image of the cumulative difference result of the present invention; Figure 4 is the first frame image of the video; Figure 5 is the result image after superimposing processing on the first frame image. Detailed implementation manners

[0018] For a better understanding of the technical solution of the present invention, the content of the present invention includes but is not limited to the following specific implementation manners. Similar technologies and methods should be regarded as falling within the scope of protection of the present invention. To make the technical problems to be solved, technical solutions and advantages of the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0019] It should be clear that the embodiments described in the present invention are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts belong to the scope of protection of the present invention.

[0020] As Figure 1 shown, an automatic detection method for view system penetration and coplanarity defects provided by the present invention includes the following steps: S1. Generate an automatic test eye point instruction sequence; S2. Send the test eye point instruction sequence to the view system software in sequence. Under the drive of the test eye point instruction, the view system automatically generates a test picture at the corresponding eye point, saves the test picture, and obtains a video file; S3. Automatically analyze and screen the video file to obtain the scene penetration and coplanarity areas; S4. Save the eye point information corresponding to the scene penetration and coplanarity areas.

[0021] Now, the definitions of the terms used in the present invention are as follows: Penetration refers to the phenomenon of incorrect spatial crossing or interpenetration on the surfaces of two or more objects in the view system, that is, the geometric model or texture part of one object penetrates through another object, resulting in an unreasonable visual effect during the rendering of the view system. Penetration occurs in a dynamic scene. When the position, perspective of the geometric model of the object or calculation errors during the animation process cause improper handling of the overlapping relationship of the objects, a visual anomaly of "geometric body interpenetration" appears locally or globally.

[0022] Coplanarity refers to the phenomenon of incorrect consistent overlap in space on the surfaces of two or more objects in the view system, that is, the surfaces of multiple objects are in the same spatial layer (or very close layers), resulting in a depth competition problem during the rendering of the view system. That is, in the depth buffer, it is difficult to determine the front and back relationship of these surfaces, resulting in unstable display effects such as flickering and jittering.

[0023] An eye point is the source position of the camera or perspective in the view system, that is, the visual center point of the observer or the imaging device. The setting of the eye point determines the display state of the object under the current perspective.

[0024] Specifically, the specific steps of the present invention are as follows: 1. Generate an automatic test eye point instruction sequence The eye point instruction sequence includes eye point position parameters for perspective sampling and other signals. The eye point position parameters include six dimensions: longitude, latitude, heading angle, altitude, pitch angle, and roll angle. After setting the ranges of the above eye point position parameters according to specific requirements, an automatic test eye point instruction sequence is generated. Each eye point instruction sequence includes the eye point position parameters shown in Table 1.

[0025] The settings of the eye point parameter ranges are as follows: Longitude and latitude ranges: In the present invention, they are set to the maximum and minimum ranges of the current test scenario; Pitch angle range: Simulating the normal pitch angle during aircraft flight, it is set from -30° to 15°; Roll angle: Fixed at 0°, regarded as a perspective test without roll angle; Heading angle: Set from 0° to 360° to cover the full range of perspectives; Altitude: Set according to the requirements of the test scenario, which is the effective altitude range of the scenario.

[0026] The eye point instruction sequence is used to set the viewing range and perspective state of the visual scene system, so that the visual scene system generates a visual scene picture that meets the specified position and perspective conditions; the eye point instruction sequence is also used to drive the visual scene system to generate scene pictures at each position. Each eye point parameter is generated by a random generation method, which can ensure the comprehensiveness of the test range and not miss key positions; the recording signal bAutoRecord in the eye point instruction sequence can start or stop video recording to ensure the accurate storage of the scene data corresponding to the eye point perspective and provide a test basis for subsequent dynamic analysis.

[0027] Table 1 Eye point parameters included in the eye point instruction

[0028] To ensure the diversity and comprehensive coverage of the test eye points, the present invention randomly generates test eye point position parameters: Among the eye point position parameters: Longitude L is randomly generated within the range Latitude B is randomly generated within the range Altitude A is randomly generated within the range Heading H is randomly generated within the range Pitch angle P is randomly generated within the range Roll angle R is randomly generated within the range The formulas for randomly generating the above position parameters are as follows:

[0029]

[0030]

[0031]

[0032]

[0033] (1), where rand(0, 1) represents generating a random number in the range [0, 1].

[0034] The above-mentioned eye point position parameters are the basis for generating a dynamic viewing angle. By randomizing the viewing angle position and detection direction, they are used to ensure full-scene coverage of the visual scene system.

[0035] 2. Record the test scene images Send the test eye point instruction sequence generated in step 1 to the visual scene system software in sequence. The visual scene system will automatically generate the test images at the corresponding eye points under the drive of the test eye point instructions; the test eye point instruction sequence is also sent to the screen recording software resident on the computer where the visual scene system is located. This software can automatically record the screen displaying the test images according to the received instructions and save them as video files. The present invention does not impose requirements on the specifications of the screen recording software.

[0036] The following is the specific recording method: a. When a new test eye point instruction is received, the screen recording software first keeps the screen content static for a certain period of time, such as 5 seconds. This static waiting is to ensure that the visual scene system has enough time to generate and fully load the test image content at the corresponding eye point, preventing image changes or flickers caused by incomplete loading. During this period, the visual scene system should have completed all necessary loading and rendering work and be ready for the next recording.

[0037] b. After keeping static for 5 seconds, the screen recording software automatically starts the recording function and begins to capture and record the test images generated by the visual scene system.

[0038] c. The recording lasts for a preset period of time, such as 10 seconds, to ensure that a long enough video segment is captured for a comprehensive evaluation of the scene under the test eye point.

[0039] d. After recording for 10 seconds, the screen recording software automatically stops the recording and generates a recording end flag. This flag not only indicates the completion of the current recording session but also serves as a signal to trigger the next recording cycle.

[0040] e. Once the recording end flag is confirmed, the screen recording software automatically saves the video file of this recording.

[0041] f. After successfully saving the video file, prepare to receive the next test eye point instruction and enter the next recording cycle.

[0042] 3. Automatically analyze and screen for scene penetration and coplanar regions For each video file generated in step 2 and used as a test, perform the following operations. The method flow is as Figure 2 shown, and the specific steps are as follows: a. Extract video frames one by one A video file consists of a series of consecutive image frames. To meet subsequent processing and analysis requirements, the video file needs to be frame-serialized, that is, extracted frame by frame. Each frame is separated from the video file stream and saved as a separate image file, giving it the form of a static image. For the extracted video frames one by one, the video content therein can be analyzed and operated on frame by frame, providing basic support for subsequent processing of scene penetration detection or coplanar region screening.

[0043] b. Convert to grayscale image After completing the extraction of frames one by one, to reduce the computational complexity and improve the processing speed, the color image is converted to a grayscale image, retaining only the luminance information and removing the RGB color information, thereby simplifying the image data structure, reducing redundant information, and providing a more efficient input basis for subsequent processing.

[0044] Assume that the pixel coordinates of the image are (x, y), where x represents the pixel position of the image on the horizontal axis (lateral), and y represents the pixel position of the image on the vertical axis (longitudinal). Extract the video frames of three color images at equal intervals, where the three frames are set as the first n frames, the current frame, and the last n frames, and n is an integer greater than or equal to 1. The converted grayscale images are respectively denoted as the first n frames t - n, the current frame t, and the last n frames t + n. Modifying different values of n is equivalent to sampling the original video file at different intervals, thereby matching different scene flicker frequencies. In the present invention, n = 1 is used for illustration.

[0045] The grayscale conversion formula is as follows: The pixel value of the grayscale image of the previous frame is: (2) The pixel value of the grayscale image of the current frame is: (3) The pixel value of the grayscale image of the next frame is: (4) Where: respectively represent the pixel values of the grayscale images of the previous frame, the current frame, and the next frame at the coordinate (x, y).

[0046] , and They respectively represent the pixel values of the red components of the previous frame, the current frame, and the next frame of the original color image at the coordinate (x, y).

[0047] , : They respectively represent the pixel values of the green components of the previous frame, the current frame, and the next frame of the original color image at the coordinate (x, y).

[0048] , , : They respectively represent the pixel values of the blue components of the previous frame, the current frame, and the next frame of the original color image at the coordinate (x, y).

[0049] c. Inter-frame difference calculation Perform inter-frame difference on the grayscale images of three consecutive frames to calculate the pixel differences between adjacent frames, thereby obtaining two difference images D1 and D2. The specific steps are as follows: The difference value between the previous frame and the current frame is: (5) The difference value between the current frame and the next frame is: (6) Where: represents the pixel value of the three grayscale images at the coordinate (x, y); , represents the difference value at the coordinate (x, y) after the two differences, reflecting the change of the pixel values of adjacent frames.

[0050] d. Merge the difference images To further remove noise and accurately retain the changed areas, perform a bitwise AND operation on each pixel of the two difference images D1 and D2 to obtain a candidate image for the changed area or a merged difference image .

[0051] Merged difference image The calculation formula for the value at the pixel point (x, y) in the (7) Where: ∧ represents the bitwise AND operation; represents the value at the pixel point (x, y) in the merged difference image.

[0052] e. Difference accumulation and binarization In the previous steps, a candidate image for the changed area is obtained through the merged difference image . To further highlight the significantly changed areas in the video, it is necessary to perform frame-by-frame accumulation on all the merged difference images. By traversing and comparing all the merged difference images of all frames and performing frame-by-frame accumulation, a cumulative difference image of all frames is constructed , the cumulative difference image integrates the information of all instantaneous changes in the entire video sequence, thus being able to highlight the regions with significant changes during the entire video. The direct manifestations of penetration and coplanarity are the flickering phenomenon. By analyzing the flickering regions, the significant change regions in the visual scene can be effectively identified and highlighted.

[0053] Since directly adding the difference images may cause the pixel values to exceed the standard range (0~1), after generating the cumulative difference image, the present invention normalizes it. Normalization remaps the pixel values of the cumulative difference image to the range of 0~1 for subsequent processing and analysis.

[0054] Calculating the cumulative difference image The formula for the pixel value at coordinates (x,y) is as follows: (8) where: N represents the total number of frames of the video. t is the frame number, representing the discrete points of time in the video, with the range ; S(x,y) is the pixel value of the cumulative difference image at coordinates (x,y).

[0055] The maximum value of the cumulative difference image is: (9) where: represents the maximum pixel value in the cumulative difference image .

[0056] The normalized cumulative difference image is: (10) where represents the pixel value of the normalized difference image at coordinates (x,y), and the pixel value is mapped to the range [0,1].

[0057] Binarization processing: The value of the normalized cumulative difference image at coordinates (x,y) is binarized to obtain a binary image. Specifically, by setting a threshold T, preferably T = 10 in the present invention, is binarized to obtain the value , so that the binary image has a region showing changes. The conversion process is as follows: (11) When the pixel value is greater than or equal to the threshold T, = 1, and the region where this 1 value is located is marked as a white region, which is regarded as a significant change region, indicating that the pixel point (x,y) belongs to the significant change region; when the pixel value is less than the threshold T = 0, the area where the 0 value is located is marked as a black area, indicating that the pixel point (x, y) does not belong to the significantly changing area. Among them: T represents the threshold value, which is used to distinguish and screen the significantly changing area, so as to obtain a binary image including a white area and a black area, as Figure 3 shown.

[0058] 4. Detection and confirmation of changing area In step 3, after analyzing and calculating the video frame difference using a set threshold value, a binary image including a significantly changing area and / or a black area is obtained. The significantly changing area may have significantly changing dynamic objects or areas, such as flashing points, occluded / occluding objects, etc. In this step, the changing area or the black area is further detected and confirmed to obtain the significantly changing area. If there is flashing in the changing area, it indicates the existence of model penetration and coplanarity. Therefore, the direct manifestation forms of model penetration and coplanarity are flashing phenomena. By analyzing whether there is a flashing phenomenon in the changing area, the significantly changing area in the visual scene can be effectively identified and highlighted. The specific steps are as follows: (1) Detect in the changing area. If no flashing phenomenon is found in this changing area, it means that no significant change is found in this changing area. Then return to the previous step 1, send a test eye point instruction sequence to the visual scene system software, randomly select a direction, and move forward along this direction with a step size of 1 / 1000 of the scene width to detect the change of the viewing angle.

[0059] (2) If a flashing phenomenon is detected in the changing area, then this changing area is judged as a significantly changing area. Then return to the previous step 1 to send a test eye point instruction sequence to the visual scene system software, but keep the current direction, and set the continuous forward step size to 1 / 1000 of the scene width, and observe the change of the flashing area in the changing area.

[0060] (3) If it is found through the observation in step (2) that the flashing area becomes larger, then repeat step (2) until the size of the flashing area exceeds the threshold value set when there was a previous picture difference, confirm that there is a flashing problem caused by coplanarity and model penetration in this changing area, then extract the first frame of the video as the reference background image, as Figure 4 shown, and use the binary image to perform superposition processing on this background image. Cover the white area (representing the significantly changing area) in the binary image onto the reference background image, while the black area keeps the original content of the reference background image unchanged, so as to visually display the changing area in the video on the reference background image, as Figure 5 shown.

[0061] Such as Figure 4The reference background image or the original image shown is the first frame image of the extracted video, which is generated during the detection process of the vision system of the present invention and is used as a reference background image for detection and analysis. Since this image is the first frame of the video, it presents a static picture in the initial state and does not contain any pixel content with dynamic changes (such as flickering, occlusion, or penetration phenomena). Therefore, it fully reflects the static layout and details of the scene. Since the first frame is stationary or close to the relatively stable initial state of the overall scene, it can intuitively reflect the environmental basic state from the starting moment of the scene test to the end of the analysis, serving as a reference for subsequent comparison of video dynamic changes. Taking the first frame, that is, the initial static picture, as the reference background image or the original image, and performing an overlay process on it using the generated binary difference map. This overlay operation can intuitively present the changed areas on the original image, facilitating the quick positioning of the coplanar and penetration positions, thereby effectively troubleshooting problems.

[0062] The picture obtained after the overlay process is as Figure 5 shown. The changed areas (white) are highlighted, covering the corresponding pixel points of the original image; and these changed areas or white areas point to the dynamic objects with significant changes, such as flickering points, occluded / occluding objects, penetration or coplanar areas, which are the scene penetration and coplanar areas obtained by analyzing and screening the test pictures generated by the vision system; the black part retains the original static content of the background, making the result intuitively show the contrast relationship between the background and the changed areas. Figure 5 The changed areas shown display the dynamic scenes with significant changes in the video (such as flickering phenomena, blurred or textured changed areas), highlighting the positions and ranges of the target areas, providing data support for detecting penetration and coplanarity, and also directly showing the distribution of dynamic changes and the severity within the time interval, thereby helping to optimize subsequent dynamic analysis and judgment.

[0063] (4) If it is observed in step (2) that the flickering area decreases or disappears, repeat step (1), randomly select a new direction and continue the detection.

[0064] (5) To ensure the comprehensiveness of perspective sampling, in the same scene, repeat the above detection operations until iterated to the set number of times, such as 100,000 times, thereby completing the inspection.

[0065] Each time the eye point position changes, this method randomly generates new eye point position information through step 1 and sends it to the vision system. Subsequently, according to steps 2 - 3, the frame difference analysis results are processed to detect whether there are significant changed areas, and then through step 4, the changed areas are deeply detected and confirmed. Through multiple loop detections, the entire scene can be comprehensively covered, and finally the detection results are output.

[0066] 5. Problem Recording and Phenomenon Reproduction For the significant change regions that are considered to have penetration and coplanarity phenomena after detection in step (3) of step 4, save the eye point information including longitude, latitude, pitch angle, roll angle, yaw angle, and altitude that is considered to be used in the significant change region process after detection in step (3), and record it in a file. When conducting subsequent tests, using these saved eye point information can reproduce the corresponding penetration and coplanarity regions. Software developers can repeat driving the visual scene according to the recorded eye point information to confirm and modify the visual scene.

[0067] As an embodiment disclosed in the present invention, the present invention also discloses an automatic detection device for penetration and coplanarity defects of a visual system. The device is used to implement the method, and includes the following modules: A generation module, used to generate an automatic test eye point instruction sequence; A recording module, used to sequentially send the test eye point instruction sequence to the visual system software. Under the drive of the test eye point instruction, the visual system automatically generates a test picture at the corresponding eye point, saves the test picture, and obtains a video file; A screening module, used to automatically analyze and screen the video file to obtain the scene penetration and coplanarity regions; A saving module, used to save the eye point information corresponding to the scene penetration and coplanarity regions.

[0068] As an embodiment disclosed in the present invention, the present invention also discloses an electronic device, which includes: A memory, storing executable instructions; A processor, the processor runs the executable instructions in the memory to implement the method of the present invention.

[0069] As an embodiment disclosed in the present invention, the present invention also discloses a computer storage medium, on which a computer program is stored, and the computer program is executed by a processor to implement the method of the present invention.

[0070] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms of "a", "the", and "said" used in the embodiments of the present invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0071] The foregoing description has shown and described several preferred embodiments of the present invention. However, as previously mentioned, it should be understood that the present invention is not limited to the forms disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the above teachings or the skills or knowledge in the relevant field. Any alterations and changes made by those skilled in the art without departing from the spirit and scope of the present invention shall fall within the protection scope of the appended claims of the present invention.

Claims

1. An automatic detection method for mold penetration and coplanar defects of a vision system, characterized in that: The method comprises the steps of: S1. Generate automatic test eye point instruction sequence; S2. The eye test command sequence is sent to the visual system software in sequence, and the visual system automatically generates a test screen at the corresponding eye point under the drive of the eye test command, and saves the test screen to obtain a video file; S3. Automatically analyze and screen the video file to obtain scene penetration and coplanar areas; S4. Save the eye point information corresponding to the scene penetration and coplanar area.

2. The method according to claim 1, characterized in that The S1 includes: S11. Set the eye point parameter range and eye point instruction content; S12. generating random eye point position parameters within the eye point parameter range; S13. Detect flickering caused by coplanarity and mold penetration. If no flickering is found, randomly select a direction and move forward along the direction at a certain step length to detect changes in viewing angle; S14. If flickering is found, maintain the current direction, continue to move forward according to the set step length, and observe the changes in the flickering area; S15. If the flickering area becomes larger, repeat S14 until the size of the flickering area exceeds the set threshold, confirming the existence of flickering caused by coplanarity and mold penetration; S16. If the flickering area decreases or disappears, repeat step S13, randomly select a new direction and continue checking; S17. In the same scene, repeat S13 to S16 until the specified number of iterations is reached, and then end the inspection.

3. The method according to claim 1, characterized in that The S2 includes: S21. When a new test eye point instruction is received, the test picture remains still for a certain period of time; S22. After the static time ends, the recording function is automatically started to capture and record the test screen generated by the visual system; S23. Recording continues for a preset period of time to obtain the desired video clip; S24. After the preset recording time, the recording is stopped and a recording end mark is generated, and the video file of this recording is automatically saved.

4. The method according to claim 1, characterized in that: The S3 includes: S31 extracts the video file frame by frame to obtain a color image of each frame; S32. Converting each color image frame into a grayscale image; S33. performing inter-frame difference on the grayscale images of three consecutive frames, calculating pixel differences between adjacent frames, thereby obtaining a first differential image and a second differential image; S34. merging the first differential image and the second differential image to obtain a merged differential image; S35. After performing difference accumulation, normalization, binarization and superposition on the merged differential image, a scene penetration and coplanar area are obtained.

5. The method according to claim 4, characterized in that The difference accumulation specifically includes: performing accumulation processing on the combined difference image to obtain a cumulative difference image; and performing maximization calculation on all pixel values ​​of the cumulative difference image to obtain a maximum pixel value of the cumulative difference image.

6. The method according to claim 5, characterized in that The normalization process includes: performing normalization process on all pixels of the cumulative difference image, and mapping all pixel values ​​to a range of [0, 1].

7. The method according to claim 6, characterized in that The binarization processing includes: comparing the normalized pixel value of the cumulative difference image with a set threshold value; when the pixel value of a point in the cumulative difference image is greater than or equal to the set threshold value, setting the value of the point in the significant change area to 1; when the pixel value of a point in the cumulative difference image is less than the set threshold value, setting the value of the point in the significant change area to 0.

8. An automatic detection device for mold penetration and coplanar defects in a visual system, characterized in that: The device is used to implement the method described in any one of claims 1 to 7, and includes the following modules: A generation module, used for generating an automatic test eyepoint instruction sequence; A recording module, used for sending the eye point test instruction sequence to the visual system software in sequence, and the visual system automatically generates a test picture at the corresponding eye point under the drive of the eye point test instruction, and saves the test picture to obtain a video file; A screening module, used for automatically analyzing and screening the video file to obtain scene penetration and coplanar areas; The saving module is used to save the eye point information corresponding to the scene penetration and coplanar area.

9. An electronic device, characterized in that: The electronic device comprises: A memory storing executable instructions; A processor, wherein the processor runs the executable instructions in the memory to implement the method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that: The medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 7.

Images