Traffic sign recognition test system and method

Abstract

A traffic sign recognition test system of the present invention comprises: an LED display panel that displays the content of a traffic sign in variable manner by using a pulse width modulation (PWM) scheme; a lighting device that adjusts illumination conditions of an experimental environment in which the LED display panel is located, by adjusting illuminance and a light source direction; an environment control device that adjusts transmittance of light emitted from the LED display panel, by using an environment filter for simulating an external environment; a windshield control device that artificially generates optical distortion for the light transmitted through the environment control device; a collimator that adjusts a focal length of the light transmitted through the windshield control device; an image capturing device that generates a captured image of the LED display panel by using the light transmitted through the collimator; and an image analysis device that determines a recognition success rate of the traffic sign by analyzing the captured image received from the image capturing device.

Description

Traffic Sign Recognition Test System and Method

[0001] The present invention relates to a traffic sign recognition test system and method.

[0002] In autonomous driving, traffic sign recognition is an essential technology for vehicles to understand their surroundings and drive safely. Since traffic signs serve as a means of conveying rules and warnings—such as speed limits, stop signals, one-way streets, and pedestrian prohibitions—it is difficult to guarantee safety on the road if the autonomous driving system fails to recognize them accurately.

[0003] Autonomous vehicles generally perceive their surrounding environment by utilizing various sensors such as cameras, LiDAR, and radar. Among these, traffic sign recognition is primarily achieved through AI-based computer vision technology using front-facing camera footage. The vehicle detects candidate traffic signs within the video, classifies them, and determines the type of sign. Recently, recognition models utilizing deep learning technology are being developed to respond robustly to various weather conditions, changes in lighting, damage or contamination of signs, and changes in angle.

[0004] These traffic sign recognition capabilities significantly enhance the accuracy and safety of autonomous driving by allowing the vehicle to make judgments on behalf of the driver regarding information they might otherwise miss, or by automatically adjusting speed when driving on highways. This is particularly important because misrecognizing or missing traffic signs in various situations, such as complex urban environments or rural roads, can lead to accidents or traffic violations. Furthermore, since the shape and meaning of traffic signs vary slightly by country or region, the ability to learn various types of signs and accurately recognize them is essential for autonomous vehicles to be utilized internationally.

[0005] A 'Traffic Sign Recognition Test System' is used to evaluate and verify the performance of such traffic sign recognition technology. This system is utilized to test how accurately a vehicle's recognition algorithms or artificial intelligence models recognize traffic signs in real-world driving environments, and typically includes sign image datasets from various countries and regions, simulation environments or actual road driving data, and evaluation criteria.

[0006] In particular, for autonomous driving technology to be commercialized, it must meet international safety standards, and traffic sign recognition testing is considered one of the key items in this process. Therefore, traffic sign recognition test systems have evolved beyond simple algorithm performance verification tools to establish themselves as essential evaluation elements for ensuring the overall driving safety and reliability of autonomous vehicles.

[0007] The technical problem that the present invention aims to solve is to provide a traffic sign recognition test system and method that measures quantitative performance such as recognition accuracy, false positive rate, and omission rate of traffic signs, and comprehensively determines recognition ability under various conditions such as weather changes, illumination, distance, and damage to the signs.

[0008] A traffic sign recognition test system according to an embodiment of the present invention comprises: an LED display panel that displays the content of a traffic sign using a Pulse Width Modulation (PWM) method to display it variably; a lighting device that adjusts the lighting conditions of an experimental environment in which the LED display panel is located by adjusting the illuminance and the direction of the light source; an environment control device that controls the transmittance of light emitted from the LED display panel using an environment filter to simulate an external environment; a front glass control device that artificially generates optical distortion for the light transmitted through the environment control device; a collimator that adjusts the focal length of the light transmitted through the front glass control device; an image capturing device that generates a captured image of the LED display panel using the light transmitted through the collimator; and an image analysis device that analyzes the captured image received from the image capturing device to determine the recognition success rate of the traffic sign.

[0009] According to an embodiment, the front windshield control device may include a glass material having the same curvature and optical and geometric characteristics as the front windshield of a vehicle, a transparent plate that transmits light transmitted through the environment control device, a lower frame structure supported on the floor with a fixing leg at the bottom, a frame structure including an upper frame structure that supports the transparent plate at the top of the lower frame structure, a transparent plate fixing part disposed at both edges of the upper frame structure to support both ends of the transparent plate and provided as a rotatable clamp, and an inclination adjustment part connected to the upper frame structure and the lower frame structure to adjust the inclination of the upper frame structure with respect to the ground.

[0010] According to an embodiment, the collimator comprises a lens disk having a plurality of first openings arranged along the circumferential direction and a support member that supports the lens disk and has a first central hole formed therein, an optical lens having different optical characteristics is inserted into the first openings, and the optical transmission characteristics can be selected by aligning the first openings with the first central hole through rotation of the lens disk.

[0011] According to an embodiment, the environment control device comprises a filter disk having a plurality of second openings arranged along the circumferential direction and a support frame that supports the filter disk and has a second central hole formed therein, wherein the environment filter having different optical characteristics is inserted into the second openings, and the environment characteristics can be selected by aligning the second openings with the second central hole through the rotation of the filter disk.

[0012] According to an embodiment, the LED display board includes LED elements arranged in a grid shape, controls individual driving at the pixel level, and can control the display of the traffic sign through PWM (Pulse Width Modulation)-based scanning signal control.

[0013] According to an embodiment, the image analysis device performs preprocessing of the captured image through a previously trained artificial intelligence model, recognizes the traffic sign based on object detection technology, determines the country in which the recognized traffic sign is used using previously stored traffic sign data, and can determine the false positive rate and false negative rate by comparing the recognized traffic sign with actual sign information.

[0014] According to an embodiment, the image analysis device may store a test data log containing a performance judgment result under implemented test conditions and convert the test data log into a training dataset for the artificial intelligence model.

[0015] A traffic sign recognition test method according to another embodiment of the present invention comprises the steps of: displaying a traffic sign by means of an LED display board; illuminating the LED display board by adjusting the illuminance and the direction of the light source; adjusting the transmittance of light emitted from the LED display board using an environment filter to simulate an external environment; artificially generating optical distortion through a transparent plate for the light transmitted through the environment filter; adjusting the focal length of the light transmitted through the transparent plate using an optical lens; generating a captured image of the LED display board using the light transmitted through the optical lens; and analyzing the captured image to determine the recognition success rate of the traffic sign.

[0016] According to an embodiment, the step of determining the recognition success rate may include: a step of performing a preprocessing operation on the captured image through a previously trained artificial intelligence model; a step of recognizing the traffic sign based on object detection technology; a step of determining the country in which the recognized traffic sign is used using previously stored traffic sign data; and a step of determining the false positive rate and false negative rate by comparing the recognized traffic sign with actual sign information.

[0017] According to an embodiment, the method may further include the step of storing a test data log containing a performance judgment result under implemented test conditions, and the step of converting the test data log into a training dataset for an artificial intelligence model.

[0018] According to the traffic sign recognition test system and method of the embodiment of the present invention, a highly reliable test environment is established to evaluate how accurately a vehicle recognizes traffic signs, and the accuracy of how accurately a vehicle recognizes traffic signs on an actual road can be evaluated by considering various environmental factors and optical changes.

[0019] FIG. 1 is a schematic block diagram of a traffic sign recognition test system according to an embodiment of the present invention.

[0020] FIG. 2 is a diagram illustrating the operation of a traffic sign recognition test system according to an embodiment of the present invention.

[0021] FIG. 3 is a drawing of a front glass control device according to an embodiment of the present invention.

[0022] FIG. 4 is a drawing for explaining a front glass control device according to an embodiment of the present invention.

[0023] FIG. 5 is a drawing of a collimator according to an embodiment of the present invention.

[0024] FIG. 6 is a drawing of an environment control device according to an embodiment of the present invention.

[0025] FIG. 7 is a drawing for explaining the operation of an LED display board according to an embodiment of the present invention.

[0026] FIG. 8 is a schematic block diagram of an image analysis device according to an embodiment of the present invention.

[0027] FIG. 9 is a flowchart illustrating a traffic sign recognition test method according to an embodiment of the present invention.

[0028] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0029] The terms used in this specification are for describing embodiments and are not intended to limit the invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms "comprises" and / or "comprising" as used in this specification do not exclude the presence or addition of one or more other components in addition to the components mentioned. Throughout the specification, the same reference numerals refer to the same components, and "and / or" includes each of the mentioned components and all combinations of one or more. Although terms such as "first," "second," etc., are used to describe various components, these components are not limited by these terms. These terms are used merely to distinguish one component from another. Therefore, the first component mentioned below may be the second component within the technical scope of the invention.

[0030] Unless otherwise defined, all terms used herein (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0031] The terms “part” or “module” as used in the specification refer to software or hardware components, such as FPGAs or ASICs, and “part” or “module” perform certain roles. However, “part” or “module” is not limited to software or hardware. “Part” or “module” may be configured to reside in an addressable storage medium or configured to run on one or more processors. Thus, by example, “part” or “module” includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and “parts” or “modules” may be combined into a smaller number of components and “parts” or “modules,” or further separated into additional components and “parts” or “modules.”

[0032] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0033] FIG. 1 is a schematic block diagram of a traffic sign recognition test system according to an embodiment of the present invention.

[0034] Referring to FIG. 1, a traffic sign recognition test system (10) according to an embodiment of the present invention can provide an integrated test environment for evaluating and verifying traffic sign recognition technology, which is an essential function of autonomous vehicles and advanced driver assistance systems (ADAS).

[0035] The traffic sign recognition test system (10) establishes a reliable test environment in which a vehicle can evaluate the recognition rate of traffic signs, and can evaluate how accurately traffic signs can be recognized by considering various environmental factors and optical changes.

[0036] In particular, the traffic sign recognition test system (10) can not only evaluate whether the sign is recognized, but also analyze the recognition performance of the traffic sign by considering variables such as lighting environment, weather conditions, degree of damage to the sign, and optical distortion.

[0037] To this end, the traffic sign recognition test system (10) includes an image analysis device (100), an image capturing device (200), a front windshield control device (300), a collimator (400), an environment control device (500), an LED display (600), and a lighting device (700).

[0038] The image analysis device (100) can analyze and evaluate the captured image received from the image capturing device (200) to measure how accurately an autonomous vehicle or an ADS recognizes traffic signs.

[0039] In particular, the image analysis device (100) can measure the error rate by comparing the recognized sign with the actual sign information using deep learning-based image analysis technology and verify whether the traffic sign recognition function meets the legal standards of each country.

[0040] The image analysis device (100) includes various functions such as detection, recognition, error analysis, performance evaluation by environment, and generation of training data for traffic signs, and is capable of quantitatively evaluating the recognition rate under various conditions, rather than simply processing image data.

[0041] The image analysis device (100) receives a captured image input from the image capturing device (200) and performs a preprocessing operation, and in this process, can perform operations such as noise removal, contrast adjustment, and color correction.

[0042] In road environments, lighting conditions are not constant, so the captured image may be too bright or too dark, and if these variables are not corrected, recognition performance may be degraded. Therefore, the image analysis device (100) can improve the readability of the sign by adjusting the brightness and contrast of the captured image and standardizing the colors to maintain the quality of the captured image constant.

[0043] In addition, the image analysis device (100) can detect traffic signs in the captured image. The image analysis device (100) can detect traffic signs based on object detection technology and can identify areas where traffic signs exist in the captured image using a pre-trained artificial intelligence model. Here, the artificial intelligence model may use a face detection algorithm using feature points (landmarks), a deep learning-based YOLO (You Look Only Once) model, MobileNet, R-CNN (Recursive Convolutional Neural Network), Faster-RCNN, SSD (Single Shot Multibox Detector), etc., but is not limited to these, and known algorithm models for object recognition may also be used.

[0044] In accordance with the implementation, the image analysis device (100) may apply a multi-scale analysis technique to enable recognition of traffic signs under various conditions. For example, to simultaneously recognize a sign far away and a sign close by, it scans areas of various sizes within the image, and even if the sign is partially obscured or damaged, it can be detected by compensating for it.

[0045] The image analysis device (100) can identify the type of traffic sign. Since the types of traffic signs may vary by country and even signs of the same country may have different designs, the image analysis device (100) can perform classification using previously stored traffic sign data. In addition, the image analysis device (100) can identify various types of signs, such as speed limit signs, stop signs, and no parking signs, and in this process, it can additionally utilize OCR (Optical Character Recognition) technology to recognize number or character-based signs (e.g., speed limit 50 km / h).

[0046] The image analysis device (100) can detect and classify traffic signs and evaluate the accuracy of the recognition results, and can calculate indicators such as the recognition success rate, false positive rate, and false negative rate by comparing the recognized sign with the actual sign information. For example, if the image analysis device (100) incorrectly recognizes a speed limit sign of 80 km / h as 30 km / h, this is considered an error and the image recognition algorithm can be improved.

[0047] According to an embodiment, the image analysis device (100) can analyze performance in a specific environment by evaluating changes in recognition rate when traffic signs are partially obscured or when light reflection is severe.

[0048] According to another embodiment, the image analysis device (100) can evaluate the accuracy of traffic sign recognition results under various weather and lighting conditions. For example, the image analysis device (100) can use an environment control device (500) to create a nighttime environment, a rainy environment, a snowy environment, a foggy environment, etc., and evaluate the traffic sign recognition rate under these conditions.

[0049] The image analysis device (100) can analyze the cause of performance degradation in a specific environment and devise improvement measures, and can store test data logs and convert them into a training dataset to be used for training an artificial intelligence model.

[0050] The dataset constructed by the image analysis device (100) can be used as data for future autonomous vehicles to perform more accurate sign recognition even in new environments, and can also provide a real-time feedback system to enable immediate evaluation during testing.

[0051] If the image analysis device (100) continuously misidentifies a specific sign, it may provide a test data log to the administrator so that the algorithm can be modified or the training data can be supplemented.

[0052] The image analysis device (100) can perform an integrated test by providing control signals to each of the image capturing device (200), front glass control device (300), collimator (400), environment control device (500), LED display board (600), and lighting device (700). For example, the image analysis device (100) can adjust the direction of light projected onto the LED display board (600) by providing a control signal to the lighting device (700) to evaluate how well a sign is recognized in backlit conditions.

[0053] The video recording device (200) can generate a captured image by capturing an LED display board (600). The video recording device (200) can be used to evaluate how accurately a traffic sign is recognized in an environment identical to when a vehicle is actually driving and capturing a traffic sign.

[0054] The video recording device (200) can capture detailed information about the sign using a high-resolution camera and can perform tests such as wide-angle shooting and telephoto shooting using various lenses.

[0055] The windshield control device (300) is a device for implementing conditions identical to the characteristics of a windshield mounted on an actual vehicle within a test environment in order to evaluate the reliability of a camera-based image recognition function used in an autonomous driving system or ADAS (Advanced Driver Assistance System). The windshield control device (300) can configure conditions identical to those of an actual vehicle, including the location where the image capturing device (200) is mounted, the path of light passing through the glass, and even the optical and geometric characteristics of the glass itself.

[0056] Since the process of recognizing traffic signs in autonomous driving systems relies on a structure where a camera receives external visual information passing through the windshield, image quality can vary depending on the type or characteristics of the windshield. For instance, light may refract according to the curvature of the glass, causing the shape of the sign to be slightly distorted, or the amount of light reaching the camera sensor may decrease if the glass transmittance is low. Furthermore, factors such as glass surface coatings or color filter arrays can selectively filter out light of specific wavelengths or cause double reflection, which can affect the accuracy of camera recognition.

[0057] The front windshield control device (300) can be designed to maintain these characteristics in an experimental environment. That is, by using glass with the same curvature, curved surface eccentricity, filter transmittance, polarization rate, and reflection characteristics as the front windshield used in actual autonomous vehicles, the traffic sign image output from the LED display can reach the image capturing device (200) under the same conditions. Through this setup, various viewing conditions and image distortions that an autonomous vehicle may encounter on actual roads can be reproduced in a laboratory, and how the recognition algorithm responds to them can be quantitatively evaluated.

[0058] The collimator (400) is an optical device for adjusting the focal length of an image capturing device (200) mounted in an autonomous driving system or ADAS (Advanced Driver Assistance System), and can be used to simulate the traffic sign recognition distance according to the change in driving speed of the autonomous vehicle in the traffic sign recognition test system (10).

[0059] The collimator (400) can be placed between the front glass control device (300) and the image capturing device (200).

[0060] The collimator (400) enables experiments to evaluate and correct how the image capturing device (200) located inside the autonomous vehicle focuses on signs at various distances. In particular, as the autonomous vehicle travels at high speeds, it must recognize signs early, and accordingly, the focus distance must be appropriately adjusted to account for the changed distance conditions.

[0061] The collimator (400) includes a lens disc and an optical alignment device and is configured to experimentally implement the focus adjustment range of the camera through various lenses of a rotatable or replaceable structure. For example, the collimator (400) can insert optical components with different focal lengths or refractive characteristics of lenses and adjust the distance perception of traffic sign images through the lenses so that recognition conditions at a distance of 30m and recognition conditions at a distance of 100m can be switched and tested within the same experimental system.

[0062] Through this, the tester can quantitatively determine when and at what distance traffic signs can be recognized in a specific driving scenario, and analyze the image quality and the accuracy of the algorithm response according to changes in recognition distance. The collimator (400) can be used to flexibly respond to such changes in conditions and precisely reproduce the operating environment of the image capturing device (200).

[0063] That is, the collimator (400) functions as a focal length adjustment tool to experimentally implement and analyze how the image capturing device (200) reacts to changes in the traffic sign recognition distance and at which point optimal recognition occurs, and through this, ADAS developers can simulate and optimize recognition performance according to actual vehicle driving speed and distance changes in a laboratory environment.

[0064] The environment control device (500) is positioned between the windshield control device (300) and the LED display (600), and can artificially provide weather conditions such as fog, rain, and snow to evaluate how accurately the autonomous vehicle recognizes signs in various environments.

[0065] For example, traffic signs may appear blurry when it rains or snows, and visibility may be limited when there is fog. In addition, the performance of the image capturing device (200) may deteriorate in extreme environments such as heatwaves or extreme cold. The environment control device (500) can reproduce these conditions in a laboratory environment to test whether the autonomous vehicle can stably recognize signs even under various weather conditions that it may encounter on actual roads.

[0066] The LED display (600) can flexibly create a test environment by digitally displaying various types of traffic signs. In traditional traffic sign testing, a method was used to evaluate recognition performance by installing actual signs and photographing them, but this method has physical limitations in testing various signs.

[0067] However, the LED display (600) can easily display traffic signs used in specific countries or regions, and can configure various test scenarios by adjusting the color, size, brightness, etc. of the signs. In addition, the LED display (600) can evaluate the recognition performance of vehicles in real time while sequentially changing traffic signs, so it can perform more efficient testing.

[0068] The LED display board (600) is a digital display device controlled by Pulse Width Modulation (PWM) to variably display the content of traffic signs, and can have a direct impact on the performance of autonomous driving and ADAS systems, particularly camera-based Traffic Sign Recognition (TSR) functions.

[0069] The PWM method is a control method based on high-speed flashing to efficiently adjust the brightness and content of an electronic display board. While this appears as a continuous image to the human eye, it can be perceived as intermittent flickering by camera sensors—particularly CMOS image sensors. This phenomenon can lead to serious recognition errors in situations where the TSR function requires accurate recognition of sign content.

[0070] In particular, the CMOS sensor adopts a rolling shutter structure, so instead of capturing the entire image at once, it forms an image by scanning sequentially from top to bottom. This structure increases the likelihood that if the timing is not matched with the LED display that flashes using the PWM method, some frames will be captured with the LEDs off, or patterns will be recognized as distorted or missing.

[0071] To solve this, the video recording device (200) has an LFM (LED Flicker Mitigation) function built in, and the LFM adjusts the exposure timing or integration time of the sensor in response to the blinking frequency of the LED display (600) and performs the function of minimizing recognition errors that occur when the display is turned off. However, the setting value of this function may vary depending on the camera manufacturer, chipset, and sensor structure, and in the actual environment, the recognition result may become unstable due to complex factors such as the vehicle's driving speed, distance, and changes in illumination.

[0072] If vehicle speed, distance, or camera settings differ under identical conditions, traffic signs may not be recognized properly, or parts of the displayed patterns may disappear or become distorted. This is a typical example of TSR performance degradation that occurs when the camera's internal LFM settings do not match the tested frequency or the actual PWM frequency of the display, or when sensor optimization is insufficient.

[0073] Therefore, the LED display (600) is not merely a display device for displaying traffic signs, but a test condition element that must provide visual output quality that matches the characteristics of the CMOS image sensor, and can display traffic sign images to perform fine tuning and verification of the LFM function for normal operation of the TSR function.

[0074] The lighting device (700) can provide various illumination conditions to the LED display board (600) by reproducing various brightness conditions using artificial light. The lighting device (700) can provide not only white light but also yellow light generated by streetlights.

[0075] FIG. 2 is a diagram illustrating the operation of a traffic sign recognition test system according to an embodiment of the present invention.

[0076] Referring to FIG. 2, to perform a traffic sign recognition test, a front windshield control device (300), a collimator (400), and an environment control device (500) may be placed between the LED display (600) and the video recording device (200), and complex and variable conditions that an autonomous vehicle may encounter in a real road environment can be reproduced as a controllable experimental environment through environment simulation, parallel light adjustment, focal length adjustment, etc.

[0077] The LED display (600) can digitally display an image of a traffic sign for testing. The LED display (600) can continuously test various traffic signs by switching between them, and can display not only the design of the sign used on the actual road, but also test images with adjusted color, brightness, and shape.

[0078] The environment control device (500) is located between the LED display panel (600) and the front glass control device (300), and can control the transmittance of light using a filter to simulate weather conditions or external environments. For example, the environment control device (500) can implement cloudy weather, fog, rain, snow, etc. through a filter to evaluate whether the traffic sign recognition algorithm works well under various conditions.

[0079] The windshield control device (300) is located between the environment control device (500) and the collimator (400), and can artificially cause optical distortion to the light passing through the environment control device. In particular, the windshield control device (300) can configure conditions identical to those of an actual vehicle, including the location where the image capturing device (200) is mounted, the path of light passing through the windshield, and the optical and geometric characteristics of the windshield itself.

[0080] The collimator (400) is located between the windshield control device (300) and the image capturing device (200) and can perform focal length adjustment of the image capturing device (200) to simulate the traffic sign recognition distance according to the driving speed change of the autonomous vehicle.

[0081] The video recording device (200) can generate a captured video by capturing a sign image displayed on an LED display board (600).

[0082] The lighting device (700) can adjust the light conditions of the experimental environment where the LED display board (600) is located by adjusting the illuminance and the direction of the light source. For example, the lighting device (700) can artificially create strong backlight, low illuminance, daytime conditions, and nighttime conditions, and allows for testing how light at various angles affects traffic sign recognition.

[0083] FIG. 3 is a drawing of a front glass control device according to an embodiment of the present invention, and FIG. 4 is a drawing for explaining a front glass control device according to an embodiment of the present invention.

[0084] Referring to FIGS. 3 and FIGS. 4, a front windshield control device (300) according to an embodiment of the present invention can control the path of light and optical and geometric characteristics by fixing a transparent plate (310) having a curved surface like a real vehicle front windshield at various angles.

[0085] The fracture degree adjustment device may include a perforated plate (310), a frame structure (320), a perforated plate fixing part (330), and an inclination adjustment part (340).

[0086] The transparent plate (310) is made of a glass material having the same curvature and the same optical and geometric properties as the front windshield of an actual autonomous vehicle, and causes light refraction, which can reproduce a situation where traffic signs appear distorted or bent when viewed through the image capturing device (200).

[0087] The frame structure (320) is a structure for stably supporting and fixing the transparent plate (310) and may include a lower frame structure (320B) supported on the floor by having a fixing leg at the bottom, and an upper frame structure (320A) that supports the transparent plate (310) on the upper part of the lower frame structure (320B).

[0088] The frame structure (320) may be made of a material such as aluminum alloy, carbon steel, or high-strength synthetic resin, and may include a clamping device, fastening bolt, fixing screw, etc., to prevent the glass plate from shaking or detaching due to vibration or external force during the experiment. Additionally, a scale or marker may be provided on one side of the frame structure (320) to help the experimenter quantitatively set the angle or position of the transparent plate (310) or maintain the same conditions during repeated experiments.

[0089] A permeable plate fixing part (330) for fixing a permeable plate (310) may be disposed on the upper frame structure (320A). The permeable plate fixing part (330) may be installed on both edges of the upper frame structure (320A) to support both ends of the permeable plate (310).

[0090] The transparent plate fixing part (330) can be implemented as a clamp designed to be rotatable and may include a curvature-corresponding groove or an uneven structure at a position in contact with the transparent plate (310). Through this, the transparent plate fixing part (340) can prevent displacement from the transparent plate (310) and simultaneously ensure mechanical compatibility with the curved shape.

[0091] The upper frame structure (320A) can be connected to the lower frame structure (320B) through an inclination adjustment unit (340). The inclination adjustment unit (340) may include a sliding rail, a rotation axis, or a hinge structure, and thereby the inclination (θ) of the transparent plate (310) can be controlled by adjusting the inclination of the upper frame structure (320A).

[0092] Such a front windshield control device (300) can secure mechanical stability of a curved transparent plate (310) while simultaneously implementing image recognition performance tests according to refraction conditions with precision and reproducibility, and by ensuring diversity and repeatability of experimental conditions, it can provide an experimental environment optimized for evaluating image recognition performance of an autonomous vehicle.

[0093] FIG. 5 is a drawing of a collimator according to an embodiment of the present invention.

[0094] Referring to FIG. 5, the collimator (400) according to an embodiment of the present invention passes light projected from an LED display board (600) and can adjust the focal length by changing light conditions such as the light path, distortion, magnification / reduction, and focusing distance depending on the type of optical lens inserted.

[0095] The collimator (400) may include a lens disk (410) having a plurality of first openings (OH1) arranged along the circumferential direction and a support member (420) supporting the lens disk, and a specific optical transmission characteristic can be selected by rotating the lens disk (410) to align the desired first openings (OH1) with the first central hole (CH1).

[0096] The lens disc (410) is installed to be rotatable about a central axis, allowing for the selective insertion and switching of multiple optical lenses. The lens disc (410) is rotatably fixed to a support member (420) and can be rotated manually or by a motor drive.

[0097] The first opening (OH1) of the lens disc (410) is a hole structure for inserting a lens, and an optical lens having desired optical characteristics such as refractive index, curvature, chromatic aberration correction, and field of view can be inserted.

[0098] According to an embodiment, optical lenses having different optical characteristics are inserted into each aperture (OH1), and these can be selectively placed in the shooting path via a lens disc (410) depending on the experimental purpose. For example, a convex lens (for light gathering), a concave lens (for simulating divergent light), an aspherical lens (for distortion correction), etc., may be inserted into each aperture, and one aperture may be left open without a lens inserted to provide a lens-free condition (reference value experiment).

[0099] The central hole (CH1) is located in the support member (420) and is formed along the reference alignment axis to guide the optical axis of the image capturing device (200) and the optical lens to be aligned.

[0100] The collimator (400) according to an embodiment of the present invention enables an experimenter to perform repeated experiments while using the same camera and signboard images, while precisely changing optical conditions by only selecting an optical lens.

[0101] FIG. 6 is a drawing of an environment control device according to an embodiment of the present invention.

[0102] Referring to FIG. 6, an environment control device (500) according to an embodiment of the present invention is installed on the optical path of a traffic sign recognition test system (10), and by inserting an environment filter into the optical path projected from an LED display board (600), it is possible to quantitatively and repeatedly test recognition performance under various environmental conditions.

[0103] The environment control device (500) may include a filter disk (510) having a plurality of second openings (OH2) arranged along the circumferential direction and a support frame (520) supporting the filter disk, and a specific environmental characteristic can be selected by rotating the filter disk (510) to align the desired second openings (OH2) with the second central hole (CH2).

[0104] The filter disk (510) is installed to be rotatable about a central axis, allowing for selective insertion and switching of environmental filters. The filter disk (510) is rotatably fixed to a support frame (520) and can be rotated manually or by motor drive.

[0105] The second opening (OH2) of the filter disk (510) may have an environment filter with different optical properties inserted therein. The environment filter is an optical filter specially designed to reflect conditions of reduced visibility that may occur in actual road environments, and can be used to evaluate how accurately an image recognition algorithm of an autonomous vehicle can recognize traffic signs even under extreme conditions.

[0106] Depending on the experimental purpose, the environmental filter may include a light reduction filter, a FOG filter, a contamination filter, and a composite filter. Here, the light reduction filter is a filter for simulating cloudy weather or nighttime environments by reducing illumination conditions; the FOG filter is a filter for simulating diffuse reflection and haze phenomena when fog occurs; the contamination filter is a filter for simulating reduced visibility caused by dust, dirt, oil film, etc.; and the composite filter may be a filter for simulating the combined occurrence of reduced visibility caused by dust, dirt, oil film, etc., along with diffuse reflection and haze phenomena when fog occurs.

[0107] The filter disk (510) is mounted on a support frame (520) so as to be rotatable along a central axis, and can position a specific second opening (OH2) on the optical path by overlapping it with the second central hole (CH2).

[0108] FIG. 7 is a drawing for explaining the operation of an LED display board according to an embodiment of the present invention.

[0109] Referring to FIG. 7, the LED display board (600) is a simulation device for testing the recognition of traffic signs, and displays an image of a traffic sign using an LED light source. By programmatically changing the display content or controlling the lighting method, scanning cycle, flashing frequency, etc., it can be provided to evaluate the recognition algorithm of an autonomous vehicle under various test conditions.

[0110] The LED display (600) includes LED elements arranged in a grid shape and can be designed to enable individual operation at the pixel level. The LED display (600) outputs an image generated through an LED matrix on the front and can display various traffic sign shapes such as stop, speed limit, and no turning.

[0111] The LED display (600) may include a PWM (Pulse Width Modulation)-based scanning signal control circuit to allow for fine setting of test conditions. The LED display (600) can control the brightness, frequency, and synchronization status of visual information input to the video recording device (200) by changing the flashing cycle, duty cycle, and phase of the LEDs in the same traffic sign shape. This can be utilized to test for moiré phenomena, non-illumination recognition errors, pixel loss recognition problems, etc., by intentionally matching or misaligning the scanning frequency of the LED display (600) when the video recording device (200) is operating at a specific frame rate.

[0112] The LED display board (600) is linked with the image analysis device (100) and can change the numbers, symbols, characters, etc. of the traffic sign in response to the control signal received from the image analysis device (100).

[0113] FIG. 8 is a schematic block diagram of an image analysis device according to an embodiment of the present invention.

[0114] Referring to FIG. 8, an image analysis device (100) according to an embodiment of the present invention includes a processor (110), a memory (120), a communication interface (130), and a storage (140).

[0115] The processor (110) controls the overall operation of each component of the image analysis device (100). The processor (110) may be configured to include a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphic Processing Unit), or any type of processor (110) well known in the art of the present invention.

[0116] Additionally, the processor (110) can perform operations for at least one application or program for executing the method according to embodiments of the present invention, and the server (100) may have one or more processors (110).

[0117] According to an embodiment, the processor (110) may further include RAM (Random Access Memory, not shown) and ROM (Read-Only Memory, not shown) for temporarily and / or permanently storing signals (or data) processed within the processor (110). Additionally, the processor (110) may be implemented in the form of a system-on-chip (SoC) comprising at least one of a graphics processing unit, RAM, and ROM.

[0118] Memory (120) stores various data, commands, or information. Memory (120) may load a program from storage (140) to execute a method according to various embodiments of the present invention. When a computer program is loaded into memory (120), the processor (110) may perform the method by executing one or more instructions that constitute the computer program. Memory (120) may be implemented as volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

[0119] The communication interface (130) supports wired or wireless communication of the image analysis device (100). Additionally, the communication interface (130) may support various communication methods other than internet communication. To this end, the communication interface (130) may be configured to include a communication module well known in the art of the present invention. In some embodiments, the communication interface (130) may be omitted.

[0120] Storage (140) can store computer programs non-temporarily. Storage (140) can store various information required for creation and processing during process execution.

[0121] The storage (140) may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, or any form of computer-readable recording medium well known in the art to which the present invention belongs.

[0122] The bus (150) provides communication functions between components of the management server (100). The bus (150) can be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

[0123] FIG. 9 is a flowchart illustrating a traffic sign recognition test method according to an embodiment of the present invention.

[0124] Referring to FIG. 9, the LED display (600) can digitally display traffic signs (S100).

[0125] The lighting device (700) can illuminate the LED display board (600) by adjusting the illuminance and the direction of the light source (S110).

[0126] The environment control device (500) can control the transmittance of light emitted from the LED display panel (600) using an environment filter to simulate an external environment (S120).

[0127] The front glass control device (300) can artificially generate optical distortion by passing light that has passed through the environment filter of the environment control device (500) through a transparent plate (S130).

[0128] The collimator (400) can adjust the focal length of the light transmitted through the transparent plate of the front glass control device (300) using an optical lens (S140).

[0129] The image capturing device (200) can generate a captured image of the LED display board (600) using light transmitted through an optical lens (S150).

[0130] The image analysis device (100) can determine the success rate of recognizing traffic signs by analyzing the captured image (S160). At this time, the image analysis device (100) can perform preprocessing of the captured image through a previously trained artificial intelligence model, recognize the traffic sign based on object detection technology, determine the country in which the recognized traffic sign is used using previously stored traffic sign data, and determine the false positive rate and false negative rate by comparing the recognized traffic sign with actual sign information.

[0131] Additionally, the image analysis device (100) can store a test data log containing the results of performance judgment under implemented test conditions and convert the test data log into a training dataset for the artificial intelligence model.

[0132] Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention may be implemented in other specific forms without altering its technical concept or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive.

Claims

1. An LED display board that uses PWM (Pulse Width Modulation) to variably display the content of traffic signs; A lighting device that adjusts the lighting conditions of an experimental environment where the LED display is located by adjusting the illuminance and the direction of the light source; An environment control device that controls the transmittance of light emitted from an LED display panel using an environment filter to simulate an external environment; A front windshield control device that artificially generates optical distortion for light transmitted through the above-mentioned environment control device; A collimator that adjusts the focal length of light transmitted through the above-mentioned front glass control device; An image capturing device that generates a captured image of the LED display using light transmitted through the collimator; and A traffic sign recognition test system comprising an image analysis device that analyzes a captured image received from the above image capturing device and determines the success rate of recognizing the traffic sign.

2. In Paragraph 1, The above-mentioned front windshield control device is, A transparent plate provided in a glass material having the same curvature and optical and geometric characteristics as the front windshield of a vehicle, which transmits light transmitted through the environment control device; A frame structure comprising a lower frame structure supported on the floor by having a fixing leg at the bottom, and an upper frame structure supporting the transparent plate at the top of the lower frame structure; A transparent plate fixing part disposed at both edges of the upper frame structure to support both ends of the transparent plate and provided as a rotatable clamp; and A traffic sign recognition test system comprising a slope adjustment unit that connects the upper frame structure and the lower frame structure and adjusts the slope of the upper frame structure with respect to the ground.

3. In Paragraph 1, The above collimator is, A lens disc having a plurality of first openings arranged along the circumferential direction; and Supporting the above lens disc, and including a support member having a first central hole formed therein, An optical lens having different optical characteristics is inserted into the first opening, and A traffic sign recognition test system that selects optical transmission characteristics by aligning the first opening with the first central hole through the rotation of the lens disc.

4. In Paragraph 1, The above-mentioned environment control device is, A filter disk having a plurality of second openings arranged along the circumferential direction; and It supports the filter disk and includes a support frame having a second central hole formed therein, The environment filter having different optical characteristics is inserted into the second opening, and A traffic sign recognition test system that selects environmental characteristics by aligning the second opening with the second central hole through the rotation of the filter disk.

5. In Paragraph 1, The above LED display is, A traffic sign recognition test system comprising LED elements arranged in a grid shape, controlling individual driving at the pixel level, and controlling the display of the traffic sign through PWM (Pulse Width Modulation)-based scanning signal control.

6. In Paragraph 1, The above image analysis device is, A traffic sign recognition test system that performs preprocessing of the above-mentioned captured video through a previously trained artificial intelligence model, recognizes the above-mentioned traffic sign based on object detection technology, determines the country in which the recognized traffic sign is used using previously stored traffic sign data, and determines the false positive rate and false negative rate by comparing the recognized traffic sign with actual sign information.

7. In Paragraph 6, The above image analysis device is, A traffic sign recognition test system that stores a test data log containing performance judgment results under implemented test conditions and converts the test data log into a training dataset for the artificial intelligence model.

8. Step of displaying traffic signs by LED display board; A step of illuminating the LED display board by adjusting the illuminance and the direction of the light source; A step of controlling the transmittance of light emitted from the LED display panel using an environment filter to simulate an external environment; A step of artificially generating optical distortion through a transparent plate for light transmitted through the above-mentioned environment filter; A step of adjusting the focal length of the light transmitted through the above-mentioned plate using an optical lens; A step of generating a captured image of the LED display using light transmitted through the optical lens; and A traffic sign recognition test method comprising the step of analyzing the above-described captured image to determine the success rate of recognizing the traffic sign.

9. In Paragraph 8, The step of determining the above recognition success rate is, A step of performing preprocessing work on the above-mentioned captured image using a previously trained artificial intelligence model; A step of recognizing the above traffic sign based on object detection technology; A step of determining the country in which a recognized traffic sign is used using previously stored traffic sign data; and A traffic sign recognition test system comprising a step of determining the false positive rate and false negative rate by comparing recognized traffic sign information with actual sign information.

10. In Paragraph 8, A step of storing a test data log including a performance judgment result under implemented test conditions; and A method for testing traffic sign recognition that further includes the step of converting the above test data log into a training dataset for an artificial intelligence model.

Images