Tunnel pavement illumination brightness detection method, device, system and storage medium

Abstract

The application discloses a tunnel pavement illumination brightness detection method, device and system and a storage medium, relates to the technical field of tunnel pavement brightness detection, and comprises the following steps: acquiring a dynamic photosensitive signal by means of a photosensitive sensor during detection of vehicle movement; restoring the dynamic photosensitive signal by means of a blind deconvolution algorithm to obtain a restored photosensitive signal; performing relative spectral correction and temperature drift compensation on the restored photosensitive signal to obtain a corrected photosensitive signal; performing end-to-end mapping on the corrected photosensitive signal by means of an end-to-end mapping model to obtain a brightness detection result of a tunnel detection pavement; and wherein the input of the end-to-end mapping model comprises the corrected photosensitive signal and a global calibration coefficient. The application improves the reliability of tunnel pavement brightness detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel pavement brightness detection, and in particular to a tunnel pavement lighting brightness detection method, device, system and storage medium. BACKGROUND

[0002] With the continuous expansion of highway tunnel construction scale, the tunnel lighting system plays a crucial role in ensuring driving safety, improving driving vision environment and reducing operation energy consumption. As the core index for evaluating the performance of the lighting system, the regular detection of tunnel lighting brightness has become an important work content in tunnel operation and management.

[0003] In related technologies, fixed measurement points are usually laid in the tunnel by using a handheld artificial brightness meter or illuminometer for sampling measurement, and an imaging measurement method based on vehicle-mounted camera image acquisition and inverse calculation through the gray value and brightness value mapping relationship. However, the manual detection has low efficiency and long operation period, and the illuminometer depends on the reflectivity of the tunnel inner wall and pavement, which is affected by factors such as material aging and pollution, making it difficult to accurately obtain, resulting in double errors of theory and engineering in the conversion from illuminance to brightness. The detection using the image processing method is easily affected by factors such as camera exposure parameters, lens dark angle and spectral response difference, and the image post-processing algorithm is complex, so the stability and traceability of the measurement results are poor. Therefore, there is an urgent need for a new tunnel lighting brightness detection technology that can realize efficient, continuous, direct and safe detection under normal driving conditions. SUMMARY

[0004] The main purpose of the present application is to provide a tunnel pavement lighting brightness detection method, device, system and storage medium, which aims to solve the technical problem of low reliability of related art tunnel pavement lighting brightness detection.

[0005] To achieve the above-mentioned purpose, the present application provides a tunnel pavement lighting brightness detection method, which comprises:

[0006] During the detection of the vehicle movement, a dynamic photosensitive signal is obtained by a photosensitive sensor;

[0007] The dynamic photosensitive signal is restored by a blind deconvolution algorithm to obtain a restored photosensitive signal;

[0008] The restored photosensitive signal is subjected to relative spectral correction and temperature drift compensation to obtain a corrected photosensitive signal;

[0009] The corrected photosensitive signal is subjected to end-to-end mapping by an end-to-end mapping model to obtain a brightness detection result of the tunnel detection pavement; wherein the input of the end-to-end mapping model comprises the corrected photosensitive signal and a global calibration coefficient.

[0010] In an embodiment, the step of restoring the dynamic photosensitive signal by a blind deconvolution algorithm includes:

[0011] constructing a convolution degradation model of the process of obtaining the dynamic photosensitive signal; wherein the input of the convolution degradation model includes the real light intensity corresponding to the detection position and the additive system noise, and the output of the convolution degradation model includes the dynamic photosensitive signal corresponding to the detection position;

[0012] by a blind deconvolution algorithm, taking the dynamic photosensitive signal as a known condition, deconvolving the convolution degradation model to calculate the real light intensity, and obtaining the restored photosensitive signal.

[0013] In an embodiment, the step of restoring the dynamic photosensitive signal by a blind deconvolution algorithm, taking the dynamic photosensitive signal as a known condition, deconvolving the convolution degradation model to calculate the real light intensity, and obtaining the restored photosensitive signal includes:

[0014] initializing the real light intensity and the blur kernel;

[0015] adopting an alternating minimization strategy to alternately estimate and update the real light intensity and the blur kernel until convergence, and obtaining the real light intensity;

[0016] wherein the update mode of the blur kernel is:

[0017] assuming that the current real light intensity is accurate, updating the blur kernel:

[0018]

[0019] wherein, denotes the updated blur kernel, denotes the current real light intensity, h denotes the current blur kernel, g denotes the dynamic photosensitive signal, and * denotes convolution, λ 1 denotes a regularization coefficient, Φ( h ) denotes the prior constraint of the kernel function.

[0020] the update mode of the real light intensity is:

[0021] assuming that the current blur kernel is accurate, updating the real light intensity:

[0022]

[0023] wherein, denotes the updated real light intensity, denotes the current blur kernel, g denotes the dynamic photosensitive signal, and * denotes convolution, λ 2 denotes a regularization coefficient, Ψ(f represents a prior constraint of the real light intensity.

[0024] In an embodiment, the step of obtaining the corrected photosensitive signal after the relative spectral correction and the temperature drift compensation of the recovered photosensitive signal comprises:

[0025] obtaining a discrete spectral response function of the photosensitive sensor to construct a response matrix;

[0026] performing matrix transformation on the spectral vector of the recovered dynamic photosensitive signal based on an inverse matrix of the response matrix to obtain a corrected spectral vector;

[0027] determining the photosensitive signal after the relative spectral correction based on the corrected spectral vector;

[0028] performing temperature drift compensation on the photosensitive signal after the relative spectral correction to obtain the corrected photosensitive signal.

[0029] In an embodiment, the step of obtaining the corrected photosensitive signal after the temperature drift compensation of the photosensitive signal after the relative spectral correction comprises:

[0030] constructing a temperature drift model to determine the relationship between the photosensitive signal measured at the current temperature and the photosensitive signal after the temperature compensation correction; the temperature drift model is represented as:

[0031]

[0032] wherein, S raw T represents the photosensitive signal measured at the temperature T, G T represents a gain proportional function, S ideal represents an ideal true value;

[0033] determining the gain proportional function based on the ratio of the photosensitive signal measured at the reference temperature to the photosensitive signal measured at the current temperature; the gain proportional function is represented as:

[0034]

[0035] wherein, S raw (T ref ) represents the photosensitive signal measured at the reference temperature T erf S raw T represents the photosensitive signal measured at the temperature T ; and ​​​​

[0036] Based on the temperature drift model, the temperature compensation is performed on the relative spectral corrected photosensitive signal to obtain a corrected photosensitive signal.

[0037] In an embodiment, the tunnel pavement is divided into three detection areas from left to right, and the three photosensitive sensors are used to collect the signals synchronously, and the end-to-end mapping model is used to perform end-to-end mapping on the corrected photosensitive signal to obtain the brightness detection result of the tunnel detection pavement.

[0038] Based on the vehicle positioning data and the vehicle speed information, the time series photosensitive data collected by the three photosensitive sensors is converted into three-channel spatial sequence photosensitive data.

[0039] The three-channel spatial sequence photosensitive data is subjected to transverse interpolation processing to fill in the blind area of perception.

[0040] The three-channel spatial sequence photosensitive data after interpolation processing is subjected to multi-channel consistency verification to generate a tunnel full-line brightness distribution curve.

[0041] In a second aspect, to achieve the above object, the present application further provides a tunnel pavement lighting brightness detection device, which comprises:

[0042] The acquisition module is configured to acquire a dynamic photosensitive signal through the photosensitive sensor during the movement of the detection vehicle.

[0043] The restoration module is configured to restore the dynamic photosensitive signal through a blind deconvolution algorithm to obtain a restored photosensitive signal.

[0044] The repair module is configured to perform relative spectral correction and temperature drift compensation on the restored photosensitive signal to obtain a corrected photosensitive signal.

[0045] The detection module is configured to perform end-to-end mapping on the corrected photosensitive signal through an end-to-end mapping model to obtain a brightness detection result of the tunnel detection pavement, wherein the input of the end-to-end mapping model comprises the corrected photosensitive signal and a global calibration coefficient.

[0046] In a third aspect, to achieve the above object, the present application further provides a tunnel pavement lighting brightness detection system, which comprises:

[0047] The data acquisition unit comprises three photosensitive sensors, a temperature sensor, an encoder circuit and a GNSS navigation module arranged in sequence, the photosensitive sensor is connected with a filter circuit, the photosensitive sensor is configured to convert a light signal into an electric signal, the filter circuit is configured to perform filter processing on the electric signal to obtain a dynamic photosensitive signal, and the encoder circuit is installed on a wheel of the detection vehicle and is configured to acquire vehicle speed information.

[0048] The lower-level processor is used to acquire dynamic photosensitive signals, vehicle positioning data, sensor temperature data and vehicle speed information from the data acquisition unit, organize and forward them to the upper-level processor.

[0049] The host computer processor is used to run computer programs, which are configured to implement the steps of the tunnel pavement lighting brightness detection method described above.

[0050] In one embodiment, the photosensor includes an optical lens, a filter, and a photoelectric conversion circuit.

[0051] Fourthly, to achieve the above objectives, this application further provides a storage medium, which is a computer-readable storage medium, and stores a computer program thereon. When the computer program is executed by a processor, it implements the steps of the above-described tunnel pavement lighting brightness detection method.

[0052] One or more technical solutions proposed in this application have at least the following technical effects:

[0053] This application improves the realism and stability of tunnel brightness detection under dynamic measurement conditions by acquiring dynamic photosensitive signals using a photosensitive sensor during normal vehicle operation and introducing a blind deconvolution algorithm to restore the signal degradation caused by vehicle movement, system response, and noise. By constructing the discrete spectral response function of the photosensitive sensor and performing relative spectral correction, the systematic color shift caused by differences in sensor spectral response is eliminated. At the same time, combined with a temperature drift compensation model, the influence of ambient temperature changes on the measurement results is suppressed, thereby significantly improving the accuracy and repeatability of brightness detection results. In addition, by using an end-to-end mapping model to directly map the corrected photosensitive signal to the road surface brightness value, the cumulative complexity of multiple models in traditional illuminance-brightness conversion is avoided, realizing efficient, continuous, and safe detection of tunnel road surface brightness under normal driving conditions. It has the technical effects of high detection efficiency, strong adaptability, and good engineering practicality. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0055] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0056] Figure 1 This is a flowchart illustrating an embodiment of the tunnel pavement lighting brightness detection method of this application.

[0057] Figure 2 The figure is a schematic diagram of the installation position of the photosensitive sensor of the present application.

[0058] Figure 3 The figure is a structural schematic diagram of the tunnel pavement illumination brightness detection system of the present application.

[0059] Figure 4 The figure is a structural schematic diagram of the tunnel pavement illumination brightness detection device of the present application.

[0060] The object realization, functional features and advantages of the present application will be further explained in combination with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0062] In order to better understand the technical solutions of the present application, the following will be described in detail in combination with the drawings of the specification and specific embodiments.

[0063] At present, whether the illuminance meter is used for measurement and conversion or the brightness meter is used for direct test, the following defects exist: for the illuminance conversion method, it depends on the reflectivity of the tunnel inner wall and pavement, and the value is affected by factors such as material aging and pollution, so it is difficult to accurately obtain, resulting in double errors of theory and engineering in the conversion from illuminance to brightness.

[0064] And the detection method using camera imaging needs to introduce an indirect conversion model to realize the conversion from image gray value to brightness value, which introduces additional error and uncertainty, and the stability and traceability of the measurement result are poor

[0065] Considering the above problems, the present embodiment provides a new solution, which discards all indirect measurement paths depending on complex intermediate links, adopts a "direct photosensitive-end-to-end mapping" brightness measurement method, does not need image processing algorithm for conversion, and does not need to convert illuminance, avoids two core error sources of reflectivity estimation and physical quantity conversion, and realizes direct from optical signal to brightness value.

[0066] The main solution of the embodiment of the application is: by using a photosensitive sensor to obtain a dynamic photosensitive signal in the process of detecting vehicle driving, a degradation model of the dynamic photosensitive signal is constructed and a blind deconvolution algorithm is used to restore the dynamic photosensitive signal, on this basis, by performing relative spectral correction and temperature drift compensation on the restored photosensitive signal, systematic errors caused by spectral response differences of the photosensitive sensor and environmental temperature changes are eliminated, finally, the corrected photosensitive signal is directly mapped to a tunnel road surface brightness detection result based on an end-to-end mapping model, so that continuous, stable and high-precision detection of tunnel lighting brightness under normal driving conditions is realized.

[0067] Specifically, the embodiment of the application provides a tunnel road surface lighting brightness detection method, referring to Figure 1 , Figure 1 FIG. 1 is a flowchart of a tunnel road surface lighting brightness detection method according to a first embodiment of the application.

[0068] In the embodiment, the tunnel road surface lighting brightness detection method comprises steps S10-S40:

[0069] Step S10: In the process of moving a detection vehicle, a dynamic photosensitive signal is obtained by using a photosensitive sensor.

[0070] It should be noted that the photosensitive sensor used in the embodiment is a brightness sensor whose physical spectral response matches the V(λ) function of the human eye, which directly perceives and measures the optical physical quantity of "brightness" from the physical layer, and eliminates the entire imaging link and complex post-processing algorithm, thereby eliminating the uncertainty caused by introducing an indirect conversion model from the root.

[0071] In a feasible implementation, the photosensitive sensor is composed of an optical lens, a filter and a photoelectric conversion circuit, and the dynamic photosensitive signal is an electrical signal. In addition, the photosensitive sensor is installed on the detection vehicle and collects the lighting brightness of different road surface positions in real time as the vehicle moves. In an example, referring to Figure 2 , the photosensitive sensor is installed at the front end of the vehicle, and in the process of vehicle driving, the vehicle speed can be set by software, the vehicle can cruise at the set speed, and the system can collect data at a fixed distance (generally 1 meter or 2 meters), for example, at a speed of 80 km / h, i.e. 22.22 m / s, 22.22 times of sampling per second can be performed to realize 1 meter of sampling data output per second, and 11.11 times of sampling per second can be performed to realize 2 meters of dynamic photosensitive signal sampling output per second.

[0072] Step S20: The dynamic photosensitive signal is restored by using a blind deconvolution algorithm to obtain a restored photosensitive signal.

[0073] Step S30: The restored photosensitive signal is subjected to relative spectral correction and temperature drift compensation to obtain a corrected photosensitive signal.

[0074] In step S40, the corrected photosensitive signal is subjected to end-to-end mapping through an end-to-end mapping model to obtain a brightness detection result of the tunnel detection pavement; wherein the input of the end-to-end mapping model includes the corrected photosensitive signal and the global calibration coefficient.

[0075] It should be noted that in a dynamic vehicle-mounted environment, direct mapping by the photosensitive sensor of the embodiment is faced with the problems of motion blur and environmental interference. To this end, the embodiment introduces blind deconvolution restoration, spectral correction and temperature drift compensation to solve the problems, and it is worth mentioning that it serves and strengthens the end-to-end mapping relationship, rather than an independent measurement path.

[0076] In a feasible implementation, step S20 includes steps A10-A20:

[0077] In step A10, a convolution degradation model of the process of acquiring the dynamic photosensitive signal is constructed; wherein the input of the convolution degradation model includes the real light intensity corresponding to the detection position and the additive system noise, and the output of the convolution degradation model includes the dynamic photosensitive signal corresponding to the detection position.

[0078] In step A20, the convolution degradation model is deconvolved and solved by taking the dynamic photosensitive signal as a known condition through a blind deconvolution algorithm to calculate the real light intensity and obtain the restored photosensitive signal.

[0079] Wherein, step A20 further includes steps A21-A22:

[0080] In step A21, the real light intensity and the blur kernel are initialized.

[0081] In step A22, an alternating minimization strategy is adopted to alternately estimate and update the real light intensity and the blur kernel until convergence is achieved to obtain the real light intensity.

[0082] Wherein, the update mode of the blur kernel is:

[0083] Assuming that the current real light intensity is accurate, the blur kernel is updated as:

[0084]

[0085] Wherein, denotes the updated blur kernel, denotes the current real light intensity, h denotes the current blur kernel, g denotes the dynamic photosensitive signal, and * denotes convolution. λ 1 denotes a regularization coefficient, and Φ( h ) denotes a prior constraint of the kernel function.

[0086] The updating method of the real light intensity is:

[0087] Assuming that the current blur kernel is accurate, the real light intensity is updated:

[0088]

[0089] wherein, represents the updated real light intensity, represents the current blur kernel, g represents the dynamic light signal, and * represents convolution, λ 2 represents a regularization coefficient, and Ψ( f ) represents a prior constraint of the real light intensity.

[0090] It can be understood that, in the vehicle-mounted dynamic measurement process, high-speed relative motion occurs between the optical lens and the tunnel road surface, which causes the light sensing sensor to collect not discrete geometric points during exposure, but spatial smearing formed along the motion trajectory, causing the light signal originally belonging to a single pixel to be diffused among multiple sensing units, thereby introducing motion blur.

[0091] From the physical mechanism, the optical lens is essentially a space-time integral of the incident light intensity within the exposure time. Vehicle motion causes the integral process to expand to a continuous path, resulting in distortion of the measurement results. In addition, inherent dark current noise, readout noise, and photon shot noise in the electronic system further exacerbate under high-speed dynamic acquisition conditions, and the system error significantly increases. Therefore, the improvement of dynamic measurement accuracy can be attributed to a signal restoration problem, i.e., reconstructing the real brightness distribution of the tunnel road surface from the degraded observation signal contaminated by motion blur and various noises.

[0092] To restore the signal, a convolution degradation model that restores the signal degradation process needs to be constructed first. The convolution degradation model constructed in the present embodiment can be represented as:

[0093]

[0094] wherein, L m x , y , t represent the dynamic light signal of the tunnel road surface at a detection position ( x , y ) collected by the light sensing sensor at time t, L ture x',y',t represent the real brightness of the tunnel road surface at a detection position ( x’ , y’ ) at time t, h(x-x',y-y',t) t ​​) is a point spread function, representing the spot distribution on the sensor after an infinitely small luminance point in the real scene passes through the optical lens, and in a dynamic case, the function is usually a linear motion stripe; represents the superimposed result of the light intensity received by a point (x, y) on the sensor after processing by the point spread function, n x,y,t represents the additive system noise.

[0095] After the convolution degradation model is constructed, the inverse process of the model, that is, the process of restoring the dynamic photosensitive signal, is performed. The core goal of the process is to find a restoration filter g x y such that:

[0096]

[0097] wherein, represents the true luminance of a detection position on the tunnel surface, g x y is the restoration filter, L m represents the dynamic photosensitive signal, and * represents the convolution operation.

[0098] As for how to find the restoration filter, the most direct restoration method in theory is inverse filtering, but this method will amplify the noise and mask the restored signal, and when the point spread function is unknown, it is difficult to accurately obtain the motion blur kernel, so it is usually difficult to implement in practical applications. Therefore, considering the complexity of the scene of the embodiment, and in the high-speed motion state of the vehicle, the direction and scale of the blur kernel are affected by multiple dynamic factors such as the instantaneous speed of the vehicle and mechanical vibration, which are difficult to accurately obtain through theoretical derivation or calibration, resulting in the unknown and time-varying nature of the blur kernel. The embodiment uses a blind deconvolution algorithm to restore the dynamic photosensitive signal.

[0099] Specifically, the embodiment uses a blind deconvolution algorithm to restore the signal, ensuring that the input photosensitive signal in the subsequent end-to-end mapping process is high-quality data close to the static real situation, so as to ensure the dynamic accuracy of the mapping result.

[0100] The process of restoring the signal using the blind deconvolution algorithm can be represented as:

[0101]

[0102] wherein, denotes is the dynamic photosensitive signal, f represents the true light intensity to be recovered, h represents the blur kernel, n represents the additive noise, and * represents the convolution operation. ​​​​​

[0103] Considering the ill-posedness of blind deconvolution, there are infinite pairs of (I, K) satisfying f , h ) that can be estimated from the blurred dynamic light intensity g = f * h . A tiny noise disturbance n can cause a significant deviation of the estimated f and h , and effective prior knowledge is needed to constrain the solution space in the restoration process. Among them, considering that the luminance gradient is statistically sparse, and the blurring process will destroy this sparsity, the true light intensity f uses a sparse gradient prior to introduce L0 or L1 norm to the restored dynamic light signal to constrain the prior of the clear edge structure and smooth area. For the prior of the blur kernel, to ensure the physical reasonableness of the blur kernel, the blur kernel needs to be subject to the prior constraints of support, non-negativity, normalization and smoothing.

[0104] In summary, the main purpose of the restoration process is to find the most likely blur kernel g and the true light intensity h from the given dynamic light intensity f , and the above process can be expressed in the following form:

[0105]

[0106] where represents the optimal estimate of P , g | f, h represents the likelihood term, P , f is the light intensity prior, P , h is the blur kernel prior.

[0107] To solve this process, an alternating minimization strategy is used to alternately estimate and update the true light intensity and the blur kernel until convergence.

[0108] where the update method of the blur kernel is:

[0109] Assuming that the current true light intensity is accurate, update the blur kernel:

[0110]

[0111] where represents the updated blur kernel, represents the current true light intensity, h represents the current blur kernel, g represents the dynamic light signal, * represents convolution, λ 1 represents the regularization coefficient, Φ( h ) represents the prior constraint of the kernel function.

[0112] The updating method of the real light intensity is:

[0113] Assuming that the current blur kernel is accurate, the real light intensity is updated as:

[0114]

[0115] wherein, represents the updated real light intensity, represents the current blur kernel, g represents the dynamic photosensitive signal, and * represents convolution, λ 2 represents a regularization coefficient, and Ψ( f ) represents a prior constraint of the real light intensity.

[0116] Exemplarily, referring to Table 1, the dynamic restoration performance comparison table, for the dynamic data collected in the 4381-meter test tunnel, the full-line data restoration processing on the processor is completed in about 185 seconds by using the above method, and the average single-frame processing time is 42 milliseconds. Compared with the original data without using the restoration algorithm, the peak signal-to-noise ratio of the restored data is improved by an average of 10.5 dB, the structural similarity index is improved by an average of 0.30, and the comprehensive performance is significantly improved.

[0117] Table 1 Dynamic restoration performance comparison

[0118]

[0119] The data in the table is derived from statistical analysis of 30 groups of effective samples of the basic section of the tunnel. The average error of the restored data is 0.72%, which meets the requirement of error control within 5% in engineering application. The improvement of the peak signal-to-noise ratio and the structural similarity proves the effectiveness of the algorithm in improving the accuracy of the data.

[0120] Further, in the application process of the mapping model, the influence of the light source spectrum difference and the system state change on the photosensitive sensor reading is corrected in real time through relative spectrum correction and temperature drift compensation. The purpose is to maintain the long-term effectiveness and stability of the global calibration coefficient, and to ensure that it can still output the value close to the real light intensity in the complex engineering environment.

[0121] Specifically, in a possible implementation, the step S30 comprises steps B10-B40:

[0122] Step B10, acquiring the discrete spectral response function of the photosensitive sensor, and constructing a response matrix.

[0123] Step B20, performing matrix transformation on the spectral vector of the restored dynamic photosensitive signal based on the inverse matrix of the response matrix, to obtain a corrected spectral vector.

[0124] Step B30, determining a relative spectral corrected photosensitive signal based on the corrected spectral vector.

[0125] Step B40, performing temperature drift compensation on the relative spectral corrected photosensitive signal to obtain a corrected photosensitive signal.

[0126] Wherein, step B40 comprises step B41~step B42:

[0127] Step B41, constructing a temperature drift model to determine the relationship between the measured photosensitive signal at the current temperature and the temperature-compensated corrected photosensitive signal. The temperature drift model is expressed as:

[0128]

[0129] Wherein, S raw ( T ) represents the measured photosensitive signal at temperature T, G ( T ) represents a gain proportional function, S ideal represents an ideal true value.

[0130] Step B42, determining the gain proportional function based on the ratio of the measured photosensitive signal at the reference temperature to the measured photosensitive signal at the current temperature; the gain proportional function is expressed as:

[0131]

[0132] Wherein, S raw (T ref ) represents the measured photosensitive signal at the reference temperature T erf S raw ( T ) represents the measured photosensitive signal at temperature T

[0133] Based on the temperature drift model, the temperature compensation is performed on the relative spectral corrected photosensitive signal to obtain the corrected photosensitive signal.

[0134] Specifically, the present embodiment obtains the discrete spectral response function of the system through laboratory monochromator calibration, constructs a response matrix, and uses deconvolution or matrix transformation method to restore the original reading of the sensor to the real spectral shape and absolute value of the target, so as to realize relative spectral correction. This effectively eliminates the influence of spectral characteristic differences of different tunnel light sources (such as LED, sodium lamp) on the measurement results.

[0135] ​​The basic purpose of the relative spectral correction is to eliminate the influence of the measuring system itself on the real physical spectral information of the template. The response sensitivity of any measuring system to light of different wavelengths is not the same, and the spectral correction is a process of restoring the subjective reading of the system to the real value of the target through physical and mathematical methods. In the embodiment, the recovered photosensitive signal S can be understood as the real spectrum L(λ) The system spectral response function R(λ) After convolution and superposition in the whole waveband and adding noise, the result can be expressed as:

[0136]

[0137] wherein S represents the recovered photosensitive signal, L(λ) represents the real spectrum, R(λ) represents the spectral response function, and λ represents the waveband.

[0138] Considering that the light intensity detection in the embodiment is a dynamic process, the relative spectral correction can restore the shape and color information of the spectrum to eliminate the color deviation caused by the spectral response R（ ） The process is a deconvolution or matrix transformation process, and the matrix transformation process can be expressed as:

[0139]

[0140] wherein S represents the recovered photosensitive signal, R represents the discrete response matrix of the photosensitive sensor, which can be obtained by laboratory monochromator calibration, represents the photosensitive signal after the relative spectral correction.

[0141] Due to the high requirement for the accuracy of the measurement data in actual engineering projects, the relative spectral correction is selected here. According to the above steps, after the relative spectral correction, the mean square error (MSE) is 0.002074, the peak position error is -2.80 nm, the amplitude error is 0.08%, the correlation coefficient is 0.987765, and the maximum absolute error is 0.128454.

[0142] Further, in the embodiment, the material of the photosensitive sensor is semiconductor silicon, and through experimental calibration, it is found that the photoelectric conversion efficiency of the photosensitive sensor has a negative temperature coefficient, and a temperature compensation model based on a gain proportional function is established accordingly to correct the measurement value in real time, thereby ensuring the stability of the system in a wide temperature range.

[0143] Specifically, in order to guarantee the consistency of the luminance measurement value at different temperatures in different environments, temperature drift correction must be performed. Through experiments, it is found that in the key working interval of 20-50°C, the light measurement index of the photosensitive sensor presents a monotonic downward trend with the increase of temperature, and the drift mode is mainly sensitivity drift rather than zero drift caused by dark current. Sensitivity temperature drift is a multiplicative error that the gain or responsivity of the measurement system changes with temperature, which is most likely caused by the negative temperature coefficient of the sensor photoelectric conversion efficiency. The sensor material used in this embodiment is semiconductor silicon, and the band gap of semiconductor silicon becomes narrower with the increase of temperature, which means that under constant light, the charge generated by the sensor will decrease with the increase of temperature, and finally the measured brightness value will decrease with the increase of temperature.

[0144] This embodiment establishes a model for sensitivity temperature drift according to experimental data and corrects it. Specifically:

[0145] A temperature drift model is constructed to determine the relationship between the photosensitive signal measured at the current temperature and the photosensitive signal after temperature compensation correction. The temperature drift model is represented as:

[0146]

[0147] Wherein, S raw ( T ) represents the photosensitive signal measured at temperature T, G ( T ) represents the gain proportion function, S ideal represents the ideal true value.

[0148] In actual calibration, the measurement value T ref at the reference temperature S raw (T ref ) is used to approximate the ideal true value S ideal， The gain proportion function is determined based on the ratio of the photosensitive signal measured at the reference temperature to the photosensitive signal measured at the current temperature; the gain proportion function is represented as:

[0149]

[0150] Wherein, S raw (T ref ) represents the photosensitive signal measured at the reference temperature T erf ,S raw ( T ) represents temperature T The photosensitive signal is measured.

[0151] After determining the gain proportional function, the relative spectral corrected photosensitive signal is temperature compensated based on the temperature drift model to obtain the corrected photosensitive signal.

[0152] After obtaining the final corrected photosensitive signal, the photosensitive signal is mapped to a luminance value through an end-to-end mapping model. Specifically, to achieve “end-to-end”, the key is to establish a global, one-to-one, deterministic mathematical mapping function. This is completed through a one-time, system-level overall calibration in a laboratory environment. Specifically, a standard luminance board traceable to the national metrological reference is used to irradiate the entire sensing system, and the correspondence between a series of known absolute luminance values (L) and the digital readings (D) of the system output is recorded, and the global calibration coefficient K is obtained by fitting, thereby establishing the core mapping model: L ref D out L ref =K× D out .

[0153] It can be understood that in the field measurement, the system only needs to obtain the digital reading D of any measurement point, and through the above unique and fixed mapping function K, the absolute luminance value L of the point can be directly and unconditionally calculated. The whole process does not need image processing algorithm intervention, nor does it need to query or estimate any external parameters (such as reflectivity p). D out

[0154] Further, in a feasible implementation, the tunnel pavement is divided into three detection areas from left to right, and three photosensitive sensors are used to synchronously collect data. After step S40, steps S50-S70 are further included:

[0155] Step S50: Based on the vehicle positioning data and vehicle speed information, the time sequence photosensitive data collected by the three photosensitive sensors is converted into three-channel spatial sequence photosensitive data.

[0156] Step S60: The three-channel spatial sequence photosensitive data is subjected to transverse interpolation processing to fill in the blind area of perception.

[0157] Step S70: The three-channel spatial sequence photosensitive data after interpolation processing is subjected to multi-channel consistency verification, and a tunnel full-line luminance distribution curve is fused and generated.

[0158] ​​​Specifically, by synchronously collecting data through three spatial position accurate calibration of photosensitive sensors, combining high-precision GNSS positioning and vehicle speed information, the collected data is converted from "time series" to "spatial series", and a complete, continuous, and position-accurate tunnel full-line brightness distribution curve is fused and generated through transverse interpolation and multi-channel consistency verification. Seamless and continuous measurement of single-vehicle road surface brightness is realized, which provides a complete data basis for the calculation of various lighting evaluation indexes (such as average brightness and uniformity) required by the "Technical Specification for Highway Tunnel Maintenance", and can intuitively locate the sections with insufficient brightness or poor uniformity.

[0159] Applying the above tunnel road surface lighting brightness detection method, the application further provides a tunnel road surface lighting brightness detection system. The system comprises a data acquisition unit, which comprises three photosensitive sensors, a temperature sensor, an encoder circuit, and a GNSS navigation module arranged in sequence. The photosensitive sensor is connected with a filter circuit. The photosensitive sensor is used to convert a light signal into an electric signal. The filter circuit is used to filter the electric signal to obtain a dynamic photosensitive signal. The encoder circuit is installed on a wheel of a detection vehicle and is used to collect vehicle speed information.

[0160] In this embodiment, by using a brightness photosensitive sensor with a physical spectral response matched with the human eye V(λ), combining blind deconvolution restoration of a dynamic photosensitive signal, relative spectral correction, and temperature drift compensation, an end-to-end direct mapping path from a light signal to a brightness value is constructed, and indirect links such as illumination conversion, reflectivity estimation, and image gray scale-brightness conversion are completely abandoned. Systematic errors introduced by material aging, spectral differences, motion blur, and environmental temperature changes in traditional methods are effectively avoided, continuous, stable, high-precision, and traceable detection of tunnel lighting brightness under normal driving conditions is realized, and the accuracy, consistency, and engineering practicability of dynamic measurement are significantly improved.

[0161] The lower machine processor is used to obtain the dynamic photosensitive signal, vehicle positioning data, sensor temperature data, and vehicle speed information from the data acquisition unit, and to sort and forward them to the upper machine processor.

[0162] The upper machine processor is used to run a computer program. The computer program is configured to realize the steps of the tunnel road surface lighting brightness detection method.

[0163] The photosensitive sensor comprises an optical lens, a filter, and a photoelectric conversion circuit.

[0164] Exemplarily, with reference to Figure 3The whole system is composed of a collection system and a data processing system. The collection system is to collect optical signals by sensors, then convert the optical signals into electrical signals, pre-process the signals (filtering, signal enhancement) through the circuit, receive the pre-processed signals by the MCU, and push the data to the host computer software through the wireless transceiver after digital filtering and arrangement by the lower computer software program. The data processing system mainly includes a data processing interface program: the data processing interface has functions of loading original data, setting segmented data, viewing segmented data curves, and exporting data.

[0165] Hardware components: photosensitive sensor (optical lens, filter and photoelectric conversion circuit), signal filtering, processor, lithium battery charging and discharging, voltage conversion circuit, wireless transceiver circuit, GNSS navigation, encoder circuit, temperature sensor, defogging circuit, etc. Lithium battery charging and discharging circuit: the system uses lithium battery for power supply, and the charging and discharging management circuit is a necessary design. Photoelectric conversion circuit: used for converting optical signals into voltage signals (dynamic photosensitive signals).

[0166] Signal filtering: the converted voltage signal has unstable signal, much noise and other conditions, so signal filtering is needed to obtain a relatively clean signal.

[0167] Processor: the lower computer processor, which receives signals, receives encoder signals, temperature and humidity sensor data, GNSS navigation data, controls the defogging device, communicates with the host computer, and sends the arranged data to the host computer through the wireless transceiver.

[0168] Wireless transceiver: connecting the lower computer processor and the host computer.

[0169] The host computer processor is set in the computer to run the computer program, and the computer program is configured to realize the steps of the above-mentioned tunnel pavement lighting brightness detection method.

[0170] Defogging circuit: since the system needs to collect light, there will be a collection window, so a window defogging circuit is needed to avoid fogging.

[0171] GNSS navigation: it is necessary to record the detection position in real time to facilitate the user to find the abnormal detection position and realize the tunnel light transformation. Temperature sensor: the system temperature state detection is also suitable for data collection correction. Encoder circuit: the encoder circuit is installed on the wheel to collect the vehicle mileage / vehicle speed in real time, and the system collects data at fixed intervals according to the vehicle speed.

[0172] In the above system: the optical signal needs to be geometrically processed by the optical lens; the optical signal is converted into an electrical signal by the photoelectric conversion circuit, needs to be converted from a current signal to a voltage signal, and then is amplified twice and transmitted to the processor for data ADC. After a series of processing such as digital filtering / ensemble averaging by the processor, the signal is transmitted to the upper computer through the wireless transceiver.

[0173] In the present example, the entire system is designed with three sensors to measure the left, middle and right three measurement points of a single lane, which can realize one-time rapid measurement of one lane data. The measurement position can be adjusted and installed through the laser indicator light on the system.

[0174] It should be noted that all the above examples are only for understanding the present application and do not constitute a limitation on the tunnel pavement illumination brightness detection method of the present application. More simple transformations based on this technical concept are within the protection scope of the present application.

[0175] The present application also provides a tunnel pavement illumination brightness detection device, please refer to Figure 4 The tunnel pavement illumination brightness detection device comprises:

[0176] The acquisition module 10 is configured to acquire a dynamic photosensitive signal by using a photosensitive sensor during detection of vehicle movement.

[0177] The restoration module 20 is configured to restore the dynamic photosensitive signal by using a blind deconvolution algorithm to obtain a restored photosensitive signal.

[0178] The repair module 30 is configured to perform relative spectral correction and temperature drift compensation on the restored photosensitive signal to obtain a corrected photosensitive signal.

[0179] The detection module 40 is configured to perform end-to-end mapping on the corrected photosensitive signal by using an end-to-end mapping model to obtain a brightness detection result of a tunnel detection pavement; wherein the input of the end-to-end mapping model comprises the corrected photosensitive signal and a global calibration coefficient.

[0180] The tunnel pavement illumination brightness detection device provided by the present application adopts the tunnel pavement illumination brightness detection method in the above embodiments, which can solve the technical problem of low reliability of related art tunnel pavement illumination brightness detection. Compared with the related art, the tunnel pavement illumination brightness detection device provided by the present application has the same beneficial effects as the tunnel pavement illumination brightness detection method provided by the above embodiments, and the other technical features in the tunnel pavement illumination brightness detection device are the same as the features disclosed in the above embodiments, which will not be repeated here.

[0181] It should be understood that various aspects of the disclosure can be implemented in hardware, software, firmware, or a combination thereof. In the description of the embodiments above, specific features, structures, materials or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0182] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0183] The present application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e. computer programs) for performing the tunnel pavement illumination brightness detection method in the above embodiments.

[0184] The computer readable storage medium provided by the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system or device, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to: an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present embodiment, the computer readable storage medium can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, system or device. The program code contained on the computer readable storage medium can be transmitted by any appropriate medium, including but not limited to: electric wire, optical cable, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0185] The above computer readable storage medium can be contained in the tunnel pavement illumination brightness detection system; or can exist separately without being assembled into the tunnel pavement illumination brightness detection system.

[0186] The computer readable storage medium described above carries one or more programs, when the one or more programs are executed by the tunnel pavement illumination brightness detection system, the tunnel pavement illumination brightness detection system is caused to: in the detection vehicle moves, the dynamic photosensitive signal is acquired through the photosensitive sensor; the dynamic photosensitive signal is restored through the blind deconvolution algorithm, and the restored photosensitive signal is obtained; the relative spectral correction and temperature drift compensation are carried out to the restored photosensitive signal, and the modified photosensitive signal is obtained; the end-to-end mapping model is used to carry out end-to-end mapping to the modified photosensitive signal, and the brightness detection result of the tunnel detection pavement is obtained;Wherein, the input of the end-to-end mapping model includes the modified photosensitive signal and the global calibration coefficient.

[0187] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0188] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0189] The modules described in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0190] The readable storage medium provided in the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., a computer program) for executing the tunnel pavement illumination brightness detection method described above, and can solve the technical problem of low reliability of tunnel pavement illumination brightness detection in the related art. Compared with the related art, the computer readable storage medium provided in the present application has the same beneficial effects as the tunnel pavement illumination brightness detection method provided in the above embodiments, and will not be described here.

[0191] The present application also provides a computer program product, which includes a computer program, and the computer program is executed by a processor to implement the steps of the tunnel pavement illumination brightness detection method described above.

[0192] The computer program product provided in the present application can solve the technical problem of low reliability of tunnel pavement illumination brightness detection in the related art. Compared with the related art, the computer program product provided in the present application has the same beneficial effects as the tunnel pavement illumination brightness detection method provided in the above embodiments, and will not be described here.

[0193] The above is only some embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation, direct / indirect application in other related technical fields based on the technical concept of the present application, and the content of the specification and drawings are included in the patent protection scope of the present application.

Claims

1. A method for detecting the brightness of tunnel pavement lighting, characterized in that, The method includes: During the detection of vehicle movement, dynamic light-sensing signals are acquired through a photosensitive sensor; The dynamic photosensitive signal is restored by a blind deconvolution algorithm to obtain the restored photosensitive signal; The restored photosensitive signal is subjected to relative spectral correction and temperature drift compensation to obtain the corrected photosensitive signal. The modified photosensitive signal is mapped end-to-end using an end-to-end mapping model to obtain the brightness detection result of the tunnel road surface; wherein the input of the end-to-end mapping model includes the modified photosensitive signal and global calibration coefficients.

2. The method for detecting tunnel pavement lighting brightness as described in claim 1, characterized in that, The step of restoring the dynamic photosensitive signal using a blind deconvolution algorithm to obtain the restored photosensitive signal includes: A convolutional degradation model for acquiring dynamic photosensitized signals is constructed; wherein the input of the convolutional degradation model includes the true light intensity and additive system noise at the corresponding detection position, and the output of the convolutional degradation model includes the dynamic photosensitized signal at the corresponding detection position; Using a blind deconvolution algorithm, the dynamic photosensitized signal is used as a known condition to deconvolve the degenerate convolution model, calculate the true light intensity, and obtain the restored photosensitized signal.

3. The method for detecting tunnel pavement lighting brightness as described in claim 2, characterized in that, The step of using a blind deconvolution algorithm, with the dynamic photosensitized signal as known conditions, to deconvolve the degenerate convolution model, calculate the true light intensity, and obtain the restored photosensitized signal includes: Initialize the real light intensity and the blur kernel; An alternating minimization strategy is adopted to alternately estimate and update the true light intensity and the blur kernel until convergence, thereby obtaining the true light intensity. The update method for the fuzzy kernel is as follows: Assuming the current true light intensity is accurate, update the blur kernel: in, This indicates the updated fuzzy kernel. This represents the current actual light intensity. h Indicates the current fuzzy kernel. g Represents dynamic photosensitized signals; * indicates convolution. λ 1 represents the regularization coefficient, Φ( h ) represents the prior constraints of the kernel function; The method for updating the actual light intensity is as follows: Assuming the current blur kernel is accurate, update the true light intensity: in, This represents the updated true light intensity. Indicates the current fuzzy kernel. g Represents dynamic photosensitized signals; * indicates convolution. λ 2 represents the regularization coefficient, Ψ( f ) represents the prior constraint on the actual light intensity.

4. The method for detecting tunnel pavement lighting brightness as described in claim 1, characterized in that, The step of performing relative spectral correction and temperature drift compensation on the restored photosensitive signal to obtain the corrected photosensitive signal includes: Obtain the discrete spectral response function of the photosensitive sensor and construct the response matrix; Based on the inverse of the response matrix, a matrix transformation is performed on the spectral vector of the restored dynamic photosensitized signal to obtain a corrected spectral vector; Based on the corrected spectral vector, the photosensitive signal after relative spectral correction is determined; Temperature drift compensation is performed on the relative spectral corrected photosensitive signal to obtain the corrected photosensitive signal.

5. The method for detecting tunnel pavement lighting brightness as described in claim 4, characterized in that, The step of performing temperature drift compensation on the relatively spectrally corrected photosensitive signal to obtain the corrected photosensitive signal includes: A temperature drift model is constructed to determine the relationship between the photosensitive signal measured at the current temperature and the photosensitive signal after temperature compensation correction; the temperature drift model is expressed as: in, S raw ( T () represents the photosensitive signal measured at temperature T. G ( T () represents the gain scaling function. S ideal Represents the ideal truth value; The gain scaling function is determined based on the ratio of the photosensitive signal measured at the reference temperature to the photosensitive signal measured at the current temperature; the gain scaling function is expressed as: in, S raw (T ref ) Indicates reference temperature T erf The measured photosensitive signal, S raw ( T () indicates temperature T The photosensitive signal measured below; Based on the temperature drift model, temperature compensation is performed on the relative spectral corrected photosensitive signal to obtain the corrected photosensitive signal.

6. The method for detecting tunnel pavement lighting brightness as described in claim 1, characterized in that, The tunnel surface is divided into three detection areas from left to right, and data is collected synchronously by three photosensitive sensors. After the step of obtaining the brightness detection result of the tunnel surface by performing end-to-end mapping on the corrected photosensitive signals using an end-to-end mapping model, the method further includes the following step: Based on vehicle positioning data and vehicle speed information, the time-series photosensitive data collected by the three photosensitive sensors are converted into three-channel spatial sequence photosensitive data; Lateral interpolation is performed on the three-channel spatial sequence photosensitive data to fill in the sensing blind spots; Multi-channel consistency verification is performed on the interpolated three-channel spatial sequence photosensitive data, and the data are fused to generate a brightness distribution curve for the entire tunnel.

7. A device for detecting the brightness of tunnel pavement lighting, characterized in that, The device includes: The acquisition module is used to acquire dynamic photosensitive signals through a photosensitive sensor during the detection of vehicle movement; The restoration module is used to restore the dynamic photosensitive signal using a blind deconvolution algorithm to obtain the restored photosensitive signal; The repair module is used to perform relative spectral correction and temperature drift compensation on the restored photosensitive signal to obtain the corrected photosensitive signal. The detection module is used to perform end-to-end mapping on the corrected photosensitive signal through an end-to-end mapping model to obtain the brightness detection result of the tunnel detection surface; wherein, the input of the end-to-end mapping model includes the corrected photosensitive signal and global calibration coefficients.

8. A tunnel pavement lighting brightness detection system, characterized in that, The system includes: The data acquisition unit includes three photosensitive sensors, a temperature sensor, an encoder circuit, and a GNSS navigation module arranged in sequence. The photosensitive sensors are connected to a filtering circuit. The photosensitive sensors are used to convert light signals into electrical signals. The filtering circuit is used to filter the electrical signals to obtain dynamic photosensitive signals. The encoder circuit is installed on the wheels of the detection vehicle to collect vehicle speed information. The lower-level processor is used to acquire dynamic photosensitive signals, vehicle positioning data, sensor temperature data and vehicle speed information from the data acquisition unit, organize and forward them to the upper-level processor; A host computer processor is used to run a computer program configured to implement the steps of the tunnel pavement lighting brightness detection method as described in any one of claims 1 to 6.

9. The tunnel pavement lighting brightness detection system as described in claim 8, characterized in that, The photosensitive sensor includes an optical lens, a filter, and a photoelectric conversion circuit.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the tunnel pavement lighting brightness detection method as described in any one of claims 1 to 6.

Images