Traffic light for image quality test and test method thereof

Abstract

The invention discloses a traffic light for an image quality test and a test method thereof, and the traffic light comprises an LED array which is composed of four rows and N columns of LED lamp beads, and the four rows of LEDs respectively adopt red, yellow, green and white LED lamp beads; and the controller is used for outputting working parameters to the LED lamp beads, the LED lamp beads in the same column adopt completely same working parameters, and the LED lamp beads in the same row adopt incompletely same working parameters. Wherein N is an integer greater than 1. The traffic light provided by the scheme of the invention meets the detection performance test requirements of the traffic light detection system under different conditions, and guarantees the test quality of the traffic light image quality test.

Description

Technical Field

[0001] The present invention relates to the field of image quality detection, and in particular to a traffic light for image quality testing and a testing method thereof. Background Art

[0002] With the increasing demand for automotive cameras in driver-assisted and autonomous driving systems, various image-based object detection systems have been proposed, such as building detection, tree detection, and traffic light detection. Traffic light recognition is a crucial component of intelligent traffic machine vision. It provides drivers with real-time intersection status, road feasibility information, and accident prevention. For autonomous driving systems, traffic light detection is a key technology for ensuring their safety.

[0003] Current traffic light detection systems typically use a color feature-based recognition method. Existing test devices for traffic light detection systems typically only simulate the structural composition of traffic lights, with a single test scenario, which cannot meet the detection performance testing requirements of traffic light detection systems under different conditions. Summary of the Invention

[0004] In response to the above-mentioned defects, the present invention provides a traffic light for image quality testing and a testing method thereof, and provides a traffic light that meets the detection performance testing requirements of the traffic light detection system under different situations, thereby ensuring the test quality of the traffic light image quality test.

[0005] An embodiment of the present invention provides a traffic light for image quality testing, comprising:

[0006] An LED array consisting of 4 rows and N columns of LED lamp beads, with the 4 rows of LEDs using red, yellow, green and white LED lamp beads respectively;

[0007] The controller is used to output operating parameters to each LED lamp bead. The LED lamp beads in the same column use exactly the same operating parameters, and the LED lamp beads in the same row use different operating parameters.

[0008] Wherein, N is an integer greater than 1.

[0009] Preferably, the operating parameters include brightness values, and the LED lamp beads in the same column all use exactly the same brightness value, while the LED lamp beads in the same row use different brightness values.

[0010] Preferably, the operating parameters include flicker parameters, and the LED lamp beads in the same column use exactly the same flicker parameters, while the LED lamp beads in the same row use different flicker parameters.

[0011] The flicker parameters include a flicker frequency and a flicker duty cycle.

[0012] An embodiment of the present invention further provides a method for testing a traffic light for image quality testing, the method being applied to the traffic light for image quality testing described in any of the above embodiments, the method comprising:

[0013] Inputting test parameters of the traffic light for image quality testing, wherein the test parameters include operating parameters of each LED lamp bead;

[0014] Using the camera to be tested to obtain a test image of the traffic light for image quality testing;

[0015] Calculating image indicators of the test image;

[0016] Output the image test result of the camera to be tested according to the image index.

[0017] Preferably, outputting the image test result of the camera to be tested according to the image index includes:

[0018] Compare the image indicators of each LED lamp bead in each row, and output the image test results of the camera to be tested on lamp beads of different colors.

[0019] As a preferred solution, when the test parameters include that each LED lamp bead in the same column adopts the same brightness value, and each LED lamp bead in the same row adopts a different brightness value;

[0020] The image indicators include the hue and color distance of each LED lamp bead.

[0021] As a preferred solution, the hue of each LED lamp bead is

[0022] The color distance of each LED lamp bead is

[0023] Among them, mea_v is the v value of the UV value of the LED lamp bead in the YUV space, mea_u is the u value of the UV value of the LED lamp bead in the YUV space, white_v is the v value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space, and white_u is the u value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space.

[0024] As a preferred solution, outputting the image test result of the camera to be tested according to the image index includes:

[0025] When the hue difference between any two LED lamp beads in the i-th row is within the preset hue difference range, and the color distance difference between any two LED lamp beads in the i-th row is within the preset color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row meets the classification standard;

[0026] When the hue difference between two LED lamp beads in the i-th row is not within the hue difference range, or the color distance difference between two LED lamp beads in the i-th row is not within the color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row does not meet the classification standard;

[0027] Among them, i=1, 2, 3, 4.

[0028] As a preferred solution, when the test parameters include different brightness values ​​for each white LED lamp bead and turning off the LED lamp beads of other colors: or,

[0029] When different brightness values ​​are used for each white LED lamp bead, and one color LED lamp bead is selected to use the same brightness value as the white LED lamp bead in the same column, and the other two colors of LED lamp beads are turned off; or,

[0030] When different brightness values ​​are used for each white LED lamp bead, and the same brightness value as the white LED lamp bead in the same column is used for any other color LED lamp bead;

[0031] The image index includes the saturation of each white LED lamp bead.

[0032] As a preferred solution, the saturation of each white LED lamp bead is s=(1-(3*min(R,G,B)) / (R+G+B))*100;

[0033] Among them, R, G, and B represent the grayscale values ​​of red, green, and blue of the pixels of the white LED lamp bead image in the RGB color space, respectively.

[0034] As a preferred solution, outputting the image test result of the camera to be tested according to the image index includes:

[0035] When the saturation difference between any two white LED lamp beads is within the preset saturation difference range, it is determined that the white balance characteristics of the camera to be tested meet the classification standard;

[0036] When the saturation difference between two white LED lamp beads is not within the saturation difference range, it is determined that the white balance characteristic of the camera to be tested does not meet the classification standard.

[0037] As a preferred solution, the 4 rows and N columns of LED lamp beads are divided into k groups of lamp beads with m columns in each group;

[0038] When the test parameters include that each group of lamp beads adopts the same brightness value, and different columns of lamps in each group adopt the same flicker frequency and different flicker duty cycles, the image indicators include the flicker modulation index, flicker detection index, modulation relief probability, average modulation relief probability and flicker beat frequency of each LED lamp bead;

[0039] Wherein, m and k are integers greater than or equal to 1, and m×k≤N.

[0040] As a preferred solution, the flicker modulation index of each LED lamp bead is

[0041] The flicker detection index of each LED lamp bead is

[0042] The modulation relief probability of each LED lamp bead is

[0043] The average modulation relief probability of each LED lamp bead is

[0044] The flashing frequency of each LED lamp bead is FBF=min[mod(f scene , f camera ), mod(-f scene , f camera )];

[0045] Among them, x max is the average pixel value of the frame with the largest average pixel value among the LED lamp beads in the column, x min is the average pixel value of the frame with the smallest average pixel value among the LED lamp beads in the column; meas Refers to the average pixel value of each frame of the LED lamp beads in the column, x refoff It refers to the average pixel value of a column of LED beads with the shortest average lighting time in the group, and τ is the preset flicker contrast threshold; It refers to the average pixel value of a row of LED lamp beads with the longest average lighting time in the group, and δ is the preset acceptable fluctuation range. Refers to the average pixel of all frames of the LED lamp bead with the longest average lighting time in the group; f scene is the flashing frequency of the LED lamp beads in this group, f camera is the shooting frequency of the camera to be tested.

[0046] As a preferred solution, outputting the image test result of the camera to be tested according to the image index includes:

[0047] When any image index of any LED lamp bead is within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested meets the classification standard;

[0048] When any image index of an LED lamp bead is not within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested does not meet the classification standard.

[0049] The present invention provides a traffic light for image quality testing and a testing method thereof, comprising: an LED array consisting of 4 rows and N columns of LED lamp beads, wherein the 4 rows of LEDs respectively use red, yellow, green, and white LED lamp beads; and a controller for outputting operating parameters to each LED lamp bead, wherein each LED lamp bead in the same column uses exactly the same operating parameters, while each LED lamp bead in the same row uses different operating parameters. Wherein, N is an integer greater than 1. This application provides a traffic light that meets the detection performance testing requirements of a traffic light detection system under different circumstances, thereby ensuring the test quality of the traffic light image quality test. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 1 is a schematic diagram of the structure of a traffic light for image quality testing provided by an embodiment of the present invention;

[0051] Figures 2 to 5 : is a schematic diagram of the appearance structure of a traffic light for image quality testing provided by an embodiment of the present invention; wherein, Figure 2 Right view of the traffic light used for image quality testing, Figure 3 This is a front view of a traffic light used for image quality testing. Figure 4 This is the left view of the traffic light used for image quality testing. Figure 5 A top view of a traffic light used for image quality testing;

[0052] Figure 6 4 is a flow chart of a method for testing a traffic light for image quality testing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0054] Current traffic light detection systems typically use a color feature-based recognition method. Existing test devices for traffic light detection systems typically only simulate the structural composition of traffic lights, with a single test scenario, which cannot meet the detection performance testing requirements of traffic light detection systems under different conditions.

[0055] To address the above technical issues, an embodiment of the present invention provides a traffic light for image quality testing, comprising:

[0056] An LED array consisting of 4 rows and N columns of LED lamp beads, with the 4 rows of LEDs using red, yellow, green and white LED lamp beads respectively;

[0057] The controller is used to output operating parameters to each LED lamp bead. The LED lamp beads in the same column use exactly the same operating parameters, and the LED lamp beads in the same row use different operating parameters.

[0058] Wherein, N is an integer greater than 1.

[0059] When implementing this embodiment, see Figure 1 , is a schematic structural diagram of a traffic light for image quality testing provided by an embodiment of the present invention, wherein the traffic light for image quality testing comprises an LED array composed of 4 rows and N columns of LED lamp beads, and a controller for controlling the lamp beads;

[0060] The 4 rows of LEDs use red, yellow, green and white LED beads respectively;

[0061] Through the special design and control logic of the LED array, the lighting changes in real traffic scenes are simulated, providing diverse test conditions for the image detection system.

[0062] For example, the color and arrangement scheme of the LED array provided in this case is: the first row: red LED, simulating the red light state of a traffic light; the second row: yellow LED, simulating the yellow light state; the third row: green LED, simulating the green light state; the fourth row: white LED, used to supplement ambient light or as a contrast condition (such as strong light during the day, reflection at night, etc.).

[0063] Using standard wavelength LEDs, the system accurately simulates the four colors of traffic lights: red, yellow, green, and white. The system detects hue variations by taking photos. Using a 4-row, 10-column layout, each row contains identical LEDs, but with varying operating parameters. Each light source can be independently controlled on and off. By combining multiple columns of LEDs, the system creates light patterns with varying brightness and color distribution, simulating the effects of traffic lights at varying distances, angles, and even in obstructed conditions.

[0064] When the controller is controlling, it performs column synchronization control: the four LEDs (red, yellow, green, and white) in the same column use the same operating parameters (such as current, voltage, pulse width, etc.) to ensure that the lights in the same column are consistent in brightness, flashing frequency, etc., which is convenient for simulating the vertical structure of traffic lights (such as the red, yellow, and green lights on the same lamppost).

[0065] Differentiated row control: LEDs in each column of the same row use different operating parameters, for example: brightness difference: simulates light attenuation or occlusion of traffic lights at different locations (such as partial damage to the lamp beads or dust coverage); flashing frequency difference: simulates scenes such as traffic light countdown and fault flashing; color saturation difference: simulates ambient light interference (such as color shift caused by rainy and foggy days).

[0066] By adjusting the operating parameters of each row and column, the solution can simulate the following typical scenarios: Light changes: Adjusting the background brightness through white LEDs simulates different lighting conditions such as daytime, dusk, and nighttime; Traffic light status changes: Independent control of the red, yellow, and green lights simulates normal switching of traffic lights, faulty and permanently lit states, and other states; Image interference factors: By differentiating row and column parameters, imaging interference such as lens jitter, pixel distortion, and partial occlusion is simulated.

[0067] It should be noted that when image quality testing is implemented using traffic lights, the LED array is constructed using 4 rows × N columns (N ≥ 2) of LED chip packages, with the same color LEDs (red / yellow / green / white) used in the same row and arranged vertically in the same column (such as red, yellow, green, and white lights on the vertical direction of a lamp post). The selected LEDs must meet the requirements of high brightness and a wide color temperature range (especially white LEDs), and support PWM (pulse width modulation) dimming to ensure that brightness and color parameters can be precisely controlled.

[0068] The controller also needs to be controlled by corresponding hardware, using ARM or FPGA chip as the main control, with multi-channel PWM drive circuit (at least 4 channels per column, corresponding to 4-color LED), supporting serial port, network port and other communication interfaces to receive test instructions; by setting the working parameters of each column (such as brightness value, flashing frequency, duty cycle, etc.).

[0069] Parameters for each column within a row can be adjusted independently. For example, the brightness of each column in the red row can be set in a gradient, forming a brightness gradient from left to right, simulating the perspective effect of traffic lights in the image (brighter near, darker far). The flashing frequency of each LED channel can be adjusted, and the brightness of each color can be adjusted independently to simulate different ambient lighting conditions. The camera module can also test the brightness level of a single column.

[0070] Fix the LED array on a bracket to simulate the installation height of a traffic light (such as 5-8 meters), and adjust the pitch angle and horizontal position; use an optical lens or scattering plate to simulate atmospheric attenuation (such as light scattering in fog and rain), or change the imaging focal length and distortion parameters through a lens simulator; combine with ambient light sources (such as a solar simulator or flash) to test the image detection system's anti-interference ability under complex lighting conditions.

[0071] This application solution supports the independent control and combination status of red, yellow and green lights, such as the red light and yellow light being on at the same time, covering all standard working modes of traffic lights; through the adjustment of white LED and row and column parameters, different test scenarios (such as direct strong light, backlight, low illumination, etc.) can be simulated to meet the requirements of traffic light detection standards. Through the differentiated control of the rows and columns of the LED array, the physical characteristics of the traffic light (color, brightness, flicker) are converted into precisely adjustable test parameters, providing a standardized and reproducible test environment for the image detection system. This application solution can provide a traffic light that meets the detection performance test requirements of the traffic light detection system under different circumstances, and ensure the test quality of the traffic light image quality test.

[0072] In another embodiment provided by the present invention, the operating parameters include brightness values, and the LED lamp beads in the same column all use exactly the same brightness value, while the LED lamp beads in the same row use different brightness values.

[0073] During the specific implementation of this embodiment, the brightness of the LED array is differentially controlled to simulate the illumination changes of traffic lights in different scenes, thereby providing multi-dimensional brightness test conditions for the image detection system.

[0074] When adjusting the LED brightness, the duty cycle of the LED driving current is adjusted through PWM (pulse width modulation) technology. The higher the duty cycle, the higher the LED brightness (for example, when the duty cycle is 50%, the brightness is 50% of the maximum value).

[0075] During specific control, the red, yellow, green and white LEDs in the same column use the same brightness value to simulate the lighting consistency at the same vertical position (such as the same height of the lamp post); the brightness of each column in the same color row is different, forming a horizontal brightness gradient or random distribution, simulating the perspective, occlusion or failure scenarios of traffic lights in the image (such as the brightness attenuation of some lamp beads).

[0076] By adjusting the column brightness value (such as 0.1-10000cd / m 2 ) to test the camera's exposure adaptability in extremely dark (nighttime) and extremely bright (direct sunlight) scenes; brightness differences within the same row can simulate surface defects such as stains and damage on traffic lights, testing the image recognition algorithm's tolerance to non-uniform brightness;

[0077] By adjusting the brightness value in real time (such as flickering and gradient), the detection system's accuracy in capturing rapid brightness changes (such as image capture at the moment when a traffic light switches) is tested.

[0078] Set the brightness of a column to the lowest possible value (e.g. 0.1cd / m 2 ), testing the camera's noise control and color reproduction in low light conditions;

[0079] Set the brightness of a column to the upper limit (e.g. 10000cd / m 2 ) to test the camera's overexposure suppression capability (such as halo and color distortion).

[0080] Through the design of column synchronization and row differentiation, the lighting characteristics of traffic lights are converted into quantifiable test parameters. Through precise brightness simulation, the performance indicators of the image detection system under dynamic lighting are exposed. And by comparing multiple colors with the same brightness, the image detection system's color recognition differences of different color traffic lights can be tested to obtain more comprehensive and accurate test results.

[0081] In another embodiment of the present invention, the operating parameters include flicker parameters, and the LED lamp beads in the same column use completely identical flicker parameters, while the LED lamp beads in the same row use different flicker parameters.

[0082] The flicker parameters include a flicker frequency and a flicker duty cycle.

[0083] During the specific implementation of this embodiment, the flickering parameters (frequency, duty cycle) of the LED array are differentiated to simulate the flickering state of traffic lights in faulty, abnormal or special scenarios, providing test conditions for dynamic flicker characteristics for the image detection system.

[0084] Flashing frequency, that is, the number of times the LED turns on and off per second. For example, 1Hz means it turns on and off once in 1 second.

[0085] The flashing duty cycle is the proportion of the duration of the LED being on in a single flash. For example, a flashing frequency of 1Hz and a duty cycle of 50% means that the LED is on for 0.5 seconds and off for 0.5 seconds.

[0086] The red, yellow, green and white LEDs in the same row use the same flashing parameters to simulate the synchronous flashing of lamps at the same height (e.g. the entire row of lights flashes synchronously when a fault occurs);

[0087] The flashing parameters of each column in the same color row are different, resulting in differences in horizontal flashing frequency or duty cycle, simulating local failures of traffic lights (such as abnormal flashing frequency of a red light in a column) or multi-light coordination scenarios (such as asynchronous flashing of traffic lights at adjacent intersections).

[0088] It should be noted that when conducting specific tests, the control variable method is generally used to conduct performance tests under a single parameter. This can be achieved by using the same brightness for all LED lamps but different flicker parameters to test the image detection system's ability to detect different color signal lights with different flicker levels. Similarly, using the same flicker parameters for all LED lamps but different brightness to test the image detection system's ability to detect different color signal lights with different brightness levels improves detection efficiency. Generally, when controlling flicker parameters, separate tests are performed by changing the flicker frequency or flicker duty cycle. To improve test efficiency, the flicker parameters can be changed simultaneously with the flicker frequency and flicker duty cycle. Alternatively, different brightness and flicker parameters can be used for each row of LED lamps to improve test efficiency.

[0089] By controlling the flicker parameters, it supports dynamic flicker capture capability testing, that is, testing whether the camera shutter speed and frame rate match through high-frequency flicker (such as above 10Hz) to avoid ghosting or flicker missed detection; performing flicker pattern recognition testing, simulating the abnormal on and off duration of the fault light through different duty cycles (such as 30%, 70%), and testing the image detection system's recognition accuracy for patterns such as "long on and short off" and "short on and long off"; performing multi-source flicker anti-interference testing, with different frequencies flashing in each column of the same row, simulating the interference scenario when multiple traffic lights work at the same time, and testing the system's signal separation capability.

[0090] In specific applications, a flashing control circuit needs to be built. Each column is equipped with an independent flashing control module, which includes a PWM driver chip (such as a NE555 timer or an MCU with built-in PWM function) and supports adjustable frequency (0.1-50Hz) and duty cycle (0-100%). The main control chip (such as Arduino Due) configures the parameters of each column uniformly through the serial port or bus protocol to ensure that the four-color LEDs in the column flash synchronously.

[0091] Fast-response LEDs (response time ≤ 10μs) are used to avoid the afterglow effect during high-frequency flickering (for example, there is no obvious trailing shadow when flickering at 50Hz); the array layout simulates the structure of a real traffic light: the red row is on the top (simulating a red light), the green row is on the bottom (simulating a green light), the yellow row is in the middle, and the white row can be used for background light or auxiliary light source.

[0092] Optimize camera shutter speed through high-frequency testing to meet highway intersection monitoring needs, improve the recognition accuracy of camera equipment in abnormal modes, and reduce traffic signal misreading caused by flicker misjudgment.

[0093] In specific applications, the traffic light for image quality testing provided by the present invention uses four rows of different light sources (red, green, blue, and white) to form 10 columns, providing a series of light sources with different brightnesses. By independently controlling the on and off of each light source and the brightness value of each light source, it helps evaluate the performance of imaging equipment when handling different brightness levels.

[0094] See also Figures 2 to 5 , is a schematic diagram of the appearance structure of a traffic light for image quality testing provided by an embodiment of the present invention. Figure 2 Right view of the traffic light used for image quality testing, Figure 3 This is a front view of a traffic light used for image quality testing. Figure 4 This is the left view of the traffic light used for image quality testing. Figure 5 Top view of a traffic light used for image quality testing.

[0095] The 40-step traffic light features a touchscreen, handle, power supply and switch, communication port, and fan. The touchscreen can adjust the brightness of the 40 light sources to 10 different levels, or it can be set to the default 40-step brightness setting. During testing, the tester places the light source in a darkroom and connects it to the power supply. The tester can control the light source via the touchscreen or by connecting a computer to the device's communication port. The camera is placed at an appropriate distance from the light source, and after adjusting the brightness, the camera simulates the effect of a traffic light in a complex and harsh environment. The results are then compared and analyzed to determine the imaging device's performance in handling different brightness levels.

[0096] Another embodiment of the present invention provides a method for testing a traffic light for image quality testing, the method being applied to the traffic light for image quality testing described in any of the above embodiments, see Figure 6 , is a flow chart of a method for testing a traffic light for image quality testing provided by an embodiment of the present invention, the method comprising the following steps:

[0097] Step S1, inputting test parameters of the traffic light for image quality testing, wherein the test parameters include operating parameters of each LED lamp bead;

[0098] Step S2, using the camera to be tested to obtain a test image of the traffic light for image quality testing;

[0099] Step S3, calculating the image index of the test image;

[0100] Step S4: outputting the image test result of the camera to be tested according to the image index.

[0101] In the specific implementation of this embodiment, the testing method provided by the present application scheme is based on the characteristics of traffic lights used for image quality testing. It simulates real traffic scenes through controllable LED lamp working parameters, and then combines image index calculation to achieve an objective evaluation of the performance of the camera to be tested.

[0102] For example, when performing the test, the test is performed in a dark room with an indoor illumination of ≤2 lux;

[0103] Script for setting 40-way brightness values; 40-level brightness value setting: default setting, brightest to darkest Lv20-Lv1 magnification 246;

[0104] Adjust the driving parameters of the camera to be tested to the best setting. Generally, the default parameters are used. The parameters related to the camera to be tested are set to normal mode, such as white balance and exposure.

[0105] Connect the traffic light used for image quality testing, the camera under test, and a control device (such as a computer) to ensure proper communication. Perform an initial calibration of the traffic light's LEDs to ensure accuracy of parameters such as brightness and color. Set basic parameters for the camera under test, such as resolution and exposure mode.

[0106] Adjust the camera under test so that the LED array is centered within the camera's field of view, approximately half the image height. Once the camera is stable and in focus, capture five images. Import the images into the RIQA-QR software for analysis.

[0107] Develop detailed test parameter combinations based on test requirements. For example, to test a camera's performance in low light conditions, set the traffic light brightness to a low level while adjusting the flicker parameters to simulate the normal operation of a nighttime traffic light. To test a camera's ability to capture dynamic scenes, set a higher flicker frequency and different duty cycles.

[0108] The test parameters are input into the controller of the traffic light used for image quality testing through the control device, and the controller adjusts the working state of each LED lamp bead according to the parameters.

[0109] Start the camera under test and, after the traffic light is operating stably, capture test images according to the pre-defined capture process. Capture images multiple times using the same parameter settings to reduce random errors. Preprocess the captured test images, including denoising and grayscale conversion, to facilitate subsequent performance calculations. Calculate various image metrics using professional image analysis software or algorithms.

[0110] For example, the clarity of an image is calculated using an edge detection algorithm; the color reproduction is calculated by comparing the color of traffic lights in the image with the standard color; and the dynamic range is evaluated by analyzing the brightness values ​​of the brightest and darkest areas in the image.

[0111] Comparing the calculated image metrics with pre-set standards or expected results can help determine the camera's performance under various test scenarios. Examples include: which channels saturate first at different light intensities; the hue and saturation values ​​of red and yellow signals across the entire brightness range; the delta threshold between hue / saturation values, which triggers an alert when the color signal becomes ambiguous or fails to meet classification criteria; and signal loss or misinterpretation under flicker conditions, particularly in multiple-exposure or rolling shutter sensor architectures.

[0112] By flexibly configuring traffic light operating parameters, various traffic scenarios, from normal to extreme, such as direct sunlight, low illumination, and traffic light malfunctions, can be simulated. This comprehensively tests camera performance in diverse environments, avoiding the single-scenario limitations of traditional testing methods and ensuring the integrity and reliability of test results. Quantitative calculation of image metrics presents camera image quality as concrete data, reducing subjectivity and errors in human judgment. Furthermore, by comparing results against standards or expectations, camera performance can be accurately assessed.

[0113] In another embodiment of the present invention, outputting the image test result of the camera to be tested according to the image indicator includes:

[0114] Compare the image indicators of each LED lamp bead in each row, and output the image test results of the camera to be tested on lamp beads of different colors.

[0115] During the specific implementation of this embodiment, the ability of the camera to be tested to capture, restore and process various colors is evaluated by comparing the index differences of the red, yellow, green and white LED lamp beads in the traffic light in the test image.

[0116] Different colored LEDs (red, yellow, green, and white) have unique spectral characteristics, resulting in varying brightness, color saturation, and contrast in images. For example, the red LED spectrum is concentrated in the long-wavelength region, making it susceptible to light interference and color shift in images. White LEDs cover the entire spectrum, placing higher demands on the camera's dynamic range. By analyzing these indicators, we can determine the accuracy and stability of the camera's reproduction of specific colors.

[0117] Comparing the image metrics of LEDs in the same row (same color) but different columns reveals the camera's consistency in processing the same color across different image regions. Comparing the metrics of LEDs in different rows (different colors) allows us to assess the camera's comprehensive processing capabilities for multiple colors. If a camera's clarity for red LEDs is significantly lower than for green LEDs, it indicates a flaw in its imaging in the red spectrum.

[0118] Quantify image indicators (such as color error ΔE, brightness uniformity, and edge clarity) and intuitively reflect the advantages and disadvantages of the camera in color processing through numerical comparison, avoid subjective judgment bias, and provide an objective basis for camera performance evaluation.

[0119] The test image is divided into regions according to the rows and columns of the traffic light LED array, and the image sub-regions corresponding to each column of red, yellow, green, and white LED beads are extracted respectively to ensure that each region only contains image information of a single color LED.

[0120] For each sub-region, an image analysis algorithm is used to calculate image metrics such as color reproduction (such as the ΔE difference from the standard color), brightness uniformity (the standard deviation of brightness within each column in the same row), and clarity (edge ​​sharpness). For example, by calculating the average ΔE value of the red LED sub-region, the camera's accuracy in reproducing the color red can be measured.

[0121] Compare the image indicators of LEDs in the same row (same color) but different columns horizontally. If the brightness uniformity index of a row varies significantly, it means that the camera has a deviation in capturing the brightness of that color in the horizontal area of ​​the image, which may be caused by lens distortion or uneven sensor response.

[0122] Compare the image metrics of LEDs in different rows (red, yellow, green, and white). For example, if the color error of yellow LEDs is generally higher than that of red and green, it indicates that the camera's processing capability for the yellow spectrum is weak and the color correction algorithm needs to be optimized.

[0123] Based on the comparison results, the imaging advantages and disadvantages of the camera to be tested for lamp beads of different colors are analyzed, and a comprehensive test result is given.

[0124] In real-world traffic scenarios, one of the most common failure modes is spectral channel saturation, where red or yellow signals overwhelm one color channel but not the others. This can make it difficult or impossible to distinguish between different colors of lights, especially when the image is tone mapped and passed through the ISP pipeline.

[0125] To further challenge perception systems, our platform also allows precise control of flicker parameters, including frequency and duty cycle. This enables testing under a wide range of signal conditions—from low-frequency flickering incandescent bulbs to Asian and high-frequency PWM-modulated LED signals. By simulating these flicker patterns, it's possible to determine how various sensors and algorithms respond to temporal instabilities, a major source of lost or ghosting signals.

[0126] By comparing image indicators by color and region, the differences in camera processing of specific colors can be accurately identified.

[0127] In another embodiment of the present invention, when the test parameters include that the LED lamp beads in the same column use the same brightness value, and the LED lamp beads in the same row use different brightness values;

[0128] The image indicators include the hue and color distance of each LED lamp bead.

[0129] During the specific implementation of this embodiment, the brightness of the traffic light LED lamp beads is controlled to be differentiated, and the hue and color distance indicators of each lamp bead in the image are combined to evaluate the color reproduction and processing capabilities of the camera to be tested under different brightness conditions.

[0130] Different brightness levels were set for the same row of LEDs to simulate the uneven brightness that can occur in real-world traffic lights due to distance, obstruction, or aging. At varying brightness levels, the spectral distribution of the LEDs can change, affecting hue (color variety) and color distance (the degree of deviation from a standard color). For example, at low brightness, red LEDs may appear dim and hue-shifted, while at high brightness, they may appear oversaturated. By analyzing hue and color distance at varying brightness levels, the camera's color adaptability and stability can be determined.

[0131] Each LED in the same column uses the same brightness value to ensure a uniform brightness benchmark in the vertical direction. This eliminates interference caused by brightness differences between columns when comparing color indicators in different rows (different colors), allowing test results to focus more on the camera's ability to process color.

[0132] Hue and color distance, as quantitative image metrics, provide a direct reflection of a camera's color reproduction accuracy. Hue measures color accuracy, while color distance (such as the ΔE value under the CIE 1976 standard) quantifies the degree of deviation from the standard value. By comparing the hue and color distance of LEDs of varying brightness and color, a comprehensive assessment of the camera's color processing performance in complex lighting environments can be achieved.

[0133] Compare the hue and color distance of LEDs with different brightness levels in the same row. For example, observe whether the hue shifts from low to high brightness in the red row, and whether the color distance increases with increasing brightness. This helps determine the camera's color stability at different brightness levels for that color.

[0134] Compare the hue and color distance differences between different rows (different colors) at the same brightness. For example, compare the color distance of red and green LEDs at 40% brightness to evaluate the camera's accuracy in reproducing different colors.

[0135] Summarize the brightness, hue, and color distance data for each LED bead in a table. Use a line graph to show how the hue and color distance change with brightness within the same row. Use a bar graph to compare the color distance differences between different rows at the same brightness. Based on the analysis results, write a test conclusion, highlighting the camera's strengths and weaknesses in color processing.

[0136] By combining brightness differentiation settings with hue and color distance indicators, we can accurately detect color reproduction problems of the camera under different brightness conditions, such as hue shift at low brightness and increased color distance at high brightness, providing a clear direction for optimizing the color algorithm.

[0137] In another embodiment provided by the present invention, the hue of each LED lamp bead is

[0138] The color distance of each LED lamp bead is

[0139] Among them, mea_v is the v value of the UV value of the LED lamp bead in the YUV space, mea_u is the u value of the UV value of the LED lamp bead in the YUV space, white-v is the v value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space, and white-u is the u value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space.

[0140] In the implementation of this embodiment, the color reproduction capabilities of each channel under different brightness conditions for the four colors red, yellow, green, and white are analyzed. This is primarily based on two metrics: hue and color distance. Hue, H, represents the position of a color on the color wheel, describing its hue. Color distance generally refers to the difference or distance between colors, and is used to describe the degree of similarity or difference between two colors.

[0141] Aim the camera under test at a traffic light. Once the light stabilizes at the set brightness, capture test images according to the pre-set process. To ensure data reliability, capture multiple sets of images using the same parameters.

[0142] The collected images are cropped to separate the image areas containing each LED lamp bead; denoising is performed to eliminate random noise in the image and provide clean data for subsequent indicator calculations.

[0143] Use a color space conversion algorithm (such as RGB to HSV) to extract the hue value of each LED bead image area. For example, after converting the RGB value of a red LED bead to HSV space, obtain its H (hue) channel value and record the hue performance of each bead.

[0144] Compare the color of each LED bead to a standard color (such as the corresponding color in the CIE standard color palette) and calculate the color difference ΔE value based on the CIE 1976 or CIE 2000 color difference formula. For example, the larger the ΔE value between the actual color of the red LED bead and the standard red, the more serious the color deviation.

[0145] The hue of each LED lamp bead is

[0146] The color distance of each LED lamp bead is

[0147] Among them, mea-v is the v value of the UV value of the LED lamp bead in the YUV space, mea-u is the u value of the UV value of the LED lamp bead in the YUV space, white-v is the v value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space, and white-u is the u value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space.

[0148] The comparison of quantified hue and color distance indicators can help R&D personnel specifically adjust the camera's color correction parameters and white balance algorithm, enabling the camera to more accurately restore colors in various brightness environments and improve the color authenticity of the image.

[0149] In another embodiment of the present invention, outputting the image test result of the camera to be tested according to the image indicator includes:

[0150] When the hue difference between any two LED lamp beads in the i-th row is within the preset hue difference range, and the color distance difference between any two LED lamp beads in the i-th row is within the preset color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row meets the classification standard;

[0151] When the hue difference between two LED lamp beads in the i-th row is not within the hue difference range, or the color distance difference between two LED lamp beads in the i-th row is not within the color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row does not meet the classification standard;

[0152] Among them, i=1, 2, 3, 4.

[0153] In the specific implementation of this embodiment, by comparing the relationship between the hue and color distance difference of LED lamp beads in the same row and the preset interval, the color reproduction stability and accuracy of the tested camera for a specific color are quantitatively evaluated. The core principle is as follows:

[0154] Hue reflects the essential properties of color (such as red, yellow, and green), while color distance measures the degree of deviation from a standard color. LEDs in the same row should theoretically display the same color. If the camera's color reproduction performance is good, the hue and color distance differences between LEDs of different brightness should be kept within a small range. By setting reasonable hue and color distance difference ranges, color reproduction can be transformed into a quantifiable and comparable criterion.

[0155] Because LEDs in a row have different brightness settings, brightness variations can affect color performance. If the camera's color reproduction capabilities are insufficient, the hue and color distance of LEDs with varying brightness levels will fluctuate significantly. By comparing the hue and color distance differences of each LED within the same row, we can assess the camera's color stability in scenes with varying brightness and determine whether it meets actual application requirements.

[0156] Independent judgment is performed for each of the four colors of a traffic light: red, yellow, green, and white. Different color spectra have distinct characteristics, placing varying demands on the camera's color processing capabilities. For example, the red spectrum is susceptible to interference from ambient light, while the yellow spectrum is more sensitive to changes in brightness. Color-based judgment allows for more detailed identification of the camera's strengths and weaknesses in color processing.

[0157] Determine reasonable hue and color distance difference ranges based on industry standards (such as CIE standards), camera application scenario requirements, and historical test data. For example, for intelligent traffic cameras, the hue difference range can be set to ±2° (HSV color wheel angle) and the color distance difference range can be set to ΔE ≤ 3 (CIE 2000 standard).

[0158] Based on the design indicators and test feedback of different camera models, the preset intervals are dynamically adjusted to ensure that the judgment criteria are both universal and meet the high-precision requirements of specific products.

[0159] According to the hue and color distance calculation method in the implementation process, the hue value and color distance value of each LED lamp bead in each row (each color) are extracted and calculated respectively.

[0160] For row i (i=1, 2, 3, 4), calculate the hue difference and color distance difference between any two LEDs. For example, row 1 (red row) has N LEDs, that is, the hue difference and color distance difference between any two LEDs.

[0161] Traverse the hue difference and color distance difference of all LED lamp beads in the i-th row. If all hue difference values ​​are within the preset hue difference value range, and all color distance difference values ​​are within the preset color distance difference value range, then it is determined that the color reproduction of the camera to be tested for the corresponding color in the i-th row meets the classification standard.

[0162] If there is at least one group of hue difference values ​​in the i-th row that exceeds the hue difference value interval, or there is at least one group of color distance difference values ​​that exceeds the color distance difference value interval, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row does not meet the classification standard.

[0163] Output four rows of color judgment results, marking whether each row meets the standard; for rows that do not meet the standard, list the specific hue difference and color distance difference data that exceed the range.

[0164] The image test result is determined based on the judgment results of the four rows of colors.

[0165] By comparing the differences and determining the intervals of different colors and indicators, we can quickly locate the color reproduction problems of the camera under specific color and brightness conditions.

[0166] In another embodiment of the present invention, when the test parameters include different brightness values ​​for each white LED lamp bead and turning off LED lamp beads of other colors: or,

[0167] When different brightness values ​​are used for each white LED lamp bead, and one color LED lamp bead is selected to use the same brightness value as the white LED lamp bead in the same column, and the other two colors of LED lamp beads are turned off; or,

[0168] When different brightness values ​​are used for each white LED lamp bead, and the same brightness value as the white LED lamp bead in the same column is used for any other color LED lamp bead;

[0169] The image index includes the saturation of each white LED lamp bead.

[0170] In the specific implementation of this embodiment, the white balance characteristics of the camera to be tested are quantitatively evaluated by comparing the saturation differences of white LED lamp beads with different brightness.

[0171] White balance is essentially the camera's ability to automatically correct for light of varying color temperatures, ensuring that white objects appear pure white in images. White LEDs cover the full spectrum, and their saturation directly reflects the camera's accuracy in reproducing white. A camera with good white balance should maintain stable saturation even when the brightness of the white LEDs varies (simulating varying light intensities). Conversely, significant saturation variations indicate a deviation in white balance, resulting in white distortion (e.g., a reddish or bluish cast).

[0172] Analyzing the white balance of the white channel in the last row, there are mainly two scenarios: turning on the white channel alone and turning on the white channel alone;

[0173] When the white channel is turned on alone, that is, the test parameters include different brightness values ​​for each white LED lamp bead, and the LED lamp beads of other colors are turned off: In this case, the test turns on the white channel alone and analyzes the white balance characteristics under different brightness conditions;

[0174] When the white channel is not turned on separately, it includes the white channel and one color channel, as well as turning on the white channel and other three color channels;

[0175] Corresponding:

[0176] The test parameters include using different brightness values ​​for each white LED lamp bead, selecting one color LED lamp bead to use the same brightness value as the white LED lamp bead in the same column, and turning off the other two colors of LED lamp beads;

[0177] In this case, the white balance characteristics in an environment that affects the white channel are tested by turning on the first row of red lights at different brightness conditions; turning on the second row of yellow lights at different brightness conditions are tested by turning on the third row of green lights at different brightness conditions.

[0178] When the white channel and the other three color channels are enabled, the test parameters include using different brightness values ​​for each white LED, and using the same brightness values ​​for all other color LEDs as the white LEDs in the same column. With all four rows of lights enabled and at different brightness levels, the white balance characteristics are tested under conditions that affect the white channel.

[0179] Saturation data under different test scenarios can form a standardized test data set to accurately evaluate the camera's color performance.

[0180] In another embodiment provided by the present invention, the saturation of each white LED lamp bead is s=(1-(3*min(R, G, B)) / (R+G+B))*100;

[0181] Among them, R, G, and B represent the grayscale values ​​of red, green, and blue of the pixels of the white LED lamp bead image in the RGB color space, respectively.

[0182] During the implementation of this embodiment, the RGB color space can be converted into other color spaces, such as the HSV color space. RGB color space: uses red (R), green (G), and blue (B) components to represent colors.

[0183] The HSV color space uses hue (H), saturation (S), and brightness (V) components to represent colors. White balance mainly refers to saturation S, and the formula is as follows:

[0184] S=(1-(3*min(R,G,B)) / (R+G+B))*100;

[0185] The saturation value ranges from 0 to 100%, indicating the purity of the color.

[0186] Using saturation as a core image metric, its numerical nature makes test results more intuitive and easier to compare. Compared to traditional subjective evaluation, saturation can be calculated through precise color space calculations (such as HSV or Lab color space), avoiding individual differences and errors in human judgment.

[0187] In another embodiment of the present invention, outputting the image test result of the camera to be tested according to the image indicator includes:

[0188] When the saturation difference between any two white LED lamp beads is within the preset saturation difference range, it is determined that the white balance characteristics of the camera to be tested meet the classification standard;

[0189] When the saturation difference between two white LED lamp beads is not within the saturation difference range, it is determined that the white balance characteristic of the camera to be tested does not meet the classification standard.

[0190] In the specific implementation of this embodiment, the preset saturation difference range is established based on industry standards, extensive test data, and actual application requirements. This range serves as a judgment threshold, converting white balance characteristics into a quantifiable and comparable standard. For example, for a smart traffic camera, if the saturation difference range is set to ±5%, when the saturation difference of all white LEDs falls within this range, the camera can be considered to meet the white reproduction consistency standard at different brightness levels.

[0191] Each white LED uses a different brightness value to simulate a variety of lighting scenarios, from low to high illumination. Testing saturation differences under these conditions comprehensively evaluates the camera's white balance stability in complex lighting environments, avoiding the limitations of traditional single brightness testing and more closely reproducing actual usage scenarios.

[0192] Calculate the saturation difference between any two white LED beads. If there are N white LED beads, then calculate C N,2 Group difference.

[0193] All saturation differences are traversed. If all differences are within the preset saturation difference range, the white balance characteristics of the camera to be tested are determined to meet the classification standards; if there is at least one group of differences outside the range, it is determined that the white balance characteristics of the camera to be tested do not meet the standards.

[0194] By comparing quantitative saturation differences, we can quickly and accurately identify any camera white balance issues. For example, if the saturation difference between low-brightness and high-brightness white LEDs is too large, it indicates that the camera's white balance correction capability is insufficient in low-light environments. Based on saturation difference analysis, we can adjust color temperature correction parameters and gain compensation algorithms, accelerating algorithm iteration and improving the camera's color reproduction accuracy in complex lighting environments.

[0195] In another embodiment provided by the present invention, the LED lamp beads having 4 rows and N columns are divided into k groups of lamp beads with m columns in each group;

[0196] When the test parameters include that each group of lamp beads adopts the same brightness value, and different columns of lamps in each group adopt the same flicker frequency and different flicker duty cycles, the image indicators include the flicker modulation index, flicker detection index, modulation relief probability, average modulation relief probability and flicker beat frequency of each LED lamp bead;

[0197] Wherein, m and k are integers greater than or equal to 1, and m×k≤N.

[0198] In the specific implementation of this embodiment, by grouping LED lamp beads and setting different flashing parameters, combining multiple image indicators, a complex traffic light flashing scene is simulated, thereby evaluating the capture, recognition and processing capabilities of the camera to be tested for dynamic flashing signals.

[0199] The LEDs in four rows and n columns are divided into k groups, each with m columns. By setting the same brightness value for each group, the basic lighting consistency of the test scene is ensured. Furthermore, the different columns within each group use the same flashing frequency but different flashing duty cycles to simulate the localized flashing variations that can occur in real-world traffic lights due to faults, synchronization anomalies, and other issues. For example, different groups of traffic lights may flash at inconsistent rhythms due to a controller malfunction, and different columns within the same group may flash for different durations, creating diverse and dynamic scenarios for camera testing.

[0200] Image metrics such as the Flicker Modulation Index (FMI), Flicker Detection Index (FDI), Modulation Relief Probability (MRP), Average Modulation Relief Probability (AMRP), and Flicker Beat Frequency (FFF) are introduced to quantify a camera's flicker signal processing performance from multiple dimensions. The Flicker Modulation Index (FMI) reflects the intensity variation of flicker signals, the Flicker Detection Index (FDI) measures the camera's ability to capture flicker signals, the MRP and AMRP assess the camera's stability in correctly identifying signals in complex flicker environments, and the Flicker Beat Frequency (FFF) determines the camera's accuracy in capturing flicker rhythms. Comprehensive analysis of these metrics allows for a comprehensive assessment of camera image quality in dynamic scenes.

[0201] Using the same flashing frequency across the same row of LEDs ensures a consistent vertical flashing rhythm, simulating the structural characteristics of a real traffic light. However, varying duty cycles across rows and the overall configuration between groups increase the complexity and diversity of the test scenarios. This coordinated parameter design allows the test to more closely replicate the complex dynamics of traffic lights in real-world traffic scenarios, enabling a more accurate assessment of the camera's performance when processing dynamic images.

[0202] According to the settings of m and k, the 4 rows and N columns of LED lamp beads are physically connected or logically divided into k groups, each group contains m columns of lamp beads. The controller sets the same brightness value for each group of lamp beads to ensure consistent basic lighting conditions. At the same time, the same flashing frequency (such as 2Hz) is set for different columns of lamps in each group, and different flashing duty cycles are assigned to each column.

[0203] There are 10 columns of lights, with the last column off. The first 9 columns are divided into three groups: columns 1, 2, and 3 form the first group, columns 4, 5, and 6 form the second group, and columns 7, 8, and 9 form the third group. The first light in each group is constantly on (ON) with a 100% duty cycle. The second light flashes (n% duty cycle) (Flicker). The third light is off (OFF) with a 100% duty cycle.

[0204] Control the camera under test to capture multiple sets of images containing the blinking status of LED lamp beads at a fixed frame rate (such as 30fps), ensuring that the acquisition time is long enough to cover the complete blinking cycle and various blinking combinations.

[0205] The collected images are subjected to pre-processing operations such as denoising and cropping to extract the effective image area containing the LED lamp beads in preparation for subsequent indicator calculations.

[0206] Calculate image indicators, including the flicker modulation index, flicker detection index, modulation relief probability, average modulation relief probability, and flicker beat frequency of each LED lamp bead. Specifically:

[0207] Analyze the brightness changes of LED lamp beads in different frames in the image, calculate the flicker modulation index through the formula, and quantify the intensity fluctuation of the flicker signal.

[0208] The actual flickering state of the LED lamp is compared with the flickering performance in the image captured by the camera, and the proportion of flicker signals correctly detected by the camera is evaluated to obtain the flicker detection index.

[0209] In complex flickering scenarios, the ratio of the number of times the camera correctly identifies flickering signals to the total number of times is counted to obtain the modulation relief probability; the modulation relief probability of multiple sets of test data is averaged to obtain the average modulation relief probability.

[0210] By analyzing the periodicity of the LED light bead flashing in the image, the flashing beat frequency is calculated and the accuracy of the camera's capture of the flashing rhythm is determined.

[0211] Comprehensively analyze the calculated image indicators, compare the differences in indicators of different groups and columns of lamp beads, and judge the performance of the camera in different flickering scenarios.

[0212] By grouping LEDs and setting different flashing parameters, it is possible to simulate the flashing states of traffic lights in various complex scenarios, such as normal operation, failure, and synchronization anomalies. Compared with a single flashing mode test, this can more comprehensively evaluate the image quality of the camera in an actual dynamic environment. The introduction of multiple image indicators allows for a quantitative assessment of the camera's ability to process flash signals from different dimensions, making the test results more objective and accurate.

[0213] In another embodiment provided by the present invention, the flicker modulation index of each LED lamp bead is

[0214] Among them, x max is the average pixel value of the frame with the largest average pixel value among the LED lamp beads in the column, x min is the average pixel value of the frame with the smallest average pixel value among the LED lamp beads in the column; max and x min It should be calculated based on the average value of the pixel ROI to average out the effect of image noise. max and x min The average value of at least 30 pixels is used for calculation.

[0215] The FMI indicator reflects the magnitude of the brightness change in the "flickering" area in the video, that is, a performance indicator of whether the flickering phenomenon is obvious or not.

[0216] The flicker detection index of each LED lamp bead is

[0217] Among them, x measRefers to the average pixel value of each frame of the LED lamp beads in the column, x refoff It refers to the average pixel value of a column of LED beads with the shortest average lighting time in the group. τ is the preset flicker contrast threshold. The flicker contrast threshold is the defined minimum acceptable Weber contrast, which is determined by the actual application scenario. The recommended value is 0.2.

[0218] The Flicker Detection Index (FDI) is the fraction of all video frames that meet the contrast threshold, with a maximum value of 1.0. Higher values ​​indicate good flicker mitigation (i.e., more measurement points are detectable compared to the background brightness level), while lower values ​​indicate poor flicker mitigation. The Flicker Detection Index (FDI) measures the likelihood that, for a given camera, LED frequency, and duty cycle, there will be sufficient contrast between the LED light and the reference light level in the captured image.

[0219] The modulation relief probability of each LED lamp bead is

[0220] in, It refers to the average pixel value of the LED column with the longest average lighting time in the group. δ is the preset acceptable fluctuation range. In the absence of specific camera application definitions, the recommended setting value is 0.1.

[0221] MMP represents the proportion of frames whose pixel values ​​in the "flicker" area of ​​the video fall within the acceptable fluctuation range, with a maximum value of 1.0, using the average value of the "on" area as the reference signal value. A higher value indicates good flicker mitigation (i.e., more measured values ​​fall within the acceptable threshold), while a lower value indicates poor flicker mitigation.

[0222] The average modulation relief probability of each LED lamp bead is

[0223] in, Refers to the average pixel of all frames of the LED lamp beads with the longest average lighting time in the group;

[0224] MMP_average represents the proportion of frames whose pixel values ​​in the flicker region fall within the acceptable fluctuation range, using the flicker region average as a relative reference signal value. The maximum value is 1.0. A higher value indicates good flicker mitigation (i.e., more measured values ​​fall within the acceptable threshold), while a lower value indicates poor flicker mitigation.

[0225] The flashing frequency of each LED lamp bead is FBF=min[mod(f scene ,f camera ),mod(-f scene ,f camera )];

[0226] Among them, f scene is the flashing frequency of the LED lamp beads in this group, f camera is the shooting frequency of the camera to be tested.

[0227] Flicker beat frequency (FBF) refers to the frequency of a modulated light source captured by a camera system. It is the modulation frequency within a video sequence, not the frequency of the physical light source. FBF is a key parameter in terms of visual saliency and its impact on the functionality of CV algorithms, and therefore should be quantified as part of flicker assessment.

[0228] In another embodiment of the present invention, outputting the image test result of the camera to be tested according to the image indicator includes:

[0229] When any image index of any LED lamp bead is within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested meets the classification standard;

[0230] When any image index of an LED lamp bead is not within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested does not meet the classification standard.

[0231] In this embodiment, the LEDs in four rows and N columns are divided into k groups, each with m columns, according to the proposed scheme. Each group is assigned the same brightness value, and the LEDs in different columns within a group are assigned the same flashing frequency but different flashing duty cycles. The camera under test is fixed at a standard test distance and angle to ensure consistent image acquisition conditions.

[0232] Control the camera to capture multiple sets of images containing the flashing status of LED lamp beads at a fixed frame rate (such as 30fps). The acquisition time must cover the complete flashing cycle and multiple flashing combinations to ensure the comprehensiveness of the data.

[0233] After preprocessing the captured images, such as denoising and cropping, the flicker modulation index, flicker detection index, modulation relief probability, average modulation relief probability, and flicker beat frequency are calculated for each LED bead. For example, the flicker modulation index is calculated by analyzing the brightness variation curve of the LED bead in different frames of the image. The flicker detection index is then derived by comparing the actual flicker state of the LED bead with the image.

[0234] Each LED lamp bead's image indicators are compared one by one with the corresponding preset range. For example, if the flicker detection index of an LED lamp bead is 92%, but the preset range is ≥95%, then the indicator does not meet the standard.

[0235] As long as any image indicator of any LED lamp bead is not within the corresponding preset range, the flicker suppression of the camera to be tested is determined to be inconsistent with the classification standard; only when all image indicators of all LED lamp beads are within the corresponding preset range, it is determined to be consistent with the classification standard.

[0236] Lists the image indicator values ​​of each LED bead, the corresponding preset range, and the judgment result of whether it meets the standard. In the case of non-compliance with the standard, the exceeded indicator and the corresponding LED bead group information are highlighted.

[0237] By comparing all indicators and all LEDs within a range, we can accurately identify any flicker suppression issues in the camera. Quantitative determination within pre-set indicator ranges avoids the errors and uncertainties of subjective judgment, making test results more objective and repeatable.

[0238] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A traffic light for image quality testing, characterized in that: include: An LED array consisting of 4 rows and N columns of LED lamp beads, with the 4 rows of LEDs using red, yellow, green and white LED lamp beads respectively; The controller is used to output operating parameters to each LED lamp bead. The LED lamp beads in the same column use exactly the same operating parameters, while the LED lamp beads in the same row use different operating parameters. Wherein, N is an integer greater than 1.

2. The traffic light for image quality testing according to claim 1, wherein: The working parameters include brightness values. The LED lamp beads in the same column all use exactly the same brightness value, and the LED lamp beads in the same row use different brightness values.

3. The traffic light for image quality testing according to claim 1, wherein: The operating parameters include flickering parameters. The LED lamp beads in the same column use exactly the same flickering parameters, while the LED lamp beads in the same row use different flickering parameters. The flicker parameters include a flicker frequency and a flicker duty cycle.

4. A method for testing traffic lights for image quality testing, characterized in that: The method is applied to the traffic light for image quality testing according to any one of claims 1 to 3, and the method comprises: Inputting test parameters of the traffic light for image quality testing, wherein the test parameters include operating parameters of each LED lamp bead; Using the camera to be tested to obtain a test image of the traffic light for image quality testing; Calculating image indicators of the test image; Output the image test result of the camera to be tested according to the image index.

5. The method for testing a traffic light for image quality testing according to claim 4, wherein: Outputting the image test result of the camera to be tested according to the image indicator includes: Compare the image indicators of each LED lamp bead in each row, and output the image test results of the camera to be tested on lamp beads of different colors.

6. The method for testing a traffic light for image quality testing according to claim 4, wherein: When the test parameters include that the LED lamp beads in the same column use the same brightness value, and the LED lamp beads in the same row use different brightness values; The image indicators include the hue and color distance of each LED lamp bead.

7. The method for testing a traffic light for image quality testing according to claim 6, wherein: The hue of each LED lamp bead is The color distance of each LED lamp bead is Among them, mea_v is the v value of the UV value of the LED lamp bead in the YUV space, mea_u is the u value of the UV value of the LED lamp bead in the YUV space, white_v is the v value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space, and white_u is the u value of the UV value of the white LED lamp bead in the same column of the LED lamp bead in the YUV space.

8. The method for testing a traffic light for image quality testing according to claim 6, wherein: Outputting an image test result of the camera to be tested according to the image indicator includes: When the hue difference between any two LED lamp beads in the i-th row is within the preset hue difference range, and the color distance difference between any two LED lamp beads in the i-th row is within the preset color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row meets the classification standard; When the hue difference between two LED lamp beads in the i-th row is not within the hue difference range, or the color distance difference between two LED lamp beads in the i-th row is not within the color distance difference range, it is determined that the color reproduction performance of the camera to be tested for the corresponding color in the i-th row does not meet the classification standard; Among them, i=1, 2, 3, 4.

9. The method for testing a traffic light for image quality testing according to claim 4, wherein: When the test parameters include different brightness values ​​for each white LED lamp bead and turning off the LED lamp beads of other colors: or, When different brightness values ​​are used for each white LED lamp bead, and one color LED lamp bead is selected to use the same brightness value as the white LED lamp bead in the same column, and the other two colors of LED lamp beads are turned off; or, When different brightness values ​​are used for each white LED lamp bead, and the same brightness value as the white LED lamp bead in the same column is used for any other color LED lamp bead; The image index includes the saturation of each white LED lamp bead.

10. The method for testing a traffic light for image quality testing according to claim 9, wherein: The saturation of each white LED is s = (1-(3*min(R,G,B)) / (R+G+B))*100; Among them, R, G, and B represent the grayscale values ​​of red, green, and blue of the pixels of the white LED lamp bead image in the RGB color space, respectively.

11. The method for testing a traffic light for image quality testing according to claim 9, wherein: Outputting an image test result of the camera to be tested according to the image indicator includes: When the saturation difference between any two white LED lamp beads is within the preset saturation difference range, it is determined that the white balance characteristics of the camera to be tested meet the classification standard; When the saturation difference between two white LED lamp beads is not within the saturation difference range, it is determined that the white balance characteristic of the camera to be tested does not meet the classification standard.

12. The method for testing a traffic light for image quality testing according to claim 4, wherein: Divide the LED lamp beads into k groups with 4 rows and N columns according to each group of m columns; When the test parameters include that each group of lamp beads adopts the same brightness value, and different columns of lamps in each group adopt the same flicker frequency and different flicker duty cycles, the image indicators include the flicker modulation index, flicker detection index, modulation relief probability, average modulation relief probability and flicker beat frequency of each LED lamp bead; Wherein, m and k are integers greater than or equal to 1, and m×k≤N.

13. The method for testing a traffic light for image quality testing according to claim 12, wherein: The flicker modulation index of each LED lamp bead is The flicker detection index of each LED lamp bead is The modulation relief probability of each LED lamp bead is The average modulation relief probability of each LED lamp bead is The flashing frequency of each LED lamp bead is FBF=min[mod(f scene , f camera ), mod(-f scene , f camera )]; Among them, x max is the average pixel value of the frame with the largest average pixel value among the LED lamp beads in the column, x min is the average pixel value of the frame with the smallest average pixel value among the LED lamp beads in the column; meas Refers to the average pixel value of each frame of the LED lamp beads in the column, x refoff It refers to the average pixel value of a column of LED beads with the shortest average lighting time in the group, and τ is the preset flicker contrast threshold; It refers to the average pixel value of a row of LED lamp beads with the longest average lighting time in the group, and δ is the preset acceptable fluctuation range. Refers to the average pixel of all frames of the LED lamp bead with the longest average lighting time in the group; f scene is the flashing frequency of the LED lamp beads in this group, f camera is the shooting frequency of the camera to be tested.

14. The method for testing a traffic light for image quality testing according to claim 12, wherein: Outputting an image test result of the camera to be tested according to the image indicator includes: When any image index of any LED lamp bead is within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested meets the classification standard; When any image index of an LED lamp bead is not within the corresponding preset index range, it is determined that the flicker suppression of the camera to be tested does not meet the classification standard.

Images