An image analysis-based light supplement control method, device and medium

Abstract

The application discloses a light supplement control method and device based on image analysis and a medium, relates to the technical field of traffic snapshot control, and comprises the following steps: acquiring a rolling shutter line exposure mapping parameter and a light supplement parameter, identifying a signal lamp region, a license plate region, a road marking region and an interference region, dividing the signal lamp region into a core lamp color area and a halo area, and determining line exposure time windows of the regions; extracting brightness and chrominance features along the image line direction, identifying transverse light and dark stripes and color bands, and generating evidence quality results; generating candidate light supplement parameters according to stripe analysis results, evidence quality results and adjustable boundary data of a light supplement device, taking the line exposure time windows of the core lamp color area, the license plate region and the road marking region as avoidance bases, and determining target light supplement parameters; and performing light supplement snapshot according to the target light supplement parameters. The application realizes hierarchical constraint of key evidence regions, matching of light supplement pulse edges and line exposure time windows, and closed-loop updating of light supplement control parameters.

Description

Technical Field

[0001] This invention relates to the field of traffic capture and control technology, and in particular to a supplementary lighting control method, device and medium based on image analysis. Background Technology

[0002] In traffic light capture and road image evidence collection scenarios, supplementary lighting control is typically configured by combining camera exposure parameters, supplementary light trigger delay, pulse width, and output level to improve the image discernibility of license plates, vehicle outlines, road markings, and traffic light status. Conventional methods often adjust the supplementary lighting intensity, timing, and mode based on ambient brightness, license plate reflectivity, or overall grayscale changes in the captured image, switching between white light, infrared, or hybrid supplementary lighting methods to meet the capture needs in nighttime, backlight, rain, fog, and complex intersection environments.

[0003] However, in the process of adjusting the lighting parameters, the above-mentioned conventional methods usually take the brightness of the whole image or local area as the main adjustment basis, making it difficult to distinguish between the core color area of ​​the traffic light and the outer halo area. This makes it easy for the lighting adjustment to lack graded constraints on the effect of color fidelity and halo diffusion. On the other hand, they do not make full use of the correspondence between the rolling shutter speed window and the edge of the lighting pulse, making it difficult to avoid the horizontal bright and dark stripes or colored bands in the license plate area, road marking area and key areas of the traffic light in a timely manner. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an image analysis-based supplementary lighting control method to solve the problems of insufficient hierarchical constraints on key evidence regions and insufficient matching between the edge of the supplementary lighting pulse and the line exposure time window in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an image analysis-based supplementary lighting control method, comprising: acquiring continuous image frames of a traffic light capture scene; obtaining rolling shutter line exposure mapping parameters and supplementary lighting parameters; identifying traffic light areas, license plate areas, road marking areas, and interference areas; dividing the traffic light area into a core light color area and a halo area; and determining the line exposure time window for each area; extracting brightness and chromaticity features along the image line direction; identifying horizontal bright and dark stripes and color bands; and generating evidence quality results based on core light color fidelity, core stripe overlap, halo diffusion, license plate clarity, and road marking contrast; generating candidate supplementary lighting parameters based on the stripe results, evidence quality results, and adjustable supplementary lighting boundaries; using the line exposure time windows of the core light color area, license plate area, and road marking area as the avoidance basis; hard avoiding the core light color area and softly constraining the halo area; and determining the target supplementary lighting parameters; performing supplementary lighting capture according to the target supplementary lighting parameters; verifying the stripe coverage and evidence quality of the captured image; saving available supplementary lighting parameters; or providing feedback on interference markers and failure reasons for the next cycle update.

[0008] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the step of obtaining the rolling shutter line exposure mapping parameters and supplementary lighting parameters includes: acquiring continuous image frames of the current traffic light capture scene, and recording the acquisition timestamp, image height, image width, and frame rate of each frame; acquiring the rolling shutter line readout time and single-line exposure duration of the camera, and acquiring the current supplementary lighting trigger delay, current supplementary lighting pulse width, current supplementary lighting peak level, and current supplementary lighting mode; establishing a correspondence between the image line number and the line exposure time window based on the image line number, rolling shutter line readout time, and single-line exposure duration; and reading the adjustable boundary data of the supplementary lighting device, wherein the adjustable boundary data includes the supplementary lighting trigger delay range, supplementary lighting pulse width range, supplementary lighting peak level range, minimum delay adjustment step size, and available supplementary lighting modes.

[0009] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the identification of traffic light areas, license plate areas, road marking areas, and interference areas, and the segmentation of the traffic light area into a core light color area and a halo area, includes: performing target detection and region segmentation on consecutive image frames to obtain traffic light areas, license plate areas, vehicle areas, road marking areas, strong road surface reflection areas, and strong background light source areas; extracting brightness, chromaticity, and saturation features within the traffic light area to determine the light panel area currently in the luminous state; selecting pixels in the traffic light area whose brightness values ​​are higher than the fixed percentile of the brightness of the traffic light area, and extracting candidate connected regions of the luminous light panel in combination with the main color channel features; determining the candidate connected regions whose area, roundness, and chromaticity stability meet preset conditions as the core light color area; performing morphological dilation on the core light color area to obtain an expanded area, and determining the portion of the expanded area after removing the core light color area as the halo area; and determining the line exposure time windows corresponding to the core light color area and the halo area, respectively.

[0010] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the specific steps of extracting brightness and chromaticity features along the image row direction and identifying horizontal bright and dark stripes and colored bands are as follows: extracting the average brightness, average red channel, average green channel, average blue channel, saturated pixel ratio, and local contrast of each row along the image row direction; calculating the brightness abrupt change based on the average brightness of each row and the median brightness of the neighborhood to form a row-oriented brightness curve; forming a row-oriented chromaticity curve based on the average red channel, average green channel, and average blue channel; when a continuous row segment meets the following conditions: the brightness abrupt change exceeds a preset brightness abrupt change threshold, it spans at least two regions in the horizontal direction, there is row-oriented drift or intensity periodic change in continuous image frames, and it does not belong to license plate characters, parking lines, vehicle body edges, or fixed road markings, the continuous row segment is determined as a candidate stripe region; and the colored band is determined based on the chromaticity change of the candidate stripe region.

[0011] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the specific steps for generating evidence quality results based on core light color fidelity, core stripe overlap, halo diffusion, license plate clarity, and road marking contrast are as follows: Calculate the average values ​​of the red, green, and blue channels within the core light color area, and calculate the core light color separation degree based on the current traffic light color; when the core light color separation degree is lower than the corresponding core light color separation threshold, generate a core light color anomaly marker; calculate the overlap ratio between the core light color area and the candidate stripe area, and when the overlap ratio exceeds a preset core overlap threshold, generate a core stripe crossing marker; calculate the ratio of the number of pixels in the halo area that share the same primary color as the current traffic light to the number of pixels in the core light color area, and obtain the halo diffusion ratio; calculate the overlap ratio between the halo area and the candidate stripe area, and generate a halo stripe result; combine the license plate edge clarity, license plate overexposure ratio, road marking contrast, and vehicle outline blur degree to generate an evidence quality result.

[0012] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the step of generating candidate supplementary lighting parameters based on stripe results, evidence quality results, and adjustable supplementary lighting boundaries includes: generating candidate trigger delay, candidate pulse width, candidate peak level, and candidate supplementary lighting mode based on supplementary lighting adjustable boundary data; determining the supplementary lighting on-edge time and supplementary lighting off-edge time based on the candidate supplementary lighting parameters; expanding the line exposure time windows corresponding to the core light color area, license plate area, and road marking area based on the stripe width to form a supplementary lighting edge avoidance window; determining that the corresponding candidate supplementary lighting parameter has a stripe risk when the supplementary lighting on-edge time or supplementary lighting off-edge time falls into the supplementary lighting edge avoidance window; and screening candidate supplementary lighting parameters based on stripe risk and evidence quality results.

[0013] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the determination of the target supplementary lighting parameters by hard avoidance of the core light color area and soft constraint of the halo area includes: when the chromaticity fidelity result of the core light color area is lower than the preset chromaticity fidelity threshold, the stripe overlap rate of the core light color area exceeds the preset core overlap threshold, or the signal light abnormality type is marked as core light color abnormality or core stripe crossing, the line exposure time window corresponding to the core light color area is used as the highest priority supplementary lighting edge avoidance window; candidate supplementary lighting parameters that enter the line exposure time window corresponding to the core light color area by the supplementary lighting opening edge or the supplementary lighting closing edge are excluded; when there is halo diffusion abnormality or halo stripe abnormality in the halo area, and the core light color area meets the chromaticity fidelity requirement, the line exposure time window corresponding to the halo area is used as the soft constraint avoidance window; without affecting the imaging quality of the license plate area and road marking area, candidate supplementary lighting parameters that reduce the white light supplementary lighting ratio, limit the supplementary lighting peak level, or enable infrared auxiliary supplementary lighting are selected; the target supplementary lighting parameter is determined from the remaining candidate supplementary lighting parameters.

[0014] As a preferred embodiment of the image analysis-based supplementary lighting control method of the present invention, the steps of performing supplementary lighting capture according to the target supplementary lighting parameters, verifying the stripe coverage and evidence quality of the captured image, saving usable supplementary lighting parameters, or providing feedback on interference markers and failure reasons for the next cycle update are as follows: performing supplementary lighting capture according to the target supplementary lighting parameters to obtain the captured image; re-performing the captured image with key evidence region identification, separation of core light color area and halo area, stripe coverage status analysis, and evidence quality evaluation; when the stripe coverage in the core light color area, license plate area, and road marking area is eliminated or weakened, and the chromaticity of the core light color area is reduced... When the fidelity result, license plate edge clarity, license plate overexposure ratio, and road marking contrast meet the capture requirements, the target supplementary lighting parameters are saved as usable supplementary lighting parameters. When the core light color area returns to normal but the halo area still has diffusion, the halo residue anomaly is recorded, and the supplementary lighting mode is adjusted in the next cycle. When the stripe position does not change with the supplementary lighting trigger delay and is related to the background strong light source area, the road surface strong reflection area, or the traffic light's own refresh status, an external light source interference mark or a traffic light's own flicker interference mark is fed back. When the verification fails for a preset number of consecutive times, the failure reason record is saved, and the corresponding candidate supplementary lighting parameters are excluded in the next cycle.

[0015] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the image analysis-based supplementary lighting control method described in the first aspect of the present invention.

[0016] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any step of the image analysis-based supplementary lighting control method described in the first aspect of the present invention.

[0017] The beneficial effects of this invention are as follows: by determining the row exposure time window of the key evidence area, the matching of supplementary lighting control and image row exposure sequence is achieved; by segmenting the core light color area and halo area of ​​the traffic light, hierarchical constraints on the key interpretation area and the outer diffusion area of ​​the traffic light are achieved; by identifying horizontal bright and dark stripes and colored bands, the distinction between supplementary lighting-related interference and external light source interference is achieved; by generating evidence quality results that include light color fidelity, license plate clarity, and road marking contrast, the correlation control between supplementary lighting parameters and evidence image quality is achieved; by hard avoidance of the core light color area and soft constraint of the halo area, differentiated adjustment of supplementary lighting parameters is achieved; by reviewing the captured image and feeding back available supplementary lighting parameters and interference markers, closed-loop update of supplementary lighting control parameters is achieved. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of an image analysis-based supplemental lighting control method.

[0020] Figure 2 A flowchart for region identification and row exposure time window determination.

[0021] Figure 3 A flowchart for generating stripe recognition and evidence quality results.

[0022] Figure 4 A flowchart for determining and verifying supplemental lighting parameters for the target.

[0023] Figure 5 A comparison curve of the stripe overlap rate in the core light color area.

[0024] Figure 6 This is a scatter plot comparing the overexposure ratio of license plates with the edge sharpness of license plates. Detailed Implementation

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0027] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0028] Reference Figures 1-6 This is one embodiment of the present invention, which provides a supplementary lighting control method based on image analysis, including the following steps:

[0029] S1: Acquire continuous image frames of the traffic light capture scene, obtain rolling shutter line exposure mapping parameters and fill light parameters, identify the traffic light area, license plate area, road marking area and interference area, divide the traffic light area into core light color area and halo area, and determine the line exposure time window for each area.

[0030] After the traffic light capture device enters the capture state, the camera captures continuous image frames of the current intersection and records the capture timestamp, image height, image width and frame rate of each image.

[0031] Furthermore, the most recent 3 to 8 frames of images are preferably collected as continuous image frames. These continuous image frames are used to determine the directional drift of the horizontal bright and dark stripes in the continuous frames. The number of continuous image frames is determined by the balance between the real-time performance of traffic capture and the stability of stripe drift determination. Too few frames are not conducive to stripe drift identification, while too many frames will increase processing delay.

[0032] Furthermore, the current operating parameters of the camera and the lighting device are obtained, including the rolling shutter line readout time, single-line exposure duration, current lighting trigger delay, current lighting pulse width, current lighting peak level, and current lighting mode.

[0033] Furthermore, if the camera can output the scroll shutter line readout time, the scroll shutter line readout time parameter is read directly; if the camera cannot directly output the scroll shutter line readout time, the fill light is controlled to emit a low-intensity test pulse during the installation and debugging phase, and the correspondence between the image line number and the exposure time is calibrated based on the horizontal bright band line number formed by the test pulse in the image.

[0034] It should be noted that the intensity of the low-intensity test pulse is derived from the calibration results of the rated output power of the supplementary light and the visible bright band in the preview image, preferably 5% to 15% of the rated output power. The test pulse is not used for formal evidence images of illegality.

[0035] Among them, the image number The start time of the exposure of a row is expressed as: ;

[0036] No. The end time of the exposure of a line is represented as: ;

[0037] in, Indicates the first The start time of the line's exposure. Indicates the first The end time of the exposure of the line, This indicates the start exposure time of the current image frame. Indicates the image row number. This indicates the scrolling read time interval between two adjacent rows. Indicates the duration of a single-line exposure.

[0038] Furthermore, target detection and region segmentation are performed on continuous images to obtain a set of key evidence regions and a set of interference regions.

[0039] The key evidence area set includes traffic light area, license plate area, vehicle area and road marking area; the road marking area includes at least the stop line area; the interference area set includes road surface strong reflection area and background strong light source area.

[0040] For any key evidence area Record its upper boundary row number and lower boundary line number And obtain the row exposure time window corresponding to the key evidence area, represented as: ;

[0041] in, Indicates the first A key area of ​​evidence, Indicates the upper boundary row number of the key evidence area. Indicates the lower boundary row number of the key evidence area. This indicates the exposure time window for the key evidence area.

[0042] Furthermore, after identifying the traffic light area, the traffic light area is divided into two levels to obtain the core light color area and the halo area.

[0043] Specifically, let the traffic light area be... The system is The brightness, chromaticity, and saturation features are extracted to identify the area of ​​the lamp panel that is currently emitting light.

[0044] For any pixel within the traffic light area The brightness value of a pixel is represented as: ;

[0045] in, Represents pixels brightness value, , , These represent the red, green, and blue channel values ​​of a pixel, respectively.

[0046] Furthermore, the system in the traffic light area Within the signal light area, pixels with brightness values ​​higher than the 80th to 90th percentile of the signal light area are selected, and candidate connected regions of the current light panel are extracted by combining the features of the red, yellow, and green primary color channels.

[0047] It should be noted that the brightness quantile range is derived from the quantile statistics of pixel brightness distribution within the current traffic light area. This is used to exclude dark background areas and ordinary lamp housing areas, while retaining stable light-emitting pixels.

[0048] Furthermore, for candidate connected regions, regions that meet preset conditions in terms of area, roundness, and chromaticity stability are selected as the core light color region, denoted as... .

[0049] in, This indicates the core color area of ​​the current light panel of the traffic light. The core color area is used to characterize the light-emitting center area that is actually involved in judging the red, yellow, or green light status.

[0050] Furthermore, the area threshold of the candidate connected region is derived from the statistical results of the pixel area of ​​the same signal light panel in the normally illuminated image during the installation and calibration phase, preferably 20% to 90% of the area of ​​the calibration panel; the circularity threshold of the candidate connected region is derived from the morphological statistical results of the outline of the circular signal light panel, and the circularity can be determined by... The calculation is preferably taken as 0.45 to 0.85; for arrow lights, countdown lights or irregularly shaped lights, the corresponding light type template matching threshold can be used instead of the roundness threshold; the color stability threshold of the candidate connected region is derived from the statistical results of the fluctuation range of the main color channel in the normal red light, yellow light and green light samples, and the coefficient of variation of the main color channel is preferably controlled between 0.05 and 0.20.

[0051] After the core light color zone is determined, for Perform morphological dilation to obtain the extended region. .

[0052] It should be noted that the expansion radius is derived from the pixel width of the traffic light panel in the image, preferably 3 to 12 pixels, or 2% to 8% of the width of the traffic light panel.

[0053] Furthermore, the portion of the extended region after removing the core light color area is designated as the halo area, represented as follows: ;

[0054] in, This indicates the area indicated by the halo around the traffic light.

[0055] The halo region is used to characterize the outer edge area of ​​the signal light formed by lens scattering, rain and fog, supplementary light reflection, or overexposure diffusion.

[0056] Furthermore, if there are arrow lights, countdown lights, or multiple light panels entering the image simultaneously within the signal light area, then two-level segmentation is performed on each luminous connected region to obtain the corresponding core light color area and halo area.

[0057] For the core light color area and the halo area, determine their corresponding row exposure time windows respectively: ; ;

[0058] in, This indicates the row exposure time window corresponding to the core light color area. This indicates the row exposure time window corresponding to the halo area. and These represent the upper and lower boundary row numbers of the core light color area, respectively. and These represent the upper and lower boundary row numbers of the halo region, respectively.

[0059] Furthermore, the system reads the adjustable boundary data of the supplementary lighting device, which includes the supplementary lighting trigger delay range, the supplementary lighting pulse width range, the supplementary lighting peak level range, the minimum delay adjustment step size, and the available supplementary lighting modes of white light, infrared, and white light plus infrared.

[0060] It should be noted that the minimum delay adjustment step size is derived from the minimum time resolution of the fill light controller and the camera's rolling shutter line readout time, and the larger of the two executable values ​​is preferred.

[0061] S2: Extract brightness and chromaticity features along the image row direction, identify horizontal bright and dark stripes and color bands, and generate evidence quality results based on core light color fidelity, core stripe overlap, halo diffusion, license plate clarity and marking contrast.

[0062] Based on the acquired continuous image frames, key evidence region set, interference region set, core light color region and halo region, the system extracts the average brightness, average red channel, average green channel, average blue channel, saturation pixel ratio and local contrast of each row along the image row direction, forming row-direction brightness curve, row-direction chroma curve, row-direction saturation curve and row-direction contrast curve.

[0063] Let the first The average brightness of the row is Then calculate the first The brightness abrupt change of the row is expressed as: ;

[0064] in, Indicates the first The amount of brightness change in the row, Indicates the first The median brightness value within the row neighborhood. The neighborhood can be the 6th row. There are 3 to 10 lines above and below each other.

[0065] It should be noted that the neighborhood median calculation window is derived from the matching relationship between the license plate character height, road marking width, and image line resolution, and is used to reduce the interference of license plate characters, lane edges, vehicle outlines, and local reflections on stripe judgment.

[0066] Furthermore, when a continuous line segment simultaneously meets the following conditions: the brightness change amount is greater than the preset brightness change threshold, it crosses at least two regions in the horizontal direction, namely the vehicle area, road marking area, road surface area, or background area, and there is line drift or intensity periodic change in continuous image frames, and it does not belong to license plate characters, parking lines, vehicle body edges, or fixed road markings, the continuous line segment is determined as a candidate stripe region.

[0067] Furthermore, the preset brightness abrupt change threshold is derived from the statistical upper limit of the row-direction brightness abrupt change in a stripe-free qualified image, preferably taken from a stripe-free qualified image. The 95th to 99th percentile values ​​are used; in an 8-bit grayscale image, 8 to 35 gray levels are preferred. The row drift threshold is derived from the row number fluctuation range of normal static markings and vehicle edges in consecutive frames, preferably 2 to 8 rows; when the row number change of a candidate stripe in consecutive frames exceeds the row number fluctuation range, it is determined that there is stripe row drift. The stripe intensity threshold is derived from the boundary statistical value of the abrupt change in row brightness between qualified and stripe-contaminated images, preferably 15 to 50 gray levels in an 8-bit grayscale image. The stripe width is derived from the actual number of consecutive rows in the candidate stripe region along the image row direction; preferably, 2 to 40 consecutive abnormal row segments are used as the stripe width range that can participate in the avoidance window expansion.

[0068] Furthermore, for the traffic light area, the system performs image quality analysis on the core light color area and the halo area respectively.

[0069] First, calculate the average values ​​of the red, green, and blue channels within the core light color zone.

[0070] When the traffic light is red, the core light color separation is expressed as: ;

[0071] When the traffic light is green, the core light color separation is expressed as: ;

[0072] When the traffic light is yellow, the core light color separation is expressed as: ;

[0073] in, Indicates the core color separation of the red light. Indicates the core color separation of the green light. Indicates the core color separation of the yellow light. This represents the average value of the red channel of all pixels within the core light color area. This represents the average green channel value of all pixels within the core light color area. This represents the average value of the blue channel for all pixels within the core light color area.

[0074] It should be noted that the core color separation threshold for red lights is derived from the statistical lower limit of the separation degree of the red channel relative to the green and blue channels in normal red light samples; the core color separation threshold for green lights is derived from the statistical lower limit of the separation degree of the green channel relative to the red and blue channels in normal green light samples; and the core color separation threshold for yellow lights is derived from the statistical lower limit of the joint separation degree of the red and green channels relative to the blue channel in normal yellow light samples. The preferred core color separation thresholds are between 0.18 and 0.35.

[0075] when , or When the color is below the corresponding core light color separation threshold, it is determined that there is a color fidelity anomaly in the core light color area.

[0076] Furthermore, let the identified striped areas be... The overlap rate of stripes in the core light color area is calculated and expressed as: ;

[0077] in, Indicates the overlap rate of stripes in the core light color area. This indicates the number of pixels in the overlapping area between the core light color area and the stripe area. This indicates the number of pixels in the core light color area.

[0078] It should be noted that the preset core overlap threshold is derived from the correspondence between the stripe coverage ratio in manually labeled core light color area stripe contamination samples and the risk of traffic light misjudgment, preferably ranging from 0.05 to 0.15. When the value exceeds the preset core overlap threshold, it is determined that the stripe has crossed the core light color area.

[0079] Furthermore, for the halo region, the system calculates the halo diffusion ratio.

[0080] Let the number of pixels within the halo region that satisfy the same type of light color condition be . The number of pixels in the core light color area is The halo diffusion ratio is expressed as: ;

[0081] in, Indicates the halo diffusion ratio. This indicates the number of pixels within the halo area that share the same primary color as the current traffic light. This indicates the number of pixels in the core light color area.

[0082] It should be noted that the preset halo diffusion threshold is derived from the statistical cutoff value of the halo diffusion ratio in samples under normal weather, rain / fog, and excessive supplemental lighting, and is preferably set between 0.30 and 1.20. If the halo diffusion value exceeds the preset threshold, the signal light is deemed to have an abnormal halo diffusion.

[0083] Furthermore, the overlap rate of the fringes in the halo region is calculated: ;

[0084] in, Indicates the overlap rate of fringes in the halo region. This indicates the number of pixels in the overlapping area between the halo area and the stripe area. This indicates the number of pixels in the halo area.

[0085] It should be noted that the preset halo overlap threshold is derived from the statistical results of the stripe coverage ratio in the halo area stripe samples and the unaffected state of the core light color area, preferably ranging from 0.15 to 0.35. The value is greater than the preset halo overlap threshold, and If the preset core overlap threshold is not exceeded, it is determined that the stripes mainly affect the outer halo area of ​​the signal light and have not directly damaged the core light color area.

[0086] Furthermore, based on the current fill light trigger delay, the current fill light pulse width, and the line exposure mapping parameters, the system determines whether the line range of the candidate stripe region matches the line range of the image corresponding to the current fill light's on or off edge. If the candidate stripe region corresponds to the current fill light edge, a fill light-related stripe marker is generated; if the candidate stripe region is mainly located in areas of strong road reflection or strong background light sources, and does not move accordingly with changes in the current fill light edge, an external light source interference marker is generated.

[0087] It should be noted that the threshold for judging external light source interference is derived from the statistical results of consecutive frames in which the stripe position does not change with the supplementary light trigger delay and coincides with the background strong light source or the road surface strong reflection area. Preferably, 2 to 3 consecutive frames are required to meet this correspondence.

[0088] Furthermore, in addition to stripe recognition, the system also evaluates the imaging quality of key evidence areas, obtaining the chromaticity fidelity results of traffic lights, the edge sharpness of license plates, the overexposure ratio of license plates, the contrast of road markings, and the degree of vehicle outline blurring. Among them, the chromaticity fidelity results of traffic lights are jointly determined by the chromaticity fidelity results of the core light color area, the stripe overlap rate of the core light color area, the halo diffusion ratio, and the stripe overlap rate of the halo area.

[0089] Furthermore, the preset license plate overexposure threshold is derived from the statistical boundary between the proportion of pixels in the license plate area close to the full-scale range of the sensor and the number of failed character recognition samples, preferably 3% to 12% of the number of pixels in the license plate area; the preset license plate edge sharpness threshold is derived from the statistical lower limit of the character edge gradient or connectivity integrity in the correctly identifiable license plate character samples, preferably 0.20 to 0.45 of the normalized edge sharpness; the preset road marking contrast threshold is derived from the statistical lower limit of the grayscale difference between the parking line or road marking and the adjacent road surface in the qualified captured image, preferably 12 to 35 grayscale levels in an 8-bit grayscale image, or 0.08 to 0.25 of the normalized contrast; the preset vehicle contour trailing threshold is derived from the statistical boundary between the vehicle edge diffusion width along the direction of motion and the failure of contour recognition in samples of different vehicle speeds, preferably 2 to 8 pixels.

[0090] Finally, the results of the stripe analysis and the quality of evidence were obtained.

[0091] The results of the stripe analysis include stripe row range, stripe width, stripe intensity, stripe color deviation direction, key evidence areas covered by the stripes, supplementary lighting-related stripe markers, and external light source interference markers.

[0092] The evidence quality results include the color fidelity of traffic lights, the color fidelity of the core light color area, the stripe overlap rate of the core light color area, the halo diffusion ratio, the stripe overlap rate of the halo area, the traffic light anomaly type marking, the license plate edge sharpness, the license plate overexposure ratio, the road marking contrast, and the degree of vehicle outline ghosting.

[0093] The abnormal signal light types are marked as follows: abnormal core light color, abnormal core stripe crossing, abnormal halo diffusion, abnormal halo stripe, and normal signal light area.

[0094] S3: Based on the stripe results, evidence quality results, and adjustable fill light boundaries, candidate fill light parameters are generated. The line exposure time windows of the core light color area, license plate area, and road marking area are used as the avoidance basis. Hard avoidance is applied to the core light color area, and soft constraint is applied to the halo area to determine the target fill light parameters.

[0095] Based on the exposure time windows of the key evidence area, the core light color area, the halo area, and the adjustable boundary data of the supplementary lighting, as well as the stripe analysis results and evidence quality results, supplementary lighting edge avoidance parameters are generated.

[0096] Furthermore, let the on-edge time of the fill light be: ;

[0097] The off-edge time of the fill light is: ;

[0098] in, Indicates the edge time of the fill light activation. Indicates the time it takes for the fill light to turn off. Indicates the base time for triggering the snapshot. Indicates the delay of the supplementary light trigger. This indicates the width of the fill light pulse.

[0099] Furthermore, for any key area of ​​evidence When the fill light is turned on edge time Exposure time window of lines falling into key evidence area Or the timing of the fill light turning off. Exposure time window of lines falling into key evidence area At that time, it was determined that the edge of the supplementary light pulse entered the key evidence area, which posed a risk of forming a horizontal bright-dark boundary or a colored band.

[0100] Furthermore, based on the stripe width, the system expands the line exposure time window corresponding to the traffic light area, license plate area, and road marking area to form a fill light edge avoidance window.

[0101] It should be noted that the stripe width expansion parameter is derived from the actual width of the candidate stripe region in the image line direction. It is preferable to expand the avoidance window by 0.5 to 1.5 times the stripe width in the stripe movement direction to prevent the supplementary lighting edge from intruding into the boundary of the key evidence region due to the stripe width, even if it does not strictly fall into the key evidence region.

[0102] Furthermore, within the traffic light area, the system distinguishes between hard avoidance of the core light color area and soft constraint of the halo area.

[0103] When any of the following results (A1-A3) are output, the line exposure time window corresponding to the core light color area will be... Set to highest priority for fill light edge avoidance window:

[0104] A1: The color fidelity result of the core light color area is lower than the preset light color fidelity threshold.

[0105] A2: Stripe overlap rate in the core light color area It exceeds the preset core overlap threshold.

[0106] A3: The signal light abnormality type is marked as core light color abnormality or core stripe crossing.

[0107] Furthermore, the preset color fidelity threshold is derived from the lower limit of the qualified condition formed by the core color separation, core stripe overlap rate and halo diffusion ratio in the normal signal light sample. It is preferably determined by a combination of conditions where the core color separation is not lower than 0.18 to 0.35, the core stripe overlap rate is not higher than 0.05 to 0.15, and the halo diffusion ratio is not higher than 0.30 to 1.20.

[0108] Furthermore, in any of cases A1-A3, any candidate supplementary lighting parameter that meets one of the following conditions B1-B2 will be preferentially excluded: B1: .

[0109] B2: .

[0110] in, This indicates the row exposure time window corresponding to the core light color area.

[0111] Furthermore, when the halo diffusion ratio in the halo region is... The halo diffusion threshold is exceeded, but the overlap rate of stripes in the core light color area is high. If the preset core overlap threshold is not exceeded and the color fidelity result of the core light color area meets the requirements, the entire signal light area will not be treated as the highest priority avoidance target. Instead, the line exposure time window corresponding to the halo area will be used. As a soft constraint avoidance window.

[0112] Furthermore, for soft constraints in the halo area, the following strategies are preferred: moderately reduce the proportion of white light supplementary illumination; prioritize candidate supplementary illumination parameters that do not amplify halo diffusion; and ensure that the supplementary illumination edges avoid [unclear - possibly referring to specific areas or regions] without affecting the imaging quality of the license plate area and road marking area. If you avoid If the license plate area has insufficient stripes or insufficient brightness, priority should be given to ensuring the core light color area and the license plate area, and the halo area should not be forcibly avoided.

[0113] It should be noted that the white light ratio limitation threshold is derived from the calibration results of the white light supplement ratio change when the chromaticity fidelity of the core light color area decreases and the halo diffusion increases. Preferably, when the soft constraint is triggered, the white light supplement ratio is reduced by 10% to 40%, or the white light output is limited to 30% to 70% of the original white light output. The infrared auxiliary activation threshold is derived from the calibration results of infrared supplementation still maintaining license plate recognition when white light supplementation causes a decrease in the chromaticity fidelity of the core light or license plate overexposure. Preferably, infrared auxiliary supplementation is activated when the license plate overexposure ratio exceeds 3% to 12% and the core light color separation is lower than 0.18 to 0.35.

[0114] Subsequently, candidate supplementary lighting parameters are generated based on the adjustable boundary data of the supplementary lighting.

[0115] Among them, the candidate supplementary lighting parameters include candidate trigger delay, candidate pulse width, candidate peak level and candidate supplementary lighting mode, and all candidate supplementary lighting parameters are within the executable range of the supplementary lighting device.

[0116] Furthermore, candidate supplementary lighting parameters are screened according to the following rules: First, candidate supplementary lighting parameters whose open or closed edges enter the core light color area's supplementary lighting edge avoidance window are excluded; second, among the remaining candidate supplementary lighting parameters, candidate supplementary lighting parameters whose open or closed edges enter the license plate area's supplementary lighting edge avoidance window are excluded; third, among the remaining candidate supplementary lighting parameters, candidate supplementary lighting parameters that cause light and dark layering or whitening in the road marking area are excluded; then, the target supplementary lighting parameter is selected based on the evidence quality results.

[0117] Specifically, when the license plate edge clarity is insufficient and the license plate overexposure ratio does not exceed the preset overexposure threshold, candidate fill light parameters that can improve the effective brightness of the license plate area are selected first; when the license plate overexposure ratio exceeds the preset overexposure threshold, the fill light peak level is reduced to reduce brightness abrupt changes; when the chromaticity fidelity result of the core light color area is lower than the preset chromaticity fidelity threshold, the white light fill light ratio is limited or infrared auxiliary fill light is enabled; when only the halo area has diffusion abnormalities and the core light color area meets the chromaticity fidelity requirements, only the white light ratio or peak level is slightly reduced without making significant phase adjustments; when the vehicle outline shadow exceeds the preset shadow threshold, the fill light pulse width is shortened or the trigger delay is adjusted without allowing the fill light edge to enter the core light color area, license plate area, and road marking area.

[0118] Furthermore, when the output shows external light source interference marks and the stripes do not correspond to the current fill light edge, the phase adjustment amplitude is reduced to avoid repeated ineffective phase adjustment of the fill light due to external strobe light sources, vehicle light pulse width modulation, or road surface reflection.

[0119] Furthermore, when the stripe intensity exceeds the preset stripe intensity threshold and continuously covers the core light color area or license plate area, switch to low-intensity wide-pulse fill light mode, or reduce the white light ratio and enable infrared auxiliary fill light.

[0120] It should be noted that the low-intensity wide pulse switching threshold is derived from the statistical results of the number of consecutive verification failures and the failure of short pulse phase adjustment when the stripe intensity exceeds the threshold. It is preferably triggered when verification fails for 2 to 3 consecutive capture cycles, or when the stripe intensity exceeds the preset stripe intensity threshold and continuously covers the core light color area or license plate area.

[0121] Among them, the fill light edge avoidance parameters include target trigger delay, target pulse width, target peak level, target fill light mode, fill light edge avoidance window, and the reason for this adjustment.

[0122] The reasons for this adjustment include abnormal core light color, core stripe crossing, abnormal halo diffusion, abnormal halo stripe, license plate overexposure, insufficient license plate clarity, insufficient road marking contrast, external light source interference, and signal light flicker interference.

[0123] S4: Perform supplementary lighting capture according to the target supplementary lighting parameters, review the stripe coverage and evidence quality of the captured image, save the available supplementary lighting parameters, or provide feedback on interference markers and failure reasons for the next cycle update.

[0124] Perform supplementary lighting capture according to the target trigger delay, target pulse width, target peak level, and target supplementary lighting mode to acquire the captured image.

[0125] Furthermore, after the capture is completed, the key evidence area identification, the separation of the core light color area and halo area of ​​the traffic light, the stripe coverage status analysis, and the evidence quality evaluation are re-performed on the captured image to obtain the stripe analysis results and the evidence quality results after review.

[0126] Furthermore, in the signal light area verification, the system recalculates the chromaticity fidelity of the core light color area, the stripe overlap rate of the core light color area, the halo diffusion ratio, and the stripe overlap rate of the halo area.

[0127] When the verification results meet the following conditions, the supplementary lighting control of the traffic light area is deemed to be effective: the overlap rate of the stripes in the core light color area is lower than the preset core overlap threshold; the chromaticity fidelity result of the core light color area meets the preset chromaticity fidelity requirements; the halo diffusion ratio does not continue to increase; and the traffic light abnormality type marker changes from core light color abnormality or core stripe crossing to normal traffic light area or only slight halo abnormality.

[0128] Furthermore, the reviewed stripe analysis results and the reviewed evidence quality results were compared with the supplementary lighting edge avoidance window and the reason mark for this adjustment.

[0129] When the stripe coverage in the core light color area, license plate area, and road marking area is eliminated or reduced, and the color fidelity result of the core light color area, the edge sharpness of the license plate, the overexposure ratio of the license plate, and the contrast of the road marking meet the capture requirements, the current target trigger delay, target pulse width, target peak level, and target fill light mode are saved as available fill light parameters for the current camera, current lane, and current environment.

[0130] Furthermore, when the verification results show that the core light color area has returned to normal, but the halo area still has diffusion, it is not directly judged as a failure of supplementary light control, but recorded as halo residue abnormality; in the next capture cycle, only the white light ratio is slightly limited or the supplementary light peak level is adjusted, without making large phase adjustments.

[0131] Furthermore, when the verification results show that there are still stripes crossing the core light color area, and the stripe position moves accordingly with the change of the supplementary light trigger delay, it is determined that the stripes are still related to the supplementary light edge, and the supplementary light trigger delay is finely adjusted based on the supplementary light edge avoidance window of the core light color area in the next capture cycle.

[0132] It should be noted that the fine-tuning step size is determined based on the stripe intensity, preferably 20 microseconds to 100 microseconds, and can be adjusted with a delay accuracy that can be stably responded to by the common line readout time of traffic capture cameras and the supplementary light controller.

[0133] Furthermore, when the verification results show that the core light color area is abnormal and does not change with the supplementary light trigger delay, and is mainly related to the background strong light source area, the road surface strong reflection area, or the signal light's own refresh status, the external light source interference mark or the signal light's own flicker interference mark is updated, and the corresponding area is used as the interference prior area in the next capture cycle to reduce stripe misjudgment and invalid supplementary light phase adjustment.

[0134] It should be noted that the threshold for judging the flicker interference of the traffic light itself comes from the statistical results of consecutive frames where the abnormality of the core color area of ​​the traffic light is unrelated to the phase of the supplementary light, but is related to the light emission cycle or refresh status of the traffic light. Preferably, it is required that the abnormal position of the core color area and the edge of the supplementary light have no corresponding movement relationship for 2 to 3 consecutive frames.

[0135] Furthermore, when the verification fails for 2 to 3 consecutive capture cycles, the short pulse phase fine-tuning is stopped, and the mode is switched to low-intensity wide pulse supplementary light mode, or the white light ratio is reduced and infrared auxiliary supplementary light is enabled, while the failure reason is saved.

[0136] It should be noted that the threshold for failure to review after 2 to 3 consecutive capture cycles is derived from the balance between the real-time requirements of traffic capture and the impact of the number of consecutive trial and error attempts on the success rate of capturing evidence of violations.

[0137] The final output includes supplementary lighting parameters, verification results, external light source interference markers, signal light self-flicker interference markers, and failure reason records.

[0138] Among them, the available supplementary lighting parameters can be fed back to the next cycle as the current supplementary lighting parameters; the external light source interference mark and the signal light's own flicker interference mark can be fed back to the next cycle as interference priors; the failure reason record can be fed back to the next cycle for the elimination of candidate supplementary lighting parameters.

[0139] In this embodiment, as Figure 5 As shown, during the continuous capture cycle, the overlap rate of the stripes in the core light color area under conventional supplementary lighting control is generally at a high level, and it increases significantly during the stages of enhanced reflection from rain and fog and interference from external light sources. In some cycles, it approaches or exceeds the preset core overlap threshold, indicating that the edge of the supplementary lighting is prone to overlap with the line exposure time window of the core light color area of ​​the signal light, thereby forming horizontal bright and dark stripes or colored bands, increasing the risk of misjudging the status of red, yellow or green lights.

[0140] After adopting this technical solution, the system determines the line exposure time window of the core light color area, license plate area and road marking area according to the rolling shutter line exposure mapping parameters, and filters the target fill light parameters by hard avoidance of the core light color area and soft constraint of the halo area, so that the fill light opening edge and closing edge avoid the key evidence area.

[0141] As can be seen from the figure, the curve corresponding to this technical solution remains consistently below the preset core overlap threshold over a long period of time, and maintains relatively small fluctuations even during the interference enhancement phase. This indicates that it can effectively reduce the probability of stripes crossing the core light color area, improve the color fidelity of the traffic lights, and enhance the stability of captured evidence.

[0142] In this embodiment, as Figure 6 As shown, the conventional supplementary lighting control samples are mainly distributed in areas with high license plate overexposure ratio and low license plate edge clarity. Some samples are close to or exceed the preset license plate overexposure threshold, while being lower than the preset license plate edge clarity threshold. This indicates that conventional supplementary lighting methods are prone to causing license plate area saturation, character edge gradient reduction, and unstable recognition quality when increasing brightness.

[0143] After adopting this technical solution, the sample points are generally concentrated in areas with low overexposure ratio and high definition. The overexposure ratio of license plates is significantly reduced, the edge definition of license plates is significantly improved, and most samples remain within the overexposure threshold and above the definition threshold.

[0144] The results show that this technical solution can dynamically adjust the target trigger delay, fill light pulse width, fill light peak level, and fill light mode according to the evidence quality results. When the license plate overexposure ratio is too high, the fill light peak level is reduced. When the license plate clarity is insufficient but not overexposed, the effective brightness is increased, thereby balancing the license plate not being overexposed and the character edges being clear, thus improving the identifiability and evidence usability of the illegal capture images.

[0145] This embodiment also provides a computer device applicable to the image analysis-based supplementary lighting control method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the image analysis-based supplementary lighting control method proposed in the above embodiment.

[0146] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0147] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the image analysis-based supplementary lighting control method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0148] In summary, this invention achieves matching between supplementary lighting control and image line exposure timing by determining the line exposure time window of the key evidence area; it achieves hierarchical constraints between the key interpretation area and the outer diffusion area of ​​the traffic light by segmenting the core light color area and the halo area; it achieves differentiation between supplementary lighting-related interference and external light source interference by identifying horizontal bright and dark stripes and colored bands; it achieves correlation control between supplementary lighting parameters and evidence image quality by generating evidence quality results that include light color fidelity, license plate clarity, and road marking contrast; it achieves differentiated adjustment of supplementary lighting parameters by hard avoidance of the core light color area and soft constraint of the halo area; and it achieves closed-loop update of supplementary lighting control parameters by reviewing the captured image and providing feedback on available supplementary lighting parameters and interference markers.

[0149] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An image analysis-based light supplement control method, characterized by, include: Collect continuous image frames of the traffic light capture scene, obtain rolling shutter line exposure mapping parameters and fill light parameters, identify the traffic light area, license plate area, road marking area and interference area, divide the traffic light area into core light color area and halo area, and determine the line exposure time window for each area; Brightness and chromaticity features are extracted along the image row direction to identify horizontal bright and dark stripes and color bands, and evidence quality results are generated based on core light color fidelity, core stripe overlap, halo diffusion, license plate clarity and marking contrast. Candidate supplementary lighting parameters are generated based on the stripe results, evidence quality results, and adjustable supplementary lighting boundaries. The line exposure time windows of the core light color area, license plate area, and road marking area are used as the basis for avoidance. Hard avoidance is applied to the core light color area, and soft constraint is applied to the halo area to determine the target supplementary lighting parameters. Perform supplementary lighting capture according to the target supplementary lighting parameters, review the stripe coverage and evidence quality of the captured image, save the available supplementary lighting parameters, or provide feedback on interference markers and failure reasons for the next cycle update.

2. The image analysis based fill light control method of claim 1, wherein, The process of obtaining the rolling shutter line exposure mapping parameters and fill light parameters includes: Acquire continuous image frames of the current traffic light capture scene, and record the acquisition timestamp, image height, image width, and frame rate of each image frame; Obtain the camera's rolling shutter line readout time and single-line exposure duration, and obtain the current fill light trigger delay, current fill light pulse width, current fill light peak level, and current fill light mode; Establish the correspondence between image line number and line exposure time window based on image line number, rolling shutter line readout time, and single line exposure duration; Read the adjustable boundary data of the supplementary lighting device, which includes the supplementary lighting trigger delay range, the supplementary lighting pulse width range, the supplementary lighting peak level range, the minimum delay adjustment step size, and the available supplementary lighting modes.

3. The image analysis based fill light control method of claim 2, wherein, The process of identifying the traffic light area, license plate area, road marking area, and interference area, and dividing the traffic light area into a core light color area and a halo area, includes: Target detection and region segmentation are performed on consecutive image frames to obtain traffic light regions, license plate regions, vehicle regions, road marking regions, strong road surface reflection regions, and strong background light source regions; Extract brightness, chromaticity, and saturation features within the signal light area to determine the area of ​​the light panel currently in an illuminating state; Pixels with brightness values ​​higher than the fixed percentile of the brightness of the signal light area are selected, and candidate connected regions of the light-emitting disk are extracted by combining the main color channel features. Candidate connected regions that meet preset conditions in terms of area, roundness, and chromaticity stability are identified as core light color areas; Morphological expansion is performed on the core light color area to obtain the extended area, and the part of the extended area after removing the core light color area is determined as the halo area. Determine the corresponding line exposure time windows for the core light color area and the halo area respectively.

4. The image analysis based fill light control method of claim 1, wherein, The specific steps for extracting brightness and chromaticity features along the image row direction and identifying horizontal bright and dark stripes and color bands are as follows: Extract the mean brightness, mean red channel value, mean green channel value, mean blue channel value, saturation pixel ratio, and local contrast of each row along the image row direction; The brightness abrupt change is calculated based on the average brightness of each row and the median brightness of its neighboring area, forming a row-to-row brightness curve; A row-to-row chromaticity curve is formed based on the mean values ​​of the red, green, and blue channels; When a continuous line segment meets the following conditions: the brightness change exceeds the preset brightness change threshold, it spans at least two regions in the horizontal direction, there is line drift or intensity periodic change in continuous image frames, and it does not belong to license plate characters, parking lines, vehicle body edges or fixed road markings, the continuous line segment is identified as a candidate stripe region. The colored bands are determined based on the chromaticity changes in the candidate stripe regions.

5. The image analysis based fill light control method of claim 4, wherein, The specific steps for generating evidence quality results based on core light color fidelity, core stripe overlap, halo diffusion, license plate clarity, and road marking contrast are as follows: Calculate the average values ​​of the red, green, and blue channels within the core light color zone, and calculate the core light color separation degree based on the current traffic light color; When the core light color separation is lower than the corresponding core light color separation threshold, a core light color abnormality marker is generated; Calculate the overlap ratio between the core light color area and the candidate stripe area. When the overlap ratio exceeds the preset core overlap threshold, generate a core stripe crossing mark. Calculate the ratio of the number of pixels in the halo area that are the same as the main color of the current traffic light to the number of pixels in the core light color area, and obtain the halo diffusion ratio; Calculate the overlap ratio between the halo region and the candidate stripe region to generate the halo stripe result; The evidence quality results are generated by combining the sharpness of the license plate edge, the overexposure ratio of the license plate, the contrast of road markings, and the degree of vehicle outline blur.

6. The image analysis based fill light control method of claim 1, wherein, The process of generating candidate supplementary lighting parameters based on the stripe results, evidence quality results, and adjustable supplementary lighting boundaries includes: Candidate trigger delay, candidate pulse width, candidate peak level, and candidate fill light mode are generated based on the adjustable fill light boundary data. Determine the fill light on edge time and fill light off edge time based on the candidate fill light parameters; Based on the stripe width, the corresponding line exposure time windows of the core light color area, license plate area and road marking area are expanded to form a fill light edge avoidance window; When the timing of the fill light on or off falls within the fill light edge avoidance window, it is determined that the corresponding candidate fill light parameter has a stripe risk. Candidate supplementary lighting parameters were selected based on stripe risk and evidence quality results.

7. The image analysis-based supplementary lighting control method as described in claim 1 or 6, characterized in that, The process of hard avoidance of the core light color area and soft constraint of the halo area to determine the target supplementary lighting parameters includes: When the color fidelity result of the core light color area is lower than the preset color fidelity threshold, the stripe overlap rate of the core light color area exceeds the preset core overlap threshold, or the signal light abnormality type is marked as core light color abnormality or core stripe crossing, the line exposure time window corresponding to the core light color area is used as the highest priority fill light edge avoidance window. Exclude candidate fill light parameters from the exposure time window corresponding to the core light color area when the fill light is turned on or off. When there is abnormal halo diffusion or abnormal halo stripes in the halo area, and the core light color area meets the color fidelity requirements, the line exposure time window corresponding to the halo area is used as a soft constraint avoidance window. Without affecting the imaging quality of the license plate area and road marking area, the candidate supplementary lighting parameters are to reduce the proportion of white light supplementary lighting, limit the peak level of supplementary lighting, or enable infrared auxiliary supplementary lighting. The target lighting parameter is determined from the remaining candidate lighting parameters.

8. The image analysis-based supplementary lighting control method as described in claim 1, characterized in that, The steps for performing supplementary lighting capture according to the target supplementary lighting parameters, verifying the stripe coverage and evidence quality of the captured image, saving usable supplementary lighting parameters, or providing feedback on interference markers and failure reasons for the next cycle update are as follows: Perform supplementary lighting capture according to the target supplementary lighting parameters to obtain the captured image; The key evidence areas, core light color areas and halo areas were re-identified, stripe coverage status was analyzed, and evidence quality was evaluated in the captured images. When the stripe coverage in the core light color area, license plate area and road marking area is eliminated or reduced, and the color fidelity result of the core light color area, the edge clarity of the license plate, the overexposure ratio of the license plate and the contrast of the road marking meet the capture requirements, the target supplementary lighting parameters are saved as usable supplementary lighting parameters. When the core light color area returns to normal but the halo area still has diffusion, record the halo residue abnormality and adjust the supplementary lighting mode in the next cycle. When the stripe position does not change with the supplementary light trigger delay and is related to the background strong light source area, the road surface strong reflection area or the signal light's own refresh status, feedback external light source interference mark or signal light's own flicker interference mark is provided. If the verification fails for a preset number of consecutive times, the reason for the failure is saved and the corresponding candidate supplementary lighting parameter is excluded in the next cycle. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. When the processor executes the computer program, it implements the steps of the image analysis-based supplementary lighting control method according to any one of claims 1 to 8.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the image analysis-based supplementary lighting control method according to any one of claims 1 to 8.

Images