Urban road cantilever type signboard reflective film retroreflection coefficient detection method

Abstract

The invention discloses an urban road cantilever type signboard reflective film retroreflection coefficient detection method, which comprises an unmanned aerial vehicle, an optical detection system and an embedded data processing system, and is characterized in that the optical detection system comprises a light emitting unit, a light receiving unit, a laser ranging unit and a temperature, humidity and illumination sensor unit. By arranging the unmanned aerial vehicle system, the problems that a traditional detection method depends on manual close-range detection and is low in efficiency, high in risk and poor in economical efficiency are solved; the detection precision is improved through matching of light sources with different wave bands and reflective films with different colors; through long-distance detection of the unmanned aerial vehicle platform, the effective area of the detection signboard can be increased, the detection area is expanded from points to surfaces, detection data are more representative, and repeated detection can be realized according to images; the light intensity sensor, the temperature sensor and the humidity sensor are arranged to collect real-time data during detection, and environment compensation is performed on an algorithm, so that the detection result is more accurate and more conforms to the intelligent traffic development trend.

Description

Technical Field

[0001] The present invention relates to the technical field of municipal inspection, and in particular to a method for detecting the retroreflective coefficient of the reflective film of a cantilever signboard on an urban road. Background Art

[0002] The retroreflective coefficient is the ratio of the luminous intensity coefficient to the surface area of the retroreflector, expressed in candelas per lux per square meter (cd·lx-1·m -2 ), and the level of the retroreflective coefficient directly determines the visibility of the signboard. The national standard GB / T18833-2012 puts forward the minimum requirements for the retroreflective coefficients of various reflective films.

[0003] A higher retroreflective coefficient can make the signboard more prominent, shorten the reaction time of road users, and reduce the occurrence of traffic accidents.

[0004] For the cantilever signboards on urban roads, the general height is more than 2.5 meters. The existing detection method is that the tester uses a handheld retroreflective coefficient detector for detection. Its principle is to calibrate the illuminance of the standard plate before the test, and then detect the signboard. The method is to hold the handheld retroreflective coefficient detector and let the end of the instrument contact the signboard for calibration with a certain pressure, and then rent an aerial work vehicle or a climbing vehicle to climb to the front of the signboard, hold the instrument and approach the signboard with a certain pressure for single-point detection, detect 3 points, and take the average value as the detection result of this signboard for quality evaluation. First of all, this traditional detection method has extremely low detection efficiency, poor economy and affects the normal traffic, and there are great potential safety hazards; secondly, this method is single-point detection, with poor repeatability and reproducibility, and the tester needs to hold and approach for detection, without considering the influence of human factors; thirdly, the algorithm of the traditional detection method is relatively simple, only considering the comparison between the standard plate and the detection plate, without considering the influence of ambient light and atmospheric environment; finally, the traditional detection method uses a single light source to detect the retroreflective coefficients of different color reflective films, without considering the applicability of different color reflective films to the light source.

[0005] Therefore, there is an urgent need to invent an efficient, safe and automated detection method for the retroreflective coefficient of sign reflective films. In view of the above problems, the present invention proposes a detection method for the retroreflective coefficient of cantilevered sign reflective films on urban roads. By setting up a drone system, it solves the problems of traditional detection methods that rely on manual close-range detection, such as low efficiency, high danger, poor economy, and the error impact caused by manual operation. By setting up a multi-spectral automatic lighting module, it realizes the matching of light sources in different bands and reflective films of different colors, improving the detection accuracy. Through the long-distance detection of the drone platform, it can also increase the effective area of the detected sign, expand the detection area from a point to a surface, make the detection data more representative, and enable repeated detection based on images. By setting up light intensity sensors and temperature and humidity sensors to collect real-time data during detection and performing environmental compensation on the algorithm, the data is made more accurate. Summary of the Invention

[0006] In view of the above problems, the present invention provides an automated long-distance detection method for the retroreflective coefficient of cantilevered sign reflective films on urban roads based on a drone platform, which can realize synchronous automated detection of the retroreflective coefficients of reflective films at long distances and in multiple colors, significantly improving the detection efficiency and safety, and conforming to the development trend of intelligent transportation.

[0007] The technical solution adopted by the present invention is as follows: A detection method for the retroreflective coefficient of cantilevered sign reflective films on urban roads includes the following steps:

[0008] S1. For the target sign, the detector operates the touch screen terminal to control the drone to take off, searches for the target sign through the FPV camera, and hovers at a certain distance directly in front of the sign;

[0009] S2. For the target sign, select the light source and the detection area on the touch screen, and use the optical detection system to take a picture of the target sign image, which is pre-stored in the embedded data processing system;

[0010] S3. When taking pictures, synchronously trigger the laser rangefinder to measure the distance, the temperature and humidity sensor to measure the temperature and humidity, and the light sensor to measure the current ambient light intensity, and input the data into the embedded processing system;

[0011] S4. The photo is pre-stored in the embedded data processing system. Analyze the grayscale value of the image and the effective area of different color regions of the image, and use the pre-stored database and model in the processing system to calculate the retroreflective coefficients of different color regions of the standard sign reflective film, and output them.

[0012] S5. Display the retroreflective coefficients of different color reflective films of the sign on the touch screen, and store the result information in the memory database.

[0013] Further, in S1, in order to achieve long-distance detection and solve the problems of traditional detection methods that rely on manual close-range detection, such as low efficiency, high risk, poor economy, and the error impact caused by manual operation. The drone used is a DJI quadcopter drone, equipped with an FPV camera for searching for targets; equipped with a 7-inch touch screen terminal with a hovering anti-shake function, and the flight system is independently powered; the drone is provided with a three-axis stabilized mechanical gimbal for stabilizing the optical detection system; a carbon fiber extension bracket is provided for fixing a laser range finder sensor, temperature and humidity sensors, and a light sensor; the target sign is a cantilever sign on an urban road.

[0014] Further, the detector controls the drone to hover at different distance positions such as 5m or 10m directly in front of the target sign through the 7-inch touch screen terminal and then conducts subsequent detection.

[0015] Further, in S2, the optical detection system includes a light emitting unit and a reflected light receiving unit; the light emitting unit includes a light source (including a narrowband filter), a collimating lens, and an APD monitor (including a beam splitter).

[0016] In order to achieve the matching of light sources in different bands and reflective films of different colors and improve the detection accuracy; the light source (including a narrowband filter) adopts a multi-spectral active lighting module, using an LED light source array. For the sensitive bands of reflective films of different colors, spectral light sources of different colors are configured to eliminate problems such as color confusion and inaccurate reflectivity measurement in traditional white light detection and improve the measurement accuracy of the retroreflective coefficient; the narrowband filter can selectively filter out light in certain wavelength ranges and retain light in a specific wavelength range.

[0017] The collimating lens is placed in front of the light source, and the light emitted by the light source becomes parallel through the refraction of the lens.

[0018] The APD monitor (including a beam splitter) is placed parallel to the collimating lens in front of the light source, and a beam splitter (allowing 90% of the light to pass through and 10% of the light to be reflected) is provided in front of the collimating lens for real-time monitoring of the stability of the reflected light.

[0019] After passing through specific band filtering, the light source passes through the collimating lens and irradiates the sign at a fixed angle.

[0020] Further, in order to increase the effective area of the detected sign, expand the detection area from a point to a surface, make the detection data more representative, and enable repeated detection based on the image, by selecting the area of the detection area on the touch screen terminal of the drone, the detection area is each color area on the target sign displayed on the touch screen terminal, with a dotted line indicating the detection range and can be manually adjusted; after the area is determined, each color within the detection area range is highlighted when irradiated by the light source.

[0021] The camera belongs to the reflected light receiving unit, which includes a light shield, a polarization filter, and a high-resolution camera. The angle between the camera axis and the normal axis of the signboard is a fixed value.

[0022] The light shield is set with a certain size and placed in front of the camera lens to reduce the influence of ambient light.

[0023] The polarization filter is placed in front of the lens, which can eliminate the reflected light from the non-metal surface emitters around the signboard, thereby ensuring the imaging quality.

[0024] Furthermore, for the detection of the retroreflective coefficient of signs with different colors, after selecting a light source with a specific spectrum, it is irradiated onto the target signboard through a collimating lens and an APD detector (including a beam splitter); the effective detection area is selected through the touch screen controller terminal, and after determining the effective area, a high-resolution camera is used to take pictures. The photo quality is ensured by the light shield and the polarization filter; finally, the photos are pre-stored in the embedded data processing system.

[0025] Furthermore, in S3, in order to eliminate the influence of ambient light and the atmospheric environment, a light intensity sensor and temperature and humidity sensors are set to collect real-time data during detection, and environmental compensation is performed on the algorithm. The laser ranging sensor is an integrated TOF type laser ranging sensor, fixed to the bottom of the drone, used to measure the distance between the detection system and the target signboard in real time, and transmit the data to the embedded data processing system through the serial port.

[0026] The temperature and humidity sensors are fixed on the carbon fiber extension bracket at the bottom of the drone, used to detect the environmental temperature (T) and humidity (H) data, and transmit the data to the embedded data processing system through the serial port for atmospheric attenuation correction of the image.

[0027] The light sensor is fixed on the carbon fiber extension bracket at the bottom of the drone, used to record the reflected light intensity (E K ) and ambient light intensity (E G ), and transmit the data to the embedded data processing system through the serial port for the correction calculation of the retroreflective coefficient.

[0028] Furthermore, in S4, the embedded data processing system includes hardware and software.

[0029] The hardware includes a Raspberry Pi chip and an accelerator for accelerating calculations, installed in a waterproof box, fixed on the top of the drone, powered by an independent 12V lithium battery through a DC-DC module for the calculation unit of the processing system; the hardware also includes a built-in micro standard plate for the retroreflective coefficient of reflective films of different colors certified by the authoritative agency NIST, and the calibrated value of the retroreflective coefficient of the standard plate is Rs.

[0030] The software includes the YOLOv5s model, the adaptive ROI algorithm, and the HSV color space model. The YOLOv5s model has been trained through deep learning, including collecting 500 - 1000 images of municipal road signs, using the HSV color space model to achieve intelligent segmentation of the reflective film area, for the purpose of different color recognition and segmentation, using the adaptive ROI algorithm to achieve the extraction of each color area and the calculation and confirmation of the corresponding area (A) of each color. By developing a Python script, calling OpenCV to package the trained YOLOv5s model algorithm into a Docker image, and writing it to the Raspberry Pi SD card to calculate the effective average gray value (I) of the image.

[0031] Furthermore, the software calculation program also includes a calculation program for the retroreflective coefficient (R) of the sign reflective film. The following is the calculation formula for the retroreflective coefficient of the reflective film:

[0032]

[0033] C - is the atmospheric transmittance (atmospheric attenuation correction) calculated based on the values of the temperature and humidity sensors, and its formula is:

[0034] C = e -(0.0003H+0.0012T) ×L

[0035] H - is the humidity percentage (%) recorded by the humidity sensor at the detection time;

[0036] T - is the temperature (°C) recorded by the temperature sensor at the detection time;

[0037] L - is the distance (m) between the detection system and the target sign recorded by the laser rangefinder at the detection time;

[0038] A - is the effective area (m 2 ) of different color areas on the collected picture;

[0039] K - is the calibration coefficient for the on-site detection of the retroreflective coefficient, which is the ratio of the known reflection coefficient of the standard plate to the measured signal intensity, and it converts the original light intensity signal measured by the optical system into a standardized retroreflective coefficient value. Its physical meaning is the retroreflective coefficient value corresponding to the unit light intensity signal, and it is a dimensionless proportional factor. Its calculation formula is as follows:

[0040]

[0041] Rs - is the calibration value of the retroreflective coefficient of the certified micro standard plate;

[0042] E K - When the light source is turned on, collect the reflected light intensity of the micro standard plate;

[0043] E G- When the light source is turned off, collect the ambient light intensity of the micro standard board.

[0044] Further, in S5, after image analysis and processing, the retroreflective coefficient values of the retroreflective films of different colors of the target signboard calculated by the processing program will be displayed on the 7-inch high-definition touch screen. According to the minimum requirements of various retroreflective films provided by GB / T 18833-2012, determine whether the test data is qualified, and store the test result information in the Sqlite embedded database.

[0045] Compared with the prior art, the beneficial effects of the present invention are:

[0046] For the detection of the retroreflective coefficient of different colors of the signboard, after selecting a light source with a specific spectrum, it is irradiated onto the target signboard through a collimating lens and an APD detector; select the effective detection area through the touch screen controller terminal, and after confirmation, take a photo with a high-resolution camera. The photo quality is guaranteed by the light shield and the polarization filter; finally, the photo is pre-stored in the embedded data processing system, and through the analysis of the average gray value and the calculation of the retroreflective coefficient, the retroreflective coefficient values of different colors of the signboard retroreflective film are output.

[0047] The present invention solves the problems of low efficiency, high danger, poor economy, etc. and the error influence caused by manual operation in the traditional detection method that relies on manual close-range detection by setting up a drone system.

[0048] By setting up a multi-spectral automatic lighting module, the matching of light sources in different bands and retroreflective films of different colors is realized, and the detection accuracy is improved.

[0049] Through the long-distance detection of the drone platform, the effective area of the detected signboard can also be increased, the detection area is expanded from a point to a surface, the detection data is more representative, and repeated detection can be realized according to the image.

[0050] By setting up a light intensity sensor and temperature and humidity sensors to collect real-time data during detection, environmental compensation is carried out on the algorithm, so that the data is more accurate. The present invention better conforms to the development trend of intelligent transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of the present invention.

[0052] Figure 2 It is a schematic diagram of the optical detection system of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0053] For a better understanding of the present invention, the present invention will be further described below in conjunction with the accompanying drawings in the embodiments of the present invention, but it is not a limitation of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope of the present invention.

[0054] A method for detecting the retroreflective coefficient of the reflective film of a cantilever signboard on an urban road specifically includes the steps:

[0055] In S1, the unmanned aerial vehicle is a DJI Mavic 3 Enterprise Edition quadcopter unmanned aerial vehicle, equipped with an FPV camera for searching and determining the target signboard; equipped with a 7-inch touch screen terminal, having a hover anti-shake function, and the flight system is independently powered; the unmanned aerial vehicle is provided with a three-axis stabilized mechanical gimbal with a jitter accuracy of ±0.01° for stabilizing the optical detection system; a carbon fiber extension bracket is provided for fixing the TOF laser rangefinder, DS181320 temperature sensor, AM2320 humidity sensor, and BH1750 light sensor. The tester controls the DJI Mavic 3 Enterprise Edition quadcopter unmanned aerial vehicle to hover 5 m directly in front of the target signboard through the 7-inch touch screen terminal. The target signboard is a cantilever signboard on an urban road, the height of the signboard is 3.0 m, the reflective film of the signboard is a Class II reflective film, and the colors of the reflective film are white, red, and blue.

[0056] In S2, the optical detection system includes a light emitting unit and a reflected light receiving unit; the light emitting unit includes a light source (including a narrowband filter), a collimating lens, and an APD monitor (including a beam splitter);

[0057] The light source (including a narrowband filter) is an NBET spectral active illumination module, which adopts an LED light source array. Different color spectral light sources are configured according to the sensitive bands of different color reflective films to eliminate problems such as color confusion and inaccurate reflectance measurement in traditional white light detection, and improve the measurement accuracy of the retroreflective coefficient; the narrowband filter can selectively filter out light in certain wavelength ranges and retain light in a specific wavelength range. For the colors of the reflective film of the signboard being white, red, and blue, select the light sources corresponding to different reflective film colors on the touch screen terminal: for white, enable 450 - 650 nm broadband white light; for red, enable 620 - 680 nm red light + 650 nm long-pass filter; for blue, enable 400 - 480 nm blue light + 500 nm short-pass filter. For the light emitted by different color light sources, use 74-VISThe collimating lens makes it parallel and ensures that the incident angle is -4° (i.e., the angle between the parallel light and the normal axis of the signboard). At the same time, the stability of the light is monitored in real time through the APD monitor (including the beam splitter); when the light shines on the target signboard, the effective area on the target signboard will be displayed on the 7-inch touch screen terminal. The effective area is within the dotted line frame, and the selection line frame can be manually stretched to expand or shrink the detection range. When each color within the effective area is illuminated by the light source, it will be highlighted on the touch screen terminal. The camera uses a Specim IQ multi-spectral high-resolution camera to take pictures of the target effective area. The reflection angle, that is, the angle between the camera axis and the normal axis of the signboard, is 0.2°. A 15-cm-long plastic light-shielding cover and a PBP01 type polarization filter are set to remove the influence of the surrounding ambient light and ensure the imaging quality.

[0058] In S3, the laser range finder sensor is an integrated TOF type laser range finder sensor, fixed to the bottom of the drone, used to measure the distance between the detection system and the target signboard in real time as L = 5.0 meters, and transmit the data to the embedded data processing system through the serial port. The DS181320 temperature sensor measures the ambient temperature as 25°C; the AM2320 humidity sensor measures the ambient humidity as 56%; the BH1750 light sensor measures the reflected light intensity of the micro standard board when the light source is turned on, which is 4500 lx for white, 2500 lx for red, and 950 lx for blue, and the ambient light intensity of the micro standard board when the light source is turned off, which is 1500 lx for white, 750 lx for red, and 200 lx for blue. The above data is transmitted to the embedded data processing system through the serial port for calculating the atmospheric attenuation correction of the image and the calibration coefficient of the retroreflective coefficient.

[0059] In S4, the embedded data processing system includes hardware and software. The hardware includes a Raspberry Pi CM4 chip and a Google Coral USB accelerator, installed in a waterproof box, fixed to the top of the drone, configured with an independent 12V lithium battery, and powered for the processing system calculation unit through a DC-DC module; the hardware also includes a built-in Labsphere retroreflective coefficient micro standard board for reflective film certified by the authoritative institution NIST, and the calibrated values of the retroreflective coefficients of its white, red, and blue standard boards are 600 cd·lx-1·m -2 、350 cd·lx-1·m -2 、150 cd·lx-1·m -2 .

[0060] Calculate the atmospheric transmittance according to the formula H = 56%, T = 25°C, L = 5 meters

[0061] C = e -(0.0003H+0.0012T)×L

[0062] = e -(0.0003*0.56+0.0012*25)×5.0= 0.86

[0063] Calculate the calibration coefficient K of the retroreflective coefficient detected on-site according to the formula, which is the ratio of the known reflectivity of the standard plate to the measured signal intensity. It converts the original light intensity signal measured by the optical system into a standardized retroreflective coefficient value. Its physical meaning is the retroreflective coefficient value corresponding to the unit light intensity signal, and it is a dimensionless scale factor. The calculation formula is as follows:

[0064]

[0065] Calculate the calibration coefficient K values of the white light source, red light source, and blue light source respectively, as follows:

[0066]

[0067] The software includes the YOLOv5s model, the adaptive ROI algorithm, and the HSV color space model. The YOLOv5s model has been trained through deep learning, including collecting 500 - 1000 images of municipal road signs, using the HSV color space model to achieve intelligent segmentation of the reflective film area, for the purpose of different color recognition and segmentation, and using the adaptive ROI algorithm to achieve the extraction of each color area and the calculation of the corresponding area of each color.

[0068] The areas of each color in the effective area of the sign determined by the embodiment are 0.25 m² for white 2 , 0.2 m² for red 2 , and 0.55 m² for blue 2 ; By developing a Python script, calling OpenCV to package the YOLOv5s model algorithm after training into a Docker image and writing it to the Raspberry Pi SD card, the effective average gray values of the images are calculated as 7.7 cd·lx⁻¹·m² for white -2 , 1.5 cd·lx⁻¹·m² for red -2 , and 1.9 cd·lx⁻¹·m² for blue -2 . Finally, the software calculation program also includes a program for calculating the retroreflective coefficient (R) of the sign reflective film. The following is the calculation formula for the retroreflective coefficient of the reflective film:

[0069]

[0070] Calculate the retroreflective coefficients of the three-color reflective films of the sign respectively as follows:

[0071]

[0072] In S5, after image analysis and processing, the reverse reflectance coefficient values of different color reflective films of the target signboard calculated by the processing program will be displayed on a 7-inch high-definition touch screen as follows: the reverse reflectance coefficient value of the white reflective film is 179 cd·lx-1·m -2 , which meets the minimum requirements of various reflective films provided by the standard GB / T 18833-2012, and the test data is determined to be qualified; the reverse reflectance coefficient value of the red reflective film is 43.6 cd·lx-1·m -2 , which meets the minimum requirements of various reflective films provided by the standard GB / T18833-2012; the reverse reflectance coefficient value of the blue reflective film is 20 cd·lx-1·m -2 , which meets the minimum requirements of Class II reflective films provided by the standard GB / T 18833-2012; the test data is determined to be qualified, and the test result information is stored in the Sqlite embedded database.

[0073] Finally, the inspectors use the traditional method to hold a handheld reverse reflectance coefficient detector to detect the reverse reflectance coefficient of the reflective film of the target signboard, which are 170 cd·lx-1·m for white -2 , 41 cd·lx-1·m for red -2 , and 19 cd·lx-1·m for blue -2 . The test values of the traditional test method are about 5% lower than those of the test method of the present invention, indicating that the test method of the present invention is more effective.

[0074] The above embodiments are only for illustrating the technical concept and characteristics of the present invention, and the purpose is to enable those who are familiar with this technology to understand the content of the present invention and implement it accordingly, and it cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit and essence of the present invention should be covered by the present invention.

Claims

1. A method for detecting the retroreflective coefficient of the reflective film of a cantilever sign on an urban road, characterized in that, It includes the following steps: S1. For the target signboard, the inspector operates the touch screen terminal to control the drone to take off, searches for the target signboard through the FPV camera, and hovers at a certain distance directly in front of the signboard; S2. For the target signboard, select the light source and the detection area on the touch screen, take photos using the optical detection system, and pre-store them in the embedded data processing system; S3. When taking photos, synchronously trigger the laser range finder to measure the distance, the temperature and humidity sensor to measure the temperature and humidity, and the light intensity sensor to measure the current ambient light intensity, realize environmental compensation and input the data into the embedded processing system; S4. The photos are pre-stored in the embedded data processing system. Analyze the gray value of the image and the effective area of different color regions of the image. Use the pre-stored database and model in the processing system to calculate the retroreflective coefficient of different color regions of the standard sign reflective film, and output it. S5. Display the retroreflective coefficient of different color reflective films of the signboard on the touch screen, and store the result information in the memory database.

2. The method for detecting the retroreflective coefficient of the reflective film of the cantilever signboard on the urban road according to claim 1, wherein, In S1: The drone is a DJI quadcopter drone, equipped with an FPV camera for searching for targets; equipped with a 7-inch touch screen terminal, with a hovering anti-shake function, and the flight system is independently powered; The drone is provided with a three-axis stabilized mechanical gimbal for stabilizing the optical detection system; a carbon fiber extension bracket is provided for fixing the laser range finder, the temperature and humidity sensor, and the light intensity sensor; The target signboard is a cantilever signboard on an urban road; The inspector controls the drone to hover at different distances such as 5m or 10m directly in front of the target signboard through the 7-inch touch screen terminal and then conducts subsequent detections.

3. A method for detecting the retroreflective coefficient of the reflective film of a cantilever sign on an urban road according to claim 1, characterized in that In S2: In S2a1, the optical detection system includes a light emitting unit and a reflected light receiving unit; the light emitting unit includes a light source with a narrowband filter, a collimating lens, and an APD monitor (including a beam splitter); The light source is a multi-spectral active illumination module, using an LED light source array. For the sensitive bands of different color reflective films, spectral light sources of different colors are configured; the narrowband filter can selectively filter out light in certain wavelength ranges and retain light in specific wavelength ranges. The collimating lens is placed in front of the light source, and the light emitted by the light source becomes parallel through the refraction of the lens; The APD monitor (including a beam splitter) is placed in parallel with the collimating lens in front of the light source. A beam splitter (allowing 90% of the light to pass through and 10% of the light to be reflected) is provided in front of the collimating lens for real-time monitoring of the stability of the reflected light; After the light source is filtered through a specific band, it passes through the collimating lens and irradiates the signboard at a fixed angle; In S2a2, the detection area is each color area on the target signboard displayed on the touch screen terminal, with a dotted line indicating the detection range, which can be manually adjusted; each color area within the detection range is highlighted when irradiated by the light source; In S2a3, the camera belongs to the reflected light receiving unit, and the reflected light receiving unit includes a light shield, a polarization filter, and a high-resolution camera, The light shield is set in front of the camera lens; The polarization filter is placed in front of the lens to eliminate the reflected light from non-metallic surface emitters around the signboard.

4. A method for detecting the retroreflective coefficient of the reflective film of a cantilever signboard on an urban road according to claim 1, characterized in that, In S3, The laser ranging sensor is an integrated TOF laser ranging sensor, which is fixed to the bottom of the UAV and used to measure the distance between the detection system and the target sign in real time, and transmit the data to the embedded data processing system through the serial port; The temperature and humidity sensor is fixed on the carbon fiber extension bracket at the bottom of the UAV and used to detect the environmental temperature (T) and humidity (H) data, and transmit the data to the embedded data processing system through the serial port for atmospheric attenuation correction of the image; The light sensor is fixed on the carbon fiber extension bracket at the bottom of the UAV and used to record the environmental light intensity (E) at the detection moment, and transmit the data to the embedded data processing system. After data processing, it is used for the correction calculation of the retroreflective coefficient.

5. A method for detecting the retroreflective coefficient of the reflective film of a cantilever signboard on an urban road according to claim 1, characterized in that, In S4: The embedded data processing system includes hardware and software. The hardware includes a Raspberry Pi chip and an accelerator for accelerating calculations. It is installed in a waterproof box, fixed to the top of the UAV, and uses an independent 12V lithium battery to supply power to the processing system's computing unit through a DC-DC module; The hardware also includes a built-in micro standard plate of the retroreflective coefficient of reflective films of different colors certified by the authoritative institution NIST, and the calibrated value of the retroreflective coefficient of the standard plate is Rs; The software includes the YOLOv5s model with an adaptive ROI algorithm and an HSV color space model. The YOLOv5s model has been trained through deep learning, including collecting 500-1000 municipal road sign images, using the HSV color space model to achieve intelligent segmentation of the reflective film area for the purpose of different color recognition and segmentation, using the adaptive ROI algorithm to achieve the extraction of each color area and the calculation and confirmation of the corresponding area (A) of each color. By developing a Python script, calling OpenCV to package the trained YOLOv5s model algorithm into a Docker image, writing it to the Raspberry Pi SD card, and calculating the effective average gray value (I) of the image. The software calculation program also includes a calculation program for the retroreflective coefficient (R) of the sign reflective film. The following is the calculation formula for the retroreflective coefficient of the reflective film: C- is the atmospheric transmittance (atmospheric attenuation correction) calculated based on the values of the temperature and humidity sensors, and its formula is: C = e -(0.0003H+0.0012T) × L H- is the percentage of humidity recorded by the humidity sensor at the detection moment (%); T- is the temperature (°C) recorded by the temperature sensor at the detection moment; L- is the distance (m) between the detection system and the target sign recorded by the laser ranging sensor at the detection moment; A - Effective area of different color regions on the captured image (m 2 ); K- is the calibration coefficient for the on-site detection of the retroreflective coefficient, which is the ratio of the known reflection coefficient of the standard plate to the measured signal intensity. It is used to convert the original light intensity signal measured by the optical system into a standardized retroreflective coefficient value. Its physical meaning is the retroreflective coefficient value corresponding to the unit light intensity signal, and it is a dimensionless scaling factor. Its calculation formula is as follows: Rs- is the calibrated value of the retroreflective coefficient of the certified micro standard plate; E K - When the light source is turned on, collect the reflected light intensity of the micro standard board; E G - When the light source is turned off, collect the ambient light intensity of the micro standard board.

6. The method for detecting the retroreflective coefficient of the reflective film of a cantilever sign on an urban road according to claim 1, wherein, In S5, after image analysis and processing, the reverse reflectance coefficient values of different color reflective films of the target signboard calculated by the processing program will be displayed on a 7-inch high-definition touch screen. According to the minimum requirements of various reflective films provided by GB / T 18833-2012, it is determined whether the test data is qualified, and the test result information is stored in the Sqlite embedded database.

Images