A test method for color development quality of photochromic paint protective film

Abstract

This invention relates to the field of image analysis technology, specifically a method for testing the color development quality of photochromic paint protective films. The method includes: acquiring initial state images and color-changing state images from multiple angles; determining candidate regions for color development anomalies and corresponding anomaly labels for the effective test areas of the initial state and color-changing state images, using the brightness and chromaticity changes of adjacent pixels within the effective test areas as the subject; comparing the candidate regions from multiple angles, verifying the deformation of the candidate regions based on the differences between the angles, and considering the verified regions as the output target regions; setting influence weights based on the brightness and chromaticity of the target regions under multiple color changes, and performing secondary detection on the target regions in conjunction with fading time to calibrate the image range information of the target regions; and configuring the test results of the target regions based on the image range information of the target regions. This method achieves both accuracy and processing efficiency in paint protective film analysis.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, specifically a method for testing the color development quality of photochromic paint protective films. Background Technology

[0002] In recent years, paint protection films have gained increasing acceptance among consumers. These films, based on traditional paint protection films, add the ability to change color under specific lighting conditions. However, problems remain in testing, including inconsistent test results, unclear area localization, and varying testing standards, leading to reduced testing accuracy for paint protection films.

[0003] For example, Chinese Patent Publication No. CN117969550A discloses a method and system for analyzing automotive defects based on image recognition. The method includes the following steps: identifying the defect range information of defects appearing on the surface of the vehicle through an imaging device and activating a lighting device, wherein the lighting device is set towards the area to be detected and the light-emitting structure of the lighting device is regular; sending a capture command to the imaging device for capturing the light area reflected by the lighting device on the surface of the vehicle; after the reflected light area is captured, acquiring image data of the area where the reflected light area completely covers the defect on the surface of the vehicle through the imaging device; and identifying and analyzing the reflected light area in the image data based on image recognition technology.

[0004] For example, Chinese Patent Publication No. CN119178770A discloses a method, device, system, and equipment for dynamic detection of vehicle body paint defects. The method includes: acquiring a first vehicle body image acquired by a fixed bracket image acquisition component and a second vehicle body image acquired by a robotic arm handheld image acquisition component; identifying a first vehicle body paint defect result in the first vehicle body image and a second vehicle body paint defect result in the second vehicle body image; jointly calibrating the camera in the fixed bracket image acquisition component and the camera in the robotic arm handheld image acquisition component to obtain a joint camera calibration result; and mapping the first vehicle body paint defect result and the second vehicle body paint defect result to the vehicle body three-dimensional coordinate system based on the joint camera calibration result to obtain the vehicle body paint defect detection result.

[0005] In existing technologies, defective areas that deform after reflection are identified by the principle of light reflection; and dynamic defects of the paint surface are identified by three-dimensional calibration of the vehicle body. These methods tend to focus on static color detection, ignoring color shifts caused by the material itself and color shifts caused by viewing angles. This results in low efficiency in locating abnormal color areas under multiple tests and insufficient accuracy in marking after multi-angle verification, which in turn affects the evaluation of the color development quality of the paint protection film. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for testing the color development quality of a photochromic paint protective film, comprising: acquiring the paint protective film to be tested, and acquiring initial state images and color-changing state images from multiple angles by arranging the film with different light inrush angles.

[0007] For the effective test areas of the initial state image and the color-changed state image, the brightness change and chromaticity change direction of adjacent pixels in the effective test areas are used as the subject to determine the candidate areas of color abnormality and the corresponding abnormal labels.

[0008] Compare candidate regions from multiple angles, check the deformation of candidate regions by the differences between multiple angles, and regard the checked region as the output target region.

[0009] The influence weights are set based on the brightness and chromaticity of the target area under multiple color changes. Each influence weight is set based on the average brightness and average chromaticity of the area. The target area is then subjected to secondary detection in combination with the fading time to determine the image range information of the target area.

[0010] Based on the image range information of the target area, the position information corresponding to different light inflow angles is determined, and the test results of the target area are configured by combining the number of color changes in the target area.

[0011] The beneficial effects of this invention are as follows: First, this invention clarifies the color abnormality areas under multiple perspectives and improves the processing accuracy for color abnormality areas by acquiring initial / color-changing state images from multiple angles → determining candidate areas and abnormal labels within the effective test area → verifying deformation of candidate areas from multiple angles and outputting target areas → secondary detection of target areas and image range calibration → integrating multi-angle position information and configuring time-series test results.

[0012] Second, this invention uses pixel brightness and chromaticity as the basis for quantization, and configures anomaly labels for multiple candidate regions by aggregating abnormal pixels. It also combines single candidate regions with filtering conditions to complete the configuration of abnormal regions, realizes hierarchical filtering within the effective test area, improves the aggregation efficiency of multiple regions under a single angle, and improves the consistency of anomaly label configuration.

[0013] Third, this invention identifies the deformation of candidate regions under different light inrush angles due to angular distortion by setting a vertical incident reference point and judging the same source candidate regions; it uses the same source candidate regions to remove normal deformation and non-same source candidate regions to filter abnormal labels, eliminates regions with significant color abnormalities, and obtains the output target region by merging multiple angles; it avoids the abnormal identification caused by angle and ensures the accuracy and non-repetition of the output target region position.

[0014] Fourth, this invention uses the fading time as the correlation value between the initial state image and the color-changed state image, and performs a secondary judgment based on the brightness and chromaticity of each group of images. Based on the fading time series, it compares the maximum brightness value and the fading time length value as constraints to separate abnormal areas under different trend combinations, ensuring that the image range information can reflect the true abnormal range of the target area. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] Figure 1 This is a flowchart illustrating a method for testing the color development quality of a photochromic paint protective film.

[0017] Figure 2 This is a flowchart illustrating step S2 of a method for testing the color development quality of a photochromic paint protective film.

[0018] Figure 3 This is a flowchart illustrating step S3 of a method for testing the color development quality of a photochromic paint protective film.

[0019] Figure 4 This is a flowchart illustrating step S4 of a method for testing the color development quality of a photochromic paint protective film. Detailed Implementation

[0020] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0021] See Figure 1 A method for testing the color development quality of a photochromic paint protective film includes: S1, acquiring the paint protective film to be tested, and acquiring initial state images and color-changing state images from multiple angles by arranging the film with different light inrush angles.

[0022] S2, for the effective test areas of the initial state image and the color-changed state image, using the brightness change and chromaticity change direction of adjacent pixels in the effective test area as the subject, determine the candidate areas of color abnormality and the corresponding abnormal labels of the candidate areas.

[0023] S3 compares candidate regions from multiple angles, checks the deformation of candidate regions based on the differences between multiple angles, and takes the checked region as the output target region.

[0024] S4: Based on the brightness and chromaticity of the target area under multiple color changes, set the influence weight, and combine the fading time to perform secondary detection on the target area to calibrate the image range information of the target area.

[0025] S5 determines the position information corresponding to different light inflow angles based on the image range information of the target area, and configures the test results of the target area by combining the number of color changes in the target area.

[0026] The initial state image is the image without activated photochromic conditions, used as a comparison of the color change of the paint protective film after irradiation; the color-changing state images at different incident angles are used to verify the color-changing results under different lighting conditions.

[0027] The core of photochromism lies in the photochromic molecules within the material. When these molecules are excited by a specific wavelength (usually ultraviolet light), their chemical structure undergoes a reversible change, altering their absorption characteristics for visible light. Macroscopically, this manifests as a darkening of color. For example, when a car is driven outdoors, ultraviolet light from sunlight shines on the car cover, activating the photochromic molecules, changing their structure, and causing the film to rapidly darken in color, commonly to dark gray, black, or blue, thus producing a matte or satin-like texture. Subsequently, when the ultraviolet light weakens, such as in an underground parking garage, tunnel, or at night, the molecular structure returns to its original state, and the film gradually returns to its original transparent or near-transparent state within minutes; this completes the photochromic effect of the paint protection film.

[0028] In a specific scenario, spiropyran / spiroxazine series are used as coating materials. Their molecules are closed-ring spirocyclic structures (colorless) in the stable state. After absorbing ultraviolet light, the carbon-oxygen bonds break, and the molecules open-ring isomerize into cyanine structures, which absorb visible light and thus become colored.

[0029] Alternatively, diarylethylene series can be used as coating materials. When its molecules are irradiated with ultraviolet light, they undergo a 6π electrocyclization reaction, changing from an open-ring state to a closed-ring state, which leads to the elongation of the conjugated system and a red shift in the absorption spectrum, resulting in color change. The color change reaction is due to bond isomerization rather than bond breakage.

[0030] In step S1, when acquiring the initial state image and the color-changing state image, it is also necessary to record the time required for the color change in order to determine the time interval between the color-changing state and the initial state. For example, the fading time from the activation of the light source being turned off to the paint protective film fading to the set reference state can be selected to identify the brightness and chromaticity corresponding to the paint protective film, so that the acquired images can be aligned in a time sequence manner and the effect produced by the color change can be verified.

[0031] Therefore, the implementation of the effective test area in step S2 also includes: obtaining the fading time of the paint protective film, using the fading time as the correlation value between the initial state image and the color-changing state image, and recording each set of initial state images and color-changing state images.

[0032] Based on the projected area of ​​the current light inflow angle, the region corresponding to each set of initial state images and color-changing state images is determined, and the spatial information of each set of images is synchronized. The synchronized region is regarded as the output valid test region.

[0033] When the light inflow angle is known, the color-changing area can be found from each set of initial state images and color-changing state images based on the normal projected area under the current incident light direction. These areas are then synchronized to determine the color-changing situation at corresponding positions under multiple incident angles. The image preprocessing methods used include, but are not limited to, angle registration, geometric correction, and image normalization to complete the spatial information synchronization of the images.

[0034] The current incident angle can be set to 0°, 15°, 30°, 45°, 60°, etc., and ultraviolet light is used as the light source for photochromism; multiple cameras are configured to complete the image acquisition of the current paint protection film through light sources with multiple incident angles.

[0035] For the acquired multiple images, the color changes in the images will be processed. For example, the image will be decomposed into RGB channels, with the values ​​of each pixel stored in three channels: R (red), G (green), and B (blue). The pixel values ​​will be converted to CIE Lab space to separate the brightness L and chromaticity (a, b) of each pixel before and after the color change. The brightness value is 0-100, with a larger value indicating brighter brightness. The chromaticity is divided into the red-green axis a (positive value is red, negative value is green) and the yellow-blue axis b (positive value is yellow, negative value is blue). Then, the direction of chromaticity change (such as from green to red) and the amount of brightness change when irradiated by specific incident light can be determined. This allows for the separation of candidate areas with color defects during color changes, thereby evaluating the color rendering quality of the current paint protection.

[0036] like Figure 2 As shown, the implementation method of determining the candidate region for color development anomalies and the corresponding anomaly label in step S2 includes: S21, for the color-changing state image at the same angle, comparing the color-changing state image at any time with the initial state image, and filtering the brightness change and chromaticity change of all pixels in the effective test area; it should be noted that the effective test area represents the area where color change occurs after alignment at multiple angles. This area will be used as a sample under different lighting conditions to verify the uniformity of color development, the range of color development gradient, the color shift, and the change in the time required for color development under color development changes, thereby determining whether the current paint protection film is effective.

[0037] S22, pixels that meet any one of the screening conditions of abnormal brightness change, abnormal chromaticity direction, and abnormal brightness gradient are regarded as abnormal pixels, and the abnormal pixels are aggregated to form candidate regions.

[0038] In the current scenario, the values ​​corresponding to the changes in brightness and the direction of color change will be checked. When the brightness change is abnormal, it means that the color is too dark, too light, or not displayed at all. When the brightness gradient is abnormal, it means that the difference in color depth between adjacent pixels is too large and the uniformity is poor. When the color direction is abnormal, it means that the difference in color direction between adjacent pixels is too large and there is a color cast.

[0039] When selecting abnormal pixels, the judgment method for the first three conditions is the same. The value range of the standard image of photochromism will be used as the benchmark. Pixels that exceed or fall below the benchmark will be regarded as abnormal pixels.

[0040] The aggregation of regions containing abnormal pixels can be done by using the minimum bounding circle method to define the region containing the abnormal pixels, or by connecting the abnormal pixels to form connected candidate regions.

[0041] S23, when each set of initial state images and color-changing state images contains multiple candidate regions, construct color features based on the amount of brightness change and the direction of chromaticity change of the candidate regions, compare the consistency of the color features, and obtain the anomaly label of the candidate regions; this step compares multiple candidate regions corresponding to the same incident angle to determine the parts with uniform color and the parts with color range after multiple color changes, so as to determine whether color difference or color deviation is likely to occur in the scene of photochromism.

[0042] Therefore, the method for achieving consistency of color features in step S23 includes: for the same candidate region, comparing the range of the candidate region under multiple color changes; if the range of the candidate region is consistent, then configuring anomaly labels for the candidate region based on the average value and standard deviation of the color features.

[0043] This comparison method compares whether the mean and standard deviation of the same candidate region are consistent after multiple color changes. When the mean and standard deviation of the color features are consistent after multiple color changes, it means that the candidate region belongs to the anomaly of the sample itself and has a fixed color difference, such as the color error caused by the uneven distribution of photochromic material concentration, rather than the dynamic change in the color change process.

[0044] When the mean and standard deviation of color features are inconsistent under multiple color changes, it indicates that the current candidate region is a dynamic anomaly, which often indicates material fatigue effect. That is, the response ability of photochromic molecules decays after multiple color changes, resulting in unstable color depth or direction.

[0045] If the candidate regions are inconsistent, it may indicate that the current sample has random distribution of minor defects in color development and that the photochromic performance has deteriorated overall. In this case, the candidate regions are marked based on the results of the color feature comparison.

[0046] When there are multiple candidate regions, correlation analysis is performed based on the color characteristics of each candidate region. The two candidate regions with the largest correlation values ​​are merged into one candidate region, and anomaly labels are set according to the distribution of the candidate regions.

[0047] When dealing with multiple candidate regions in the same set of samples, clustering can be performed based on the abnormal pixels corresponding to them during screening. Alternatively, correlation can be calculated directly using the mean and standard deviation of their color features. That is, the Pearson correlation coefficient can be used to calculate the color feature similarity between any two candidate regions. When calculating color features, they need to be converted into feature vectors. Then, the two most relevant candidate regions are selected and merged into one region. Anomaly labels are set based on their location to reflect the labels corresponding to the location where the color deviation occurs, such as continuous distribution, local discrete distribution, edge concentrated distribution, and other anomaly labels that include spatial location.

[0048] S24, when each set of initial state images and color-changing state images contains only one candidate region, anomaly labels are set according to the filtering conditions corresponding to the candidate region, in combination of filtering conditions; when each set of images contains only one candidate region, in addition to setting anomaly labels according to multiple color changes, anomaly labels can also be set for the current candidate region according to the filtering conditions involved, in combination of filtering conditions, to explain the abnormal situation of abnormal pixels in the current scene.

[0049] Since step S24 sets abnormal labels for the aggregation of abnormal pixels, the implementation of setting abnormal labels by combining filtering conditions also includes: viewing abnormal pixels in the current candidate area, classifying them based on the amount of brightness change and the direction of color change of the abnormal pixels, identifying the proportion of the number of abnormal pixels in each category relative to the candidate area after classification, and obtaining the abnormal label of the current candidate area.

[0050] At this point, the classification will yield categories such as luminance-dominated anomaly, chrominance-dominated anomaly, luminance gradient-dominated anomaly, luminance-chrominance anomaly, luminance-luminance gradient anomaly, and all-condition anomaly. Each category represents the pixel situation corresponding to its screening conditions. Then, based on the proportion of its abnormal pixels, it is used to indicate whether there is a composite distribution of anomalies in the candidate region, or whether it presents an overall single classification, thereby indicating the candidate region aggregated by the abnormal pixels.

[0051] In one embodiment of the present invention, in step S3, multiple incident angles are used as the analysis object to check the correlation between candidate regions at different angles, thereby determining whether the candidate regions will have the same offset under light illumination, and then statistically analyzing the target regions that are highly correlated under multiple angles.

[0052] like Figure 3 As shown, the implementation method of checking the deformation of the candidate region by the difference between multiple angles in step S3 includes: S31, taking the center coordinates of the candidate region with vertical incidence as the reference point, calculating the Euclidean distance between the center coordinates of the candidate region and the reference point at other angles, and judging the candidate regions belonging to the same source based on the Euclidean distance.

[0053] The candidate region for vertical incidence refers to the candidate region when the incident angle is 0°. This region is used as a verification reference point to determine whether the candidate regions obtained at other angles are the same region, so as to mark the candidate regions where color shift occurs at a specific angle.

[0054] Since only the effective test area was aligned, some candidate areas still have some offsets when acquired from multiple angles. At this time, it is necessary to check whether each candidate area is in a specific area of ​​the same origin or different origin, and then to obtain the subsequent target area.

[0055] When determining the same source, the offset corresponding to the incident angle and the error during image registration are generally used as the standard for judging the current Euclidean distance. At this time, the average value of the offset of the center coordinates between the corresponding incident angle and 0° angle under normal conditions and the error corresponding to the system alignment (such as 0.5 pixel value) can be combined. Images with less than this value are regarded as candidate regions of the same source, otherwise they are regarded as candidate regions under a specific angle.

[0056] S32, for candidate regions belonging to the same source, calculate the area change rate, center offset, and cross-union ratio between the current candidate region and the vertically incident candidate region; the calculated area change rate, center offset, and cross-union ratio can quantify the shift of the current color anomaly under different light refraction, and thus identify whether the current anomaly is shifted due to different incident angles.

[0057] S33, take the area change rate, center offset and intersection-to-union ratio of the current candidate region as deformation features, combine the brightness difference ratio and chromaticity direction deviation angle between the current candidate region and the vertically incident candidate region, remove the candidate regions that belong to normal deformation, and input the other candidate regions into the target region to be selected.

[0058] At this point, these five variables will be used to check the abnormal deformation of the current candidate region, distinguish between normal deformation caused by angle and abnormal deformation caused by color abnormality, and finally determine the target region; these five variables are all calculated based on the candidate region at the incident angle of 0°.

[0059] At this point, the parts where these five indicators are in a normal state will be excluded. The abnormal candidate areas are essentially normal color differences caused by the angle of light incidence, rather than color defects.

[0060] Assuming that the range of values ​​for normal deformation is ≤15% for area change rate, ≤2mm for center offset, ≥0.8 for crossover ratio, ≤10% for brightness difference ratio, and ≤8° for chromaticity direction deviation angle, then regions in the candidate area that meet these values ​​will be removed. The set range of values ​​will be adjusted according to the size, environment, and shape of the sample used during the test.

[0061] S34. For candidate regions that are not from the same origin, check the anomaly labels of the candidate regions and filter the candidate target regions by anomaly labels.

[0062] In specific candidate regions that are not homologous, it is mainly necessary to examine regions that show dynamic instability or obvious color change and degradation after multiple comparisons. Based on the data part corresponding to their abnormal labels, these regions are selected as the main target regions for verification. As for other candidate regions, the specific characteristics they exhibit under multiple angles and multiple comparisons can be directly output based on their abnormal labels to explain the abnormal color display at the corresponding positions.

[0063] When filtering by anomaly labels, the main focus is on removing non-significant single-angle anomaly areas. These color differences are mostly accidental anomalies under a single angle and do not constitute global anomalies, thus having no actual impact on color abnormalities.

[0064] S35: For candidate target areas from multiple perspectives, remove duplicate candidate target areas, merge the candidate target areas from the remaining perspectives, and regard the merged area as the output target area. At this time, remove the areas that are marked at different angles, and merge the target areas that are not repeated at multiple angles to complete the filtering and processing of the target area.

[0065] The method of merging the candidate target areas of the remaining perspectives also includes: taking the candidate target area with vertical incidence as the reference area, calculating the intersection-union ratio between the candidate target area and the reference area under other incidence angles, and considering the candidate target areas with an intersection-union ratio greater than the preset intersection-union ratio as duplicate areas; the preset intersection-union ratio can be set to 0.7 to remove most of the identical areas.

[0066] For non-repeating candidate target regions, position correction is performed based on the center offset of the candidate target region, and the position-corrected candidate target regions are combined into the output target region.

[0067] For non-repeating regions, their spatial positions need to be synchronized. For example, the center offset calculated in step S32 is used for reverse correction. The candidate target regions under other incident angles are synchronized to the position of the candidate target regions with vertical incidence. That is, the center coordinates of other incident angles are subtracted from the center offset to complete the correction of candidate target regions under other incident angles. Since the candidate target regions include the parts selected according to deformation features, the removed repetitive regions are mostly candidate regions belonging to the same source. As for non-repeating regions, they can be some regions belonging to the same source candidate regions and regions not belonging to the same source. The spatial coordinates of these regions can be corrected according to their center offset. Then, the boundary can be fitted according to these spatial coordinates to finally obtain a set of target regions with complete spatial coordinates.

[0068] It should be noted that after the candidate region processing is completed from multiple angles, the target region for subsequent processing will be processed from a single angle, that is, the target region with perpendicular incidence will be selected to complete the subsequent verification and calculation.

[0069] In one embodiment of the present invention, such as Figure 4 As shown, in step S4, the fading time of the target area under multiple color changes will be emphasized, with chroma and luminance as its vertical axis and the timestamp corresponding to the fading time as its horizontal axis, in order to verify the fading effect and the degradation caused by fading.

[0070] Therefore, the method for determining the image range information of the target area in step S4 includes: S41, using the chroma and brightness of the target area when it changes color as its vertical axis and the fading time as its horizontal axis to obtain the fading time sequence.

[0071] S42 uses the maximum brightness value and the length of the fading time at each color change in the fading time series as constraints, and compares them with the constraints during continuous color changes. The maximum brightness value represents the value at which photochromism is fully activated. The current fading time series starts with complete photochromism and verifies its fading process. This process can reflect the maximum value of photochromism and the required fading process to check whether the performance of the paint protective film has degraded.

[0072] S43, when there are differences in the constraints, set the influence weight of a single color change based on the average brightness and average chromaticity of all pixels in the target area, and determine whether the trend of the influence weight changes is consistent with the trend of the length of the fading time.

[0073] S44, if the trend is consistent, the corresponding target area is regarded as a valid target area, and the spatial position corresponding to the valid target area is regarded as the output image range information; when the trend is consistent, it means that the output image range directly reflects the significant degradation area where brightness and fading time increase or decrease at the same time; even if the aforementioned part completes the labeling of the relevant area through the method of abnormal labeling, this is to further verify whether the labeled area meets the characteristics of significant degradation, so as to obtain the valid image range.

[0074] S45, if the trends are inconsistent, the average anomaly score of multiple color changes is calculated using the influence weight of a single color change, and the decay rate of the fading time value after multiple color changes is calculated. Based on the decay rate and the average anomaly score, the output image range information is determined. In regions with inconsistent trends, the fading changes in these regions exhibit multi-layered complexity, requiring multiple comprehensive evaluations to interpret the output range of that region.

[0075] Regarding the aforementioned influence weights, each influence weight is set based on the average luminance and chrominance values ​​within the region. The weight is set by selecting the data corresponding to the time point of the maximum luminance value, using the average luminance and chrominance values ​​at that time as a ratio to the sum of the average luminance values ​​at all times corresponding to the maximum luminance values. This is equivalent to setting the two influence weights in this way: a smaller influence weight indicates a fading phenomenon in the current single color change, while a larger influence weight indicates a significant shift or fluctuation in the overall color change. In this case, output can be based on the parts exhibiting consistent behavior.

[0076] As for the average anomaly score, it will be calculated by taking the weighted average of the normalized mean luminance and mean chrominance and the influence weight to obtain the average anomaly score under multiple color changes. As for the decay rate, it represents the rate of change of the fading time from the first color change to the last color change.

[0077] It should be noted that to determine if the trend is consistent, the corresponding value needs to fluctuate by more than ±5%. Otherwise, it is considered to be in a fluctuating trend. A fluctuation of more than 5% indicates an upward trend, and a fluctuation of less than 5% indicates a downward trend. At the same time, the range of attenuation rate values ​​can be divided into the normal degradation range of attenuation rate ≤ 5%, the critical degradation range of attenuation rate between 5% and 10%, and the abnormal degradation range of attenuation rate ≥ 10%. The current range can also be adjusted based on the attenuation rate of abnormally colored samples in multiple tests.

[0078] The average anomaly score can also be selected as the no-anomaly interval (average anomaly score ≤ 0.2), the slightly abnormal interval (average anomaly score between 0.2 and 0.5), and the significantly abnormal interval (average anomaly score ≥ 0.5) to determine the image range for inconsistent trends. The configured value can also be divided according to the intervals preset in the database to obtain the preference test part in the current scenario.

[0079] The implementation of constraints during continuous color changes also includes: if there are no differences in the constraints, comparing the chromaticity corresponding to the maximum brightness value, and using the target area of ​​the maximum chromaticity value as the output image range information. When there is no significant change, the range of the maximum chromaticity will correspond to poor color uniformity, thus allowing for the quantification of relative anomalies under multiple color changes.

[0080] After completing the location positioning of various image range information, the current test results will be summarized based on the multiple light inflow angles corresponding to this part of the data and the number of color change detections, and finally the test results of the paint protection film will be obtained.

[0081] The implementation method of configuring the test results of the target area in step S5 also includes: determining the timestamp when the image range information of the target area is updated; and splicing the image range information of the target area according to the time sequence to obtain the output test results.

[0082] After completing the testing and processing of each paint protection film sample, the test results from multiple time series are summarized according to the target area output, and strung together to form a complete time series trajectory to interpret the test results for multiple sets of samples.

[0083] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A method for testing the color development quality of a photochromic paint protective film, characterized in that, include: The paint protection film to be tested was obtained, and initial state images and color-changing state images were acquired from multiple angles using different light inflow angles. For the effective test areas of the initial state image and the color-changed state image, the amount of brightness change and the direction of color change of adjacent pixels in the effective test area are used as the subject to determine the candidate areas of color abnormality and the corresponding abnormal labels of the candidate areas. Compare candidate regions from multiple angles, check the deformation of candidate regions by the differences between multiple angles, and take the checked region as the output target region; The influence weights are set according to the brightness and chromaticity of the target area under multiple color changes. Each influence weight is set based on the average brightness and average chromaticity of the area. The target area is then subjected to secondary detection in combination with the fading time to determine the image range information of the target area. Based on the image range information of the target area, determine the position information corresponding to different light inflow angles, and combine the number of color changes in the target area to configure the test results of the target area; The implementation of the effective test area also includes: obtaining the fading time of the paint protective film, using the fading time as the correlation value between the initial state image and the color-changing state image, and recording each set of initial state images and color-changing state images; determining the area corresponding to each set of initial state images and color-changing state images based on the projected area of ​​the current light inflow angle, synchronizing the spatial information of each set of images, and taking the synchronized area as the output effective test area. The method for determining candidate regions for color anomalies and their corresponding anomaly labels includes: for color-changing images at the same angle, comparing the color-changing images at any time with the initial state images, and filtering the brightness and chromaticity changes of all pixels within the valid test area; considering pixels that meet any one of the filtering conditions—brightness change anomaly, chromaticity direction anomaly, and brightness gradient anomaly—as abnormal pixels, aggregating these abnormal pixels to form candidate regions; when each set of initial and color-changing images contains multiple candidate regions, constructing color features based on the brightness and chromaticity change directions of the candidate regions, comparing the consistency of the color features, and obtaining anomaly labels for the candidate regions; when each set of initial and color-changing images contains only one candidate region, setting anomaly labels based on the filtering conditions corresponding to the candidate regions, using a combination of filtering conditions. The methods for achieving consistency in color features include: for the same candidate region, comparing the range of the candidate region under multiple color changes; if the range of the candidate regions is consistent, configuring anomaly labels for the candidate regions based on the mean and standard deviation of the color features; for multiple candidate regions, performing correlation analysis based on the color features of each candidate region, merging the two candidate regions with the largest values ​​after correlation analysis into one candidate region, and setting anomaly labels according to the distribution location of the candidate regions. The method for verifying the deformation of candidate regions based on differences between multiple angles includes: using the center coordinates of the candidate region with vertical incidence as a reference point, calculating the Euclidean distance between the center coordinates of the candidate regions at other angles and the reference point, and determining candidate regions belonging to the same origin based on this Euclidean distance; for candidate regions belonging to the same origin, calculating the area change rate, center offset, and intersection-union ratio between the current candidate region and the candidate region with vertical incidence; using the area change rate, center offset, and intersection-union ratio corresponding to the current candidate region as deformation features, and combining the brightness difference ratio and chromaticity direction deviation angle between the current candidate region and the candidate region with vertical incidence, removing candidate regions belonging to normal deformation, and inputting other candidate regions into the candidate target region; for candidate regions that are not from the same origin, checking the abnormal labels of the candidate regions, and filtering the candidate target regions by abnormal labels; for candidate target regions from multiple perspectives, removing duplicate candidate target regions, merging the candidate target regions from the remaining perspectives, and taking the merged region as the output target region; The method for determining the image range information of the target area includes: using the chroma and luminance of the target area during color change as its vertical axis and the fading time as its horizontal axis to obtain a fading time series; using the maximum luminance value and the length of the fading time at each color change as constraints, and comparing them with the constraints for continuous color changes; when there are differences in the constraints, setting the influence weight of a single color change using the average luminance and chroma values ​​of all pixels in the target area, and determining whether the trend of the influence weight changes is consistent with the trend of the length of the fading time; if the trends are consistent, the corresponding target area is considered a valid target area, and the spatial location corresponding to the valid target area is considered as the output image range information; if the trends are inconsistent, using the influence weight of a single color change, the average anomaly score of multiple color changes is obtained, and the weighted average of the normalized average luminance and chroma values ​​with the influence weight is calculated to obtain the average anomaly score under multiple color changes, and the decay rate of the fading time length value after multiple color changes is calculated. The decay rate represents the rate of change of the fading time length from the first color change to the last color change. Based on the decay rate and the average anomaly score, the output image range information is determined.

2. The method for testing the color development quality of a photochromic paint protective film according to claim 1, characterized in that, Other methods for setting exception labels by combining filtering conditions include: Examine the abnormal pixels within the current candidate region, classify them based on the amount of brightness change and the direction of color change, identify the proportion of abnormal pixels in each category relative to the candidate region, and obtain the abnormal label of the current candidate region.

3. The method for testing the color development quality of a photochromic paint protective film according to claim 1, characterized in that, Other methods for merging candidate target regions from remaining perspectives include: Using the vertically incident target area as the reference area, calculate the intersection-union ratio between the target area and the reference area at other incident angles, and consider the target area with a cross-union ratio greater than the preset cross-union ratio as a duplicate area; For non-repeating candidate target regions, position correction is performed based on the center offset of the candidate target region, and the position-corrected candidate target regions are combined into the output target region.

4. The method for testing the color development quality of a photochromic paint protective film according to claim 3, characterized in that, The implementation methods for constraints during continuous color changes also include: If there are no differences in the constraints, compare the chromaticity corresponding to the maximum luminance value, and take the target area with the maximum chromaticity value as the output image range information.

5. The method for testing the color development quality of a photochromic paint protective film according to claim 1, characterized in that, The implementation methods for configuring test results for the target region also include: When updating the image range information of the target area, determine the timestamp when the image range information is output; The image range information of the target area is stitched together according to time sequence to obtain the output test results.

Images