A method of assessing solar glare combining all-sky radiance and street view photos

Abstract

The embodiment of the application provides a kind of solar glare evaluation method combined with all-sky radiation and street view photo, belong to urban light environment evaluation technical field, first acquisition meteorological data and obtain multi-view street view image of road network, rely on meteorological parameter to solve solar position and sky state feature;Dynamic light visual efficiency empirical model is constructed according to sky type, coupled with multiple meteorological parameters to solve dynamic conversion coefficient, and radiation, luminosity double scale data are output simultaneously;Based on the sky type correction Perez model, the anisotropic all-sky field is reconstructed by discretization;Street view image is spliced, semantic segmentation and fish-eye projection transformation, generate binary sky visibility mask matching human eye hemispherical field of view, combined with sky light field and projection factor integral to solve the vertical illuminance of human eye after obstruction correction;Finally, coupled with human eye visual physiological mechanism, subjective discomfort glare is evaluated using DGP model, and retinal irradiance is quantified to represent disability glare using human eye imaging optical model, which can realize accurate evaluation of glare risk.

Description

Technical Field

[0001] This application relates to the field of urban light environment assessment technology, and in particular to a method for assessing solar glare by combining all-sky radiation and street view photographs. Background Technology

[0002] Solar glare, a common and significantly harmful transient visual disturbance, poses a serious threat to urban road traffic safety. When solar radiation at a specific angle directly enters a driver's field of vision, it triggers light scattering within the eye and physiological stress responses in retinal cells. Direct sunlight acting on the pupil can trigger photobleaching of retinal cone cells, leading to a temporary decrease in visual sensitivity and the formation of a dark spot in the center of the visual field lasting for several seconds, thus causing temporary blindness, i.e., disabling glare. Furthermore, the uncomfortable glare from the sun can also cause psychological disturbance to drivers, weakening their perception and emergency response capabilities to sudden traffic situations, thereby significantly increasing the probability and risk of traffic accidents. However, current technologies for urban glare risk assessment and monitoring mostly focus on determining whether glare has occurred, lacking a comprehensive physical inversion of the physiological and psychological mechanisms of the human eye, making it difficult to objectively and quantitatively assess the risks of disabling glare and blindness. For example: First, the rough estimation method based on low solar altitude angle ignores the real shading effect of urban elements such as buildings and street trees. The oversimplified geometric assumptions result in low simulation accuracy in complex urban environments, making it difficult to provide reliable quantitative results. Second, although methods based on lidar point clouds or high-resolution surface models have high geometric accuracy, the data acquisition cost is high and the computational complexity is extremely high. They are usually only applicable to local areas and are difficult to extend to efficient evaluation of large-scale road networks at the city level. Third, traditional static sky models and isotropic assumptions lack the ability to respond to dynamic weather conditions, making it difficult to accurately characterize the anisotropic scattering characteristics of the atmosphere under the interaction of aerosols and clouds. In particular, they ignore the bright radiation region in the vicinity of the sun and systematically underestimate the risk of glare under complex weather conditions such as cloudy skies.

[0003] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0004] The main objective of this application is to propose a solar glare assessment method that combines all-sky radiation and street view photos to achieve accurate inversion of urban glare risk while taking into account high spatiotemporal resolution, dynamic meteorological response, and physiological mechanism explanation capabilities.

[0005] To achieve the above objectives, one aspect of this application proposes a method for assessing solar glare by combining all-sky radiation and street view photographs, the method comprising: Acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network; Calculate the solar position parameters and sky state parameters based on the meteorological data; A type-specific dynamic optical performance empirical model is constructed, and the dynamic conversion coefficient from the radiation domain to the luminosity domain is solved by combining the radiation flux component, the solar position parameter and the sky state parameter, and the dual-domain physical quantity results are output. Based on the dual-domain physical quantity results, the anisotropic all-sky absolute sky field containing direct solar radiation spots is reconstructed by discretized gridding based on the Perez all-sky model with sky type correction. A panoramic image is generated by stitching together multi-view road network street scene images. The panoramic image is then subjected to pixel-level semantic segmentation and fisheye projection transformation to generate a binary fisheye sky visibility mask. Based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky, the vertical illuminance of the human eye is calculated by constructing a projection factor. Dynamic glare risk assessment is performed based on the vertical illuminance of the human eye and coupled with visual mechanisms.

[0006] In some embodiments, the solar position parameters include solar altitude angle, solar azimuth angle, solar declination angle, and solar hour angle; the sky state parameters include atmospheric mass, sky brightness parameter, sky clarity parameter, and atmospheric precipitable water vapor content, wherein the atmospheric precipitable water vapor content is calculated based on dew point temperature.

[0007] In some embodiments, a type-specific dynamic optical performance empirical model is constructed, and the dynamic conversion coefficient from the radiation domain to the luminosity domain is solved by combining the radiative flux components, the solar position parameters, and the sky state parameters, outputting dual-domain physical quantity results, including: Sky conditions are categorized based on the clarity parameter, and a corresponding sky category index is matched for each sky condition category. Based on the sky category index, match the scattering coefficient and direct radiation coefficient corresponding to the dynamic light and visual performance empirical model; Based on the scattering coefficient and the direct light coefficient, the ratio of scattered light illuminance to direct light illuminance is calculated according to the light performance formula. Based on the ratio of scattered illuminance to direct illuminance, the horizontal scattered illuminance and normal direct illuminance are obtained by inversion of radiance and illuminance. By combining radiometric and photometric parameters, the results of physical quantities in both domains are output.

[0008] In some embodiments, based on the dual-domain physical quantity results, and using a Perez all-sky model modified according to sky type, an anisotropic all-sky absolute sky field containing direct solar radiation spots is discretized and reconstructed, including: Construct the Perez coefficient matrix and match the Perez coefficients corresponding to the brightness distribution parameters based on the sky category index; The brightness distribution parameters are solved differently based on different clearness conditions, and the parameters are corrected under low clearness conditions to obtain a set of corrected parameters. Using zenith angle and azimuth angle as the dividing directions, the sky hemisphere is discretized based on a regular grid to obtain discrete sky grid cells; For each of the sky grid cells, the relative sky brightness distribution is calculated based on the set of correction parameters; The scattering components are determined according to the output mode and the normalization factor is calculated. The relative sky brightness distribution is then normalized to generate a scattered sky field. The solar direct component is assigned to the corresponding sky grid cell where the sun is located, and the scattered sky field is superimposed to generate the absolute sky field of the entire sky.

[0009] In some embodiments, after generating the absolute sky field of the entire sky, the process includes: Construct a global horizontal radiation energy model to verify energy conservation. The energy conservation test includes: Calculate the total radiant energy of the reconstructed sky field on the horizontal plane; The theoretical value of total radiant energy is calculated based on meteorological data; The normalization factor is adjusted based on the deviation between the total radiant energy and the theoretical value of the total radiant energy until the energy conservation requirement is met, and the reconstructed sky field is output.

[0010] In some embodiments, a panoramic image is generated by stitching together multi-view street view images of the road network, and pixel-level semantic segmentation and fisheye projection transformation are performed on the panoramic image to generate a binary fisheye sky visibility mask, including: The multi-view road network street scene images are stitched together to generate an equidistant cylindrical projection panoramic image, and a calibration image is obtained after orientation calibration. The calibration image is used to predict the semantic category based on a deep learning semantic segmentation method, and a panoramic binary sky mask is generated after assignment. By establishing a reverse projection coordinate mapping relationship, a fisheye projection transformation is performed on the panoramic binary sky mask to obtain a binary fisheye sky visibility mask.

[0011] In some embodiments, the vertical illuminance of the human eye is calculated by constructing a projection factor based on the binary fisheye sky visibility mask and the absolute sky field of the entire sky, including: Based on the field of view corresponding to the main viewing direction, construct the glare perception field of the human eye; Construct the vertical viewing plane normal vector and the incident light projection within the glare receptive field, and calculate the projection factor; Based on the projection factor, and combined with the binary fisheye sky visibility mask, a grid-weighted integral is performed to calculate the vertical illuminance of the human eye.

[0012] In some embodiments, the glare includes discomfort glare and disability glare; When assessing the risk of uncomfortable glare, the total illuminance and effective luminance are coupled, based on the probability of sunlight glare. The model calculates the probability of sunlight glare, and then combines it with preset grading thresholds to classify the risk level of uncomfortable glare; When assessing the risk of disability glare, the vertical illuminance received at the cornea is converted into retinal irradiance based on the human eye imaging optical model, and the risk level of disability glare is classified in combination with a preset risk threshold.

[0013] To achieve the above objectives, another aspect of this application proposes a solar glare assessment system that combines all-sky radiation and street view photographs, the system comprising: The data acquisition module is used to acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network. The data calculation module is used to calculate the solar position parameters and sky state parameters based on the meteorological data; The dynamic optical performance conversion module is used to construct a type-specific dynamic optical performance empirical model, and combine the radiation flux component, the solar position parameter and the sky state parameter to solve the dynamic conversion coefficient from the radiation domain to the luminosity domain, and output the dual-domain physical quantity results. The sky field reconstruction module is used to reconstruct the anisotropic all-sky absolute sky field containing direct sunlight spots by discretizing and meshing based on the results of the dual-domain physical quantities and the Perez all-sky model modified according to sky type. The projection transformation module is used to stitch together multi-view road network street scene images to generate a panoramic image, and to perform pixel-level semantic segmentation and fisheye projection transformation on the panoramic image to generate a binary fisheye sky visibility mask. The vertical illuminance calculation module is used to calculate the vertical illuminance of the human eye by constructing a projection factor based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky. The glare risk assessment module is used to perform dynamic glare risk assessment based on the vertical illuminance of the human eye and coupled with the visual mechanism.

[0014] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0015] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0016] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0017] The embodiments of this application include at least the following beneficial effects: This application provides a solar glare assessment method combining all-sky radiation and street view photos. This scheme extracts radiative flux and dew point temperature from meteorological data and acquires multi-view street view images of the road network. It calculates the solar position and sky state characteristics based on meteorological parameters; constructs a dynamic light and vision efficiency empirical model for different sky types, and couples multiple meteorological parameters to solve the dynamic conversion coefficient from the radiation domain to the photometric domain, simultaneously outputting radiation and photometric dual-scale data; and discretizes and reconstructs the anisotropic all-sky field containing direct light spots based on the modified Perez model for different sky types. The street view images are stitched, semantically segmented, and transformed by fisheye projection to generate a binary sky visibility mask that matches the hemispherical field of view of the human eye. The vertical illuminance of the human eye after building and vegetation shading correction is solved by combining the sky light field and projection factor integral; finally, it couples the human eye's visual physiological mechanism, simultaneously using the DGP model to assess subjective discomfort glare and the human eye imaging optical model to quantify retinal irradiance to characterize disabling glare, realizing a multi-time period and multi-season dual-dimensional dynamic quantitative assessment of urban road network glare risk. This scheme takes into account both complex atmospheric optical variations and the shading effect of urban built-up areas, effectively reducing simulation errors of traditional static models and providing quantitative support for optimizing urban road lighting environment and traffic safety planning. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating the steps of a solar glare assessment method combining all-sky radiation and street view photographs provided in an embodiment of this application; Figure 2 This is a flowchart of a solar glare assessment method combining all-sky radiation and street view photographs provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a solar glare assessment system that combines all-sky radiation and street view photos, provided in an embodiment of this application. Figure 4a This is a schematic diagram of the panoramic street view image after orientation calibration; Figure 4b This is a schematic diagram showing the geometric relationship between the glare perception field and the sun from a driving perspective. Figure 4c This is a schematic diagram of a fisheye projection image and a sky visibility mask; Figure 4dThis is a schematic diagram of the sky light field distribution and glare assessment results under clear weather conditions; Figure 4e This is a schematic diagram of the sky light field distribution and glare assessment results under cloudy weather conditions; Figure 5a This is a schematic diagram of the duration of uncomfortable glare in Guangzhou throughout the year; Figure 5b This is a schematic diagram of the annual glare duration for disabled individuals in Guangzhou. Figure 6a This is a schematic diagram of the glare index of the average vertical illuminance of the human eye throughout the year in Guangzhou. Figure 6b This is a schematic diagram of the glare index map of the average annual solar glare probability in Guangzhou. Figure 7a This is a schematic diagram of the annual average vertical illuminance of the human eye in the overall eastward direction of Guangzhou. Figure 7b This is a schematic diagram of the annual average vertical illuminance of the human eye in the westward direction of Guangzhou. Figure 8a This is a schematic diagram of the glare monitoring map in Guangzhou at 07:30 AM on March 21st; Figure 8b This is a schematic diagram of the glare monitoring map in Guangzhou at 16:30 on March 21st; Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0020] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0021] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0023] Before providing a detailed description of the embodiments of this application, some of the nouns and terms used in the embodiments of this application will be explained first. The nouns and terms used in the embodiments of this application shall be interpreted as follows: Disability glare: refers to the scattering of light into the eye when strong light enters the eye, which reduces the contrast between the target and the background on the retina, directly reducing the ability of a person to distinguish objects. Even if the person does not feel the glare subjectively, they will experience blurred vision and be unable to see details. This type of glare that impairs visual function is called disability glare.

[0024] Uncomfortable glare: refers to the subjective discomfort, glare, soreness, and irritability caused by excessive brightness of the light source or the difference between the light source and the background. It does not necessarily lead to a decrease in visual clarity. The core criterion is the subjective visual experience of the person.

[0025] Vertical illuminance of the human eye: refers to the illuminance incident on the human eye with the pupil plane as the receiving surface and perpendicular to the line of sight of the human eye. It directly represents the total amount of light entering the human eye. Illuminance refers to the luminous flux received per unit area.

[0026] Current technologies for urban glare risk assessment and monitoring largely focus on determining whether glare occurs, lacking a comprehensive physical inversion approach that considers the physiological and psychological mechanisms of the human eye. This makes it difficult to objectively and quantitatively assess the risk of disabling glare and blindness. For example: First, coarse estimation methods based on low solar altitude angles ignore the actual shading effects of urban elements such as buildings and roadside trees. Oversimplified geometric assumptions result in low simulation accuracy in complex urban environments, making it difficult to provide reliable quantitative results. Second, while methods based on lidar point clouds or high-resolution surface models have high geometric accuracy, data acquisition costs are high and computational complexity is extremely high. They are usually only applicable to local areas and are difficult to extend to efficient assessment of large-scale urban road networks. Third, traditional static sky models and isotropic assumptions lack responsiveness to dynamic weather conditions, making it difficult to accurately characterize the anisotropic scattering characteristics of the atmosphere under the influence of aerosols and clouds. In particular, they ignore the high-brightness radiation area near the sun and systematically underestimate glare risk under complex weather conditions such as cloudy skies. In view of this, this application provides a method for assessing urban glare that integrates all-sky radiation modeling, deep learning, and human visual mechanisms. Under the constraints of low data cost and high computational efficiency, it breaks through the limitations of traditional single-point measurement and static light environment simulation. By constructing a full-link physical inversion framework that connects macroscopic atmospheric radiation processes, microscopic urban geometric occlusion, and human visual physiological responses, it achieves accurate inversion of urban glare risk while taking into account high spatiotemporal resolution, dynamic meteorological response, and physiological mechanism explanation capabilities. This provides reliable quantitative technical support for the design of autonomous driving vision systems and urban traffic safety early warning.

[0027] This application provides a method for assessing solar glare by combining all-sky radiation and street view photographs, relating to the field of urban light environment assessment. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle terminal, but is not limited thereto. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing a method for assessing solar glare by combining all-sky radiation and street view photographs, but is not limited to the above forms.

[0028] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0029] Please see Figure 1 and Figure 2 As shown, Figure 1 This is a schematic diagram illustrating an optional step in a solar glare assessment method combining all-sky radiation and street view photographs, as provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S1 to S7: S1, acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network; This application uses meteorological data to obtain surface radiation flux components and related environmental parameters, thus providing an input source for constructing a high-precision sky radiation field. The meteorological data uses the ERA5 reanalysis dataset or typical meteorological year data from TMYx, and includes at least the radiation flux components, dew point temperature, and time and latitude / longitude information required for solar geometry calculations, used for inverting direct normal radiation (DNI) and diffuse horizontal radiation (DHI). To match the high-frequency dynamic requirements of instantaneous glare analysis and traffic flow, this application introduces a time-weighted linear interpolation algorithm to resample the radiation flux at non-integer times to ensure the continuity of the inversion process in the time dimension, thereby extracting the direct normal radiation (DNI) and diffuse horizontal radiation (DHI) at the current moment, as shown in the formula: (1) (2) In the formula, weight represents the time weight. and These represent the direct normal radiation and the horizontal scattered radiation at the previous full hour, respectively. and These represent the direct normal radiation and horizontal scattered radiation at the next full hour, respectively. Furthermore, dew point temperature is extracted as an auxiliary meteorological parameter for subsequent inversion of atmospheric precipitable water vapor content.

[0030] This application uses the OpenStreetMap road network to generate sampling points at 50m intervals and acquire street view images of the corresponding locations. For each sampling point, multi-view images at 0°, 90°, 180° and 270° along the direction of travel are extracted. Street view images located in bridge and tunnel areas are removed, while retaining metadata such as the street view ID, latitude and longitude coordinates and acquisition time of the images.

[0031] S2, calculate the solar position parameters and sky state parameters based on the meteorological data; In some embodiments, the solar position parameter includes the solar altitude angle. Sun azimuth Solar declination angle And solar hour angle H; the sky state parameters include atmospheric mass m, sky brightness parameters Sky clarity parameters and atmospheric sedimentable water vapor content .

[0032] First, this application calculates the solar altitude angle based on the geographical location and time information of the observation point. With solar azimuth This is to provide a unified solar geometric basis for subsequent sky state parameter calculations, photometric conversion, and full-sky brightness field reconstruction. The Sun's position in the sky can be calculated using the following formula: (3) (4) In equations (3) and (4), Indicates the geographical latitude of the streetscape measurement point; The solar declination angle (H) represents the angle between the sun's rays and the Earth's equatorial plane, and it varies with the date throughout the year. The solar hour angle (H) is derived from local solar time and reflects the sun's position relative to noon. Solar altitude angle... Convert to angle system Subsequently, the angle of the solar zenith angle was further obtained. The formula is: (5) Then calculate the solar declination angle based on the accumulated days of the year. And according to local solar time The formula for calculating the solar hour angle H is: (6) (7) In equations (6) and (7), n represents the nth day of the year, LST represents the local solar time, when LST=12 corresponds to noon, H=0°; when LST<12, H is negative; when LST>12, H is positive.

[0033] Then, after obtaining the solar position parameters, the relative atmospheric mass is further calculated. This formula is used to correct for the path attenuation effect of light rays in the atmosphere under low solar altitude angle conditions. (8) In equation (8), It is the zenith angle of the sun. Angle system representation; when the angle system represents the solar altitude angle When the relative atmospheric mass m is infinite, it is denoted as infinity.

[0034] Based on this, the sky brightness parameter and sky clarity parameter are further calculated. First, continuous daily values ​​are defined. It is expressed as the sum of the accumulated days of the year and the hourly fractions, that is: (9) In equation (9), Indicates accumulated days over a year. This indicates the hour corresponding to the current time, including the decimal part after converting minutes.

[0035] Then based on the continuous daily sequence values The formula for calculating the orbital eccentricity correction term is as follows: (10) (11) In equations (10) and (11), e represents the orbital angle parameter corresponding to the consecutive daily sequence values. This represents the Earth-Sun distance orbital eccentricity correction factor, used to correct periodic fluctuations in extraterrestrial solar irradiance caused by variations in Earth-Sun distance.

[0036] Then calculate the sky brightness parameters and sky clarity parameters The formula is: (12) (13) In equations (12) and (13), The solar constant; when DHI=0 or the relative atmospheric mass m is infinite. . Represents the solar zenith angle, when hour, Recorded as 1; when the solar altitude angle is less than or equal to 0°, Also denoted as 1. To avoid numerical divergence, The final upper limit of the value is truncated to 12.

[0037] Subsequently, based on dew point temperature Estimate the content of atmospheric sedimentable water vapor , This represents the total amount of condensable water vapor in a unit zenith column, used to characterize the effect of water vapor on the solar spectrum distribution and apparent brightness. The formula is: (14) S3. Construct a type-specific dynamic optical performance empirical model, and combine the radiation flux components, the solar position parameters and the sky state parameters to solve the dynamic conversion coefficient from the radiation domain to the luminosity domain, and output the dual-domain physical quantity results. This step stratifies atmospheric optical states by classifying them according to sky clarity parameters, constructs a categorized dynamic optical performance empirical model, and integrates multiple meteorological variables such as water vapor, solar angle, and sky brightness to solve for dynamic conversion coefficients, achieving accurate dynamic conversion of radiation to photometric quantities under complex weather conditions. Simultaneously, it outputs two types of datasets: radiation and photometric, which respectively support the calculation of objective irradiance for disabled glare and the quantification of physiological perception of uncomfortable glare by the human eye. This overcomes the problem of large conversion errors in traditional fixed optical performance constant models under variable weather scenarios, providing a reliable photometric input basis for high-precision urban glare risk inversion.

[0038] In some embodiments, a type-specific dynamic optical performance empirical model is constructed, and the dynamic conversion coefficient from the radiation domain to the luminosity domain is solved by combining the radiative flux components, the solar position parameters, and the sky state parameters, and the dual-domain physical quantity results are output, including the following steps: S31, classify the sky state categories according to the clearness parameter, and match the corresponding sky category index for the sky state categories; The preset threshold values ​​for the eight types of sky conditions are as follows: 1.065, 1.23, 1.5, 1.95, 2.8, 4.5, and 6.2. hour, ; hour, ; hour, ; hour, ; hour, ; hour, ; hour, ; hour, This yields eight sky state categories and their corresponding sky category indices. The sky category index serves as the unique index for looking up empirical coefficients in the table, distinguishing between different atmospheric light transmission states such as clear skies, cloudy skies, and overcast skies.

[0039] S32, Match the scattering coefficient and direct radiation coefficient corresponding to the dynamic light performance empirical model according to the sky category index; Construct two sets of 4-parameter empirical coefficient matrices, including the coefficient matrix of the ratio of scattered light illuminance. The coefficient matrix of the ratio of direct illuminance to direct illuminance The i-th row of the two sets of empirical coefficient matrices corresponds to the sky category index. Sky status. Based on the current sky category index. Extract the parameters and assign values ​​accordingly, i.e., based on the current sky category index. The coefficient matrix for extracting the ratio of scattered light illuminance The four coefficients in the i-th row are assigned values ​​in sequence. Based on the current sky category index The coefficient matrix for extracting the ratio of direct illuminance. The four coefficients in the i-th row are assigned values ​​in sequence. .For example, At that time, scattering coefficient The direct radiation coefficient is taken as .

[0040] Coefficient matrix of the ratio of scattered light illuminance for: (18) Coefficient matrix of direct illuminance ratio for: (19) S33, Based on the scattering coefficient and the direct illuminance coefficient, the ratio of scattered light illuminance to direct light illuminance is calculated according to the light performance formula; This step calculates the luminous efficacy coefficient from the luminous efficacy formula. The luminous efficacy coefficient is essentially a conversion factor for illuminance per unit of radiant radiation, used to achieve real-time correction of luminous efficacy under dynamic atmospheric conditions, unlike a fixed luminous efficacy constant. (Ratio of diffused light to illuminance) Ratio to direct illuminance The calculation formulas are as follows: (15) (16) In equations (15) and (16), Indicates the zenith angle of the sun. and All are determined by the empirical parameter matrix corresponding to the sky state category; This indicates the amount of atmospheric water vapor that can settle. This represents the sky brightness parameter after lower limit truncation. (The sentence is incomplete and requires further context.) To perform numerical limiting and eliminate abnormal interference from minimum values, the formula is as follows: (17) S34. Based on the ratio of scattered illuminance to direct illuminance, the horizontal scattered illuminance and normal direct illuminance are obtained by inversion of radiance and illuminance. Using the calculated illuminance ratio parameter, i.e., the ratio of diffuse illuminance to direct illuminance, the broadband radiance is converted from the radiative domain to the photometric domain to obtain the horizontal diffuse illuminance. Direct illuminance in the normal direction The formula is: (20) (twenty one) In equations (20) and (21), DHI represents horizontal scattered radiation, and DNI represents normal direct radiation, with units of _____. ; Indicates horizontal diffuse illuminance. Indicates direct illuminance in the normal direction, its unit is . .

[0041] S35 integrates radiometric and photometric parameters to output dual-domain physical quantity results.

[0042] This application retains both radiometric and photometric results in this step. Radiometric quantities are as follows: , , , Isoparameters are used to solve for the sky radiance field, and for quantitative physical calculations of solar radiation flux and retinal irradiance, such as in disability glare and light damage assessment; photometric quantities such as , This tool is used to solve for the sky brightness field, matching the human visual perception mechanism, and is geared towards assessing vertical illuminance, probability of uncomfortable glare, and visual comfort. Both types of values ​​can be input seamlessly into the entire process of full-sky distribution reconstruction, urban street scene field of view projection, and quantitative inversion of glare risk, enabling parallel support from both physical radiation analysis and human physiological visual analysis.

[0043] It should be noted that traditional methods use a fixed optical performance constant for conversion, ignoring spectral variations caused by weather, atmosphere, and solar angle. This application, however, employs a dynamic optical performance empirical model, calculating the conversion ratio based on real-time meteorological parameters. It serves as a dedicated conversion factor for converting scattered and direct radiation into illuminance, achieving accurate mapping between the two types of physical quantities and outputting the result. The categorized empirical coefficients can specifically correct the solar spectral composition under different cloud cover and turbidity conditions, significantly improving conversion accuracy. The model couples multiple meteorological factors to dynamically correct apparent performance, specifically through the content of atmospheric desiccant water vapor. Correcting the absorption and attenuation of visible light by water vapor, solar zenith angle. Correcting the spectral shift caused by changes in the length of the solar light path, and truncating the sky brightness parameter. It is used to characterize the intensity of diffuse light distribution in the sky. Through multivariate joint fitting, the conversion coefficient changes dynamically in real time with meteorological conditions, rather than a single fixed value, so as to adapt to complex dynamic weather scenarios.

[0044] The dynamic optical performance empirical model is a piecewise regression empirical mathematical system based on fitting measured meteorological and optical environment data. Essentially, it is a mapping relationship from multiple inputs to a single output, capable of... Mapped to Specifically, the first layer of this model is a pre-determination layer, used to determine the sky's clarity parameter. Classify sky states into categories and output the sky category index. To adapt to various complex weather conditions; the second layer is a parameter storage layer, used to store the coefficient matrix of the ratio of scattered light intensity to illuminance. Coefficient matrix of ratio to direct illuminance The third layer is the mapping calculation layer, used to input multiple meteorological variables. The dynamic conversion ratio is obtained by mapping calculation. Because the formula incorporates linear, logarithmic, and exponential terms, it is a multivariate nonlinear empirical regression formula, and all parameters are obtained by fitting measured data; therefore, it is called an empirical model. The model has two sets of mapping logic: the first layer discrete mapping "sky clarity parameter"... →Sky Category Index → Extract the radiation coefficients (A, B, C, D) from the empirical coefficient matrix; the second segment is a continuous nonlinear mapping. → "; A complete dynamic light and light performance empirical mapping system is constructed by superimposing two layers. Then, a linear mapping is performed on the output to realize the mapping from radiance to illuminance, that is..." .

[0045] S4. Based on the results of the dual-domain physical quantities, and using the Perez all-sky model modified according to different sky types, the anisotropic all-sky absolute sky field containing direct sunlight spots is reconstructed by discretization and gridding. This step corrects the Perez coefficient using eight sky state categories, addressing the limitation of traditional unified Perez coefficients in adapting to low-sunlight scenarios such as cloudy days and haze. It fully reconstructs the anisotropic brightness gradient of the sky and solar halo spots through high-precision 180×720 grid discretization. It distinguishes between scattering and direct sunlight components, modeling and superimposing them separately, while supporting both radiometric and photometric output modes. Through GHI global energy conservation iterative verification, it strictly constrains the total energy of the reconstructed sky field to be consistent with the measured meteorological input, ensuring the accuracy of subsequent physical calculations for street scene occlusion projection, human eye vertical illuminance, and glare quantitative assessment.

[0046] In some embodiments, based on the dual-domain physical quantity results, and using a Perez all-sky model modified according to sky type, an anisotropic all-sky absolute sky field containing direct solar radiation spots is discretized and reconstructed, including: S41, construct the Perez coefficient matrix and match the Perez coefficients corresponding to the brightness distribution parameters according to the sky category index; Based on the value range of the sky clarity parameter, the sky state is discretely divided into 8 predefined types, and the corresponding Perez coefficients are matched from the constructed Perez coefficient matrix. (Sky state category index) The Perez coefficients in the i-th row of the Perez coefficient matrix represent the set of Perez coefficients for the i-th type of sky state. This Perez coefficient matrix contains five brightness distribution parameters A, B, C, D, and E, which are used to describe the sky brightness distribution pattern, brightness gradient change, solar neighborhood halo intensity, halo range, and background brightness level, respectively.

[0047] Perez coefficient matrix Written as:

[0048] (twenty two) In equation (22), the i-th row corresponds to the set of Perez coefficients for the i-th type of sky state. The Perez coefficient matrix... The Perez coefficients in the i-th row are grouped into sets of four, corresponding to the linear combination coefficients of the luminance distribution parameters A, B, C, D, and E, respectively. That is, each luminance distribution parameter corresponds to a Perez coefficient matrix. Four consecutive Perez coefficients in a row form a quaternion coefficient set. For example, the first four Perez coefficients in the i-th row form the quaternion coefficient set corresponding to the luminance distribution parameter A; the fifth to eighth Perez coefficients form the quaternion coefficient set corresponding to the luminance distribution parameter B; and so on, with the 17th to 20th Perez coefficients used to calculate the luminance distribution parameter E. Thus, for any luminance distribution parameter X, it can be calculated from the corresponding quaternion coefficient set in the row of the Perez coefficient matrix corresponding to that sky state.

[0049] S42, Solve the brightness distribution parameters differently according to different clearness conditions, and correct the parameters under low clearness conditions to obtain a set of corrected parameters; Moderate to high sunshine levels, similar to partly cloudy skies. At that time, the five brightness distribution parameters were calculated using a linear form, and the formula was: (twenty three) In equation (23), X represents any one of the brightness distribution parameters A, B, C, D, or E. Indicates the zenith angle of the sun. Z represents the sky brightness parameter, and Z represents meteorological variables, such as the content of atmospheric precipitable water vapor. , This indicates the relationship between the Perez coefficient matrix and the Perez coefficient matrix. The quaternion coefficient set of brightness distribution parameters corresponding to the sky state category read from the database.

[0050] Low visibility, like overcast skies or smog. At this time, the luminance distribution parameters A, B, and E are still calculated in a linear form, while the luminance distribution parameters C and D are respectively calculated using the following exponential correction forms: (twenty four) (25) In equations (24) and (25), These are the quaternion coefficient sets for the brightness distribution parameters C and D corresponding to the sky state categories. An exponential function is used to enhance the uniform diffuse characteristics of cloudy skies, correcting the bias of the Perez all-sky model in low-clearance scenes.

[0051] S43, using zenith angle and azimuth angle as the dividing directions, the sky hemisphere is discretized based on a regular grid to obtain discrete sky grid cells; After obtaining five brightness distribution parameters, the sky hemisphere is discretized. This application uses a regular grid to divide the entire sky, dividing it into 180 equally spaced cells at the zenith angle and 720 equally spaced cells at the azimuth angle, thus constructing a discrete sky grid with a resolution of 180×720. Each sky grid cell corresponds to a specific spatial direction. and its solid angle ,in, The zenith angle represents the center of the sky grid cell. The azimuth angle of the center of the sky grid cell is represented by: (26) S44, For each of the sky grid cells, calculate the relative sky brightness distribution according to the set of correction parameters; First, calculate the position of the sun for each sky grid cell. The scattering angle between them, where Indicates the zenith angle of the sun. Indicates the solar azimuth angle and the scattering angle. The formula is: (27) Then, the modified parameter set, namely the brightness distribution parameters A, B, C, D, and E, is substituted into the relative sky brightness distribution function to calculate the relative sky brightness distribution. This function fully characterizes three types of sky optical features: zenith brightness gradient, solar halo, and background brightness in regions far from the sun. The relative sky brightness distribution function is expressed as: (28) S45, determine the scattering component according to the output mode and calculate the normalization factor, normalize the relative sky brightness distribution, and generate the scattered sky field. To ensure that the total energy of the all-sky scattering integral is consistent with the input scattering flux, a normalization factor N is introduced: (29) In equation (29), This represents the scattering component; when the output mode is radiance field, When the output mode is brightness field, The relative sky brightness distribution is normalized by a normalization factor, i.e., by multiplying by the normalization factor grid by grid, to generate a scattered sky field containing only the scattering component. The formula is: (30) S46, locate the sun corresponding to the sky grid cell, assign the direct sunlight component, and superimpose the scattered sky field to generate the absolute sky field of the whole sky.

[0052] First, locate the grid cell closest to the sun's position within the sky grid. This can be done using the minimum angular distance principle, which involves calculating the angle between the center direction vector of each sky grid cell and the sun's direction vector, and selecting the grid cell with the smallest angle as the cell for direct solar energy injection. For regular grids, an index approximation method can also be used. (31) (32) In equations (31) and (32), This indicates the index of the sky grid cell corresponding to the sun in the zenith angle direction. This indicates the index of the sky grid cell corresponding to the sun in the azimuth direction; This represents the distance of the sky grid cell in the zenith angle direction. This represents the distance between the sky grid cells in the azimuth direction. Indicates the zenith angle of the sun. It indicates the azimuth angle of the sun.

[0053] This allows us to locate the grid cell corresponding to the sun, which represents the grid cell closest to the sun's position, and obtain the center of that grid cell. ,in This represents the zenith angle at the center of the grid cell corresponding to the sun. This represents the azimuth angle of the center of the corresponding grid cell. Then, a direct sunlight component is assigned to the located grid cell, and the corresponding grid value is... It can be represented as: (33) In equation (33), This represents the solid angle of the grid cell corresponding to the sun; This represents the direct solar radiation component; when the output mode is radiance field... When the output mode is brightness field, .

[0054] By superimposing the direct solar component onto the scattered sky field, an absolute sky field for the entire sky is generated. The formula is: (34) In some embodiments, after generating the absolute sky field of the entire sky, the process includes: Construct a global horizontal radiation energy model to verify energy conservation. The energy conservation test includes: Calculate the total radiant energy of the reconstructed sky field on the horizontal plane; The theoretical value of total radiant energy is calculated based on meteorological data; The normalization factor is adjusted based on the deviation between the total radiant energy and the theoretical value of the total radiant energy until the energy conservation requirement is met, and the reconstructed sky field is output.

[0055] Specifically, to ensure the physical consistency of the inversion results, this application constructs a global horizontal radiation energy... Energy conservation is verified. On one hand, the total radiant energy of the sky field on the horizontal plane is statistically reconstructed using numerical integration methods. The formula is: (35) In equation (35), Indicates spatial direction The absolute sky field of the entire sky corresponding to the grid cell is output in radiometric mode. Indicates spatial direction The solid angle corresponding to the grid cell. On the other hand, the theoretical value of total radiant energy is calculated based on meteorological data. The formula is: (36) Set a deviation threshold r between the total radiant energy and the theoretical value of the total radiant energy. For example, if the deviation threshold r = 0.1%, and the deviation between the total radiant energy and the theoretical value of the total radiant energy is greater than the deviation threshold (e.g., r > 0.1%), return to step S45 to correct the normalization factor and iteratively recalculate the scattered sky field and the absolute sky field of the entire sky. If the deviation between the total radiant energy and the theoretical value of the total radiant energy is less than or equal to the deviation threshold (e.g., r ≤ 0.1%), it indicates that physical self-consistency is satisfied, the loop is exited, and the final 180×720 absolute brightness / radiance field of the entire sky is output.

[0056] It should be noted that the commonly used Perez model in existing technologies takes horizontal scattered radiation DHI, solar zenith angle, and atmospheric turbidity as inputs and outputs the relative brightness distribution of the sky hemisphere. To address the needs of adapting to low clear weather conditions in urban road glare assessment scenarios, discrete representation of direct sunlight spots, and consistency between the reconstructed sky field and meteorological input energy, this application's sky-type-modified Perez all-sky model is based on the existing Perez model. Improvements are made through segmented parameter correction for different atmospheric conditions, the addition of a low clear weather index correction, the addition of discrete grid direct sunlight overlay, and energy conservation iterative verification. This results in a multi-level segmented nonlinear mapping model, which is divided into four main levels: The first layer is the discrete sky classification mapping layer, used for sky type correction, i.e., based on the sky clarity parameter. Classify sky states into categories and output the sky category index. At the same time, an 8-row, 20-column exclusive Perez coefficient matrix is ​​constructed. Each row corresponds to a complete set of fitting coefficients for a type of sky, used to match parameter combinations under different sky conditions. The 20 Perez coefficients in each row are split into groups of 4, corresponding to the quaternion coefficient groups required for the linear calculation of brightness distribution parameters A, B, C, D, and E respectively.

[0057] The second layer is the parameter solution layer for high and low sunshine levels, used to correct parameters for low sunshine levels, i.e., for medium and high sunshine levels. At that time, the brightness distribution parameters A, B, C, D, and E are uniformly mapped linearly; for low sunshine levels, i.e. At this time, the brightness distribution parameters A, B, and E are still linearly mapped, while C and D are exponentially corrected to adapt to the optical characteristics of cloudy days with no obvious solar halo and uniform diffuse sky height. By compressing the contribution of the solar halo on cloudy days through the exponential function, the linear formula is avoided from excessively amplifying the brightness around the sun, which greatly improves the simulation accuracy of cloudy day scenes.

[0058] The third layer is a discrete grid brightness field generation and mapping layer, used to reconstruct the sky field. It constructs a 180×720 high-precision sky grid cell, calculates the scattering angle cell by cell, and substitutes the corrected A, B, C, D, and E to obtain the relative sky brightness. The normalization factor is determined according to the input mode to obtain the scattered sky field. After superimposing the direct solar component, a complete all-sky absolute brightness field containing strong solar light is generated. It can simultaneously output the radiance field (radiometry, used for disabling glare / retinal irradiance) and the brightness field (photometry, used for discomfort glare / sunlight glare probability).

[0059] The fourth layer is the energy conservation closed-loop verification iteration layer, which is used for physical self-consistent correction mapping. That is, the global horizontal radiation energy GHI is constructed for energy conservation verification. Based on the deviation between the reconstruction result of GHI and the theoretical value, the normalization factor is iteratively corrected in reverse, and the sky field is regenerated until the energy conservation constraint is met. This solves the physical defects of the traditional Perez model, such as the lack of energy closed loop and the mismatch between the reconstructed radiation flux and the measured meteorological data.

[0060] The complete mapping chain of the Perez all-sky model with sky type correction includes: the first-level discrete piecewise mapping "clearness". →Sky Category Index → Exclusive Perez coefficient; the second layer is a piecewise nonlinear mapping of "multiple meteorological parameters, exclusive Perez coefficient → 5 brightness distribution parameters A, B, C, D, E"; the third layer is a spatial geometric mapping of "grid zenith angle, azimuth angle, scattering angle → single grid relative brightness"; the fourth layer is an energy-normalized linear mapping of "single grid relative brightness → scattered sky field"; the fifth layer is a superposition mapping of "scattered sky field, direct solar component → absolute sky field"; the sixth layer is a closed-loop correction mapping for reconstructing radiation bias inverse correction normalization coefficients, which converges iteratively.

[0061] S5, stitch together multi-view road network street scene images to generate a panoramic image, perform pixel-level semantic segmentation and fisheye projection transformation on the panoramic image to generate a binary fisheye sky visibility mask. In some embodiments, a panoramic image is generated by stitching together multi-view street view images of the road network, and pixel-level semantic segmentation and fisheye projection transformation are performed on the panoramic image to generate a binary fisheye sky visibility mask, including: S51, the multi-view road network street scene images are stitched together to generate an equidistant cylindrical projection panoramic image, and a calibration image is obtained after orientation calibration. This application fuses and stitches street view images from four perspectives relative to the direction of travel to generate an equidistant cylindrical projection panoramic image covering a 360° field of view. To ensure consistency between the sun's position and the actual spatial orientation in subsequent calculations, the panoramic image is calibrated for azimuth. Based on the azimuth metadata inherent in the street view images, the column index position corresponding to true north in the panoramic image is determined, and the panoramic image is cyclically translated, moving all pixels located on the left side of true north to the right side of the panoramic image. This establishes a strict monotonic correspondence between the horizontal coordinates of the panoramic image and the azimuth, ensuring that the azimuth continuously increases within the range of [0°, 360°], ultimately yielding the azimuth-calibrated panoramic image, i.e., the calibration image.

[0062] S52, based on the deep learning semantic segmentation method, the semantic category prediction of the calibration image is performed, and a panoramic binary sky mask is generated after assignment. To obtain the effective unobstructed sky area in the urban environment, this application employs a deep learning semantic segmentation method to process the calibration image. For example, by introducing a DeepLabv3+ model pre-trained on the Cityscapes dataset, the calibration image is classified at the pixel level, dividing the scene into semantic categories such as sky, buildings, vegetation, roads, vehicles, and pedestrians. Pixels labeled as sky represent unobstructed areas where light can propagate directly, while the categories of buildings, vegetation, and other opaque objects correspond to the geometric occlusion boundaries of light propagation.

[0063] The pixel-level classification process mainly includes: scaling the calibration image to the standard input resolution specified by the DeepLabv3+ model to match the input channels of the model's convolutional network and the image size requirements; and performing standard normalization on all pixels of the scaled image according to the preset image mean and standard deviation to eliminate input bias caused by differences in illumination and imaging brightness. The preprocessed image is then input into the pre-trained DeepLabv3+ semantic segmentation network. This network uses dilated convolution and ASPP (Spatial Pyramid Pooling) modules to extract multi-scale image features, independently performing multi-class predictions for each pixel, and outputting a pixel category prediction heatmap of the same size as the scaled image. For example, each pixel corresponds to the classification confidence of six semantic categories: sky, buildings, vegetation, roads, vehicles, and pedestrians.

[0064] Then, upsampling interpolation is performed on the low-resolution predicted heatmap output by the model inference to restore it to the original resolution of the original panoramic calibration image. The semantic category with the highest confidence is taken as the final label for each pixel, generating a semantic label map that corresponds one-to-one with the pixels of the panoramic image. Each pixel in the semantic label map is assigned only a single category label, such as one of sky, building, vegetation, road, vehicle, or pedestrian.

[0065] Next, all pixels in the semantic label image are traversed and binary values ​​are assigned. If the pixel label is sky, the pixel value is assigned as 1, indicating that there is no obstruction in that direction and the sky light can reach the human eye directly. If the pixel label is a non-sky-obstructing object such as a building, vegetation, road, vehicle, or pedestrian, the pixel value is assigned as 0, indicating that there is geometric obstruction in that line of sight and the sky radiation cannot penetrate. The sky brightness in the corresponding direction is set to zero. This is used to accurately deduct the glare contribution of the obstructed area and improve the accuracy of glare assessment in complex urban built-up areas. After the assignment is completed, a panoramic binary sky mask is obtained.

[0066] Before subsequent joint calculations with the 180×720 sky grid cells, the panoramic binary sky mask is resampled and aligned. The panoramic mask is resampled according to the angular resolution of the full sky model, so that the mask rows and columns strictly correspond one-to-one with the sky grid cells, i.e., 180 rows at the zenith angle and 720 columns at the azimuth angle. This ensures that each sky grid cell can read the corresponding location's visible and occluded binary sky identifiers, which are used for subsequent occlusion removal and calculation of the effective sky irradiance integral for the human eye.

[0067] S53, by establishing a reverse projection coordinate mapping relationship, fisheye projection transformation is performed on the panoramic binary sky mask to obtain a binary fisheye sky visibility mask.

[0068] The human eye's field of view when observing the sky is a hemispherical fisheye wide-angle field of view, where the center corresponds to the zenith, the surrounding areas correspond to the horizon, and the radial distance corresponds to the elevation angle. However, step S51 generates an equidistant cylindrical panoramic image, and the spatial geometric mapping rules of the two are completely different. To simulate the distribution of the human eye's field of view in a real driving environment, since only the upper hemisphere has sky and generates glare, this application only extracts the upper half of the panoramic binary sky mask. Through the constructed inverse projection coordinate mapping relationship, the cylindrical panoramic binary mask is converted into a fisheye format binary mask that fits the real field of view of the human eye, resulting in a binary fisheye sky visibility mask. This mask is then used for subsequent pixel-by-pixel and grid-by-grid coupling with the 180×720 full sky brightness field to accurately calculate the irradiance and glare contribution of the unobstructed sky for the human eye.

[0069] The image size of the upper half of the panoramic binary sky mask is defined as width W and height H. The width W of the panoramic image corresponds to a complete 0-2π omnidirectional angle. After conversion, the maximum radial boundary of the fisheye projection image is obtained, corresponding to the horizon elevation angle position, and the radius of the fisheye projection image is also calculated. Defined as: (37) The center coordinates of the fisheye projection image correspond to the zenith center, and the radial outward tilt angle gradually decreases until it reaches the horizon. Represented as: (38) Any pixel in a fisheye projection image The larger the radial distance r to the center coordinate, the lower the corresponding elevation angle, and the closer to the horizon. r=0 represents the zenith. The radial distance r is expressed as: (39) The corresponding azimuth angle θ is calculated based on the pixel's position relative to the center, using the following formula: (40) By using the inverse projection coordinate mapping relationship, the fisheye coordinates are converted back to the panoramic image coordinates. The formula is: (41) The entire fisheye projection image is traversed pixel by pixel. The corresponding position of the upper half of the panoramic binary sky mask is obtained through the inverse projection coordinate mapping relationship. The corresponding binary value is read, i.e., 0 for occlusion and 1 for sky. This 0 / 1 value is then assigned to the current fisheye pixel. After traversing all pixels, output the complete binary fisheye sky visibility mask. In this way, the binary sky mask of the upper half of the panoramic image is resampled into a binary fisheye sky visibility mask. In this fisheye image, according to the above mapping relationship, it can be seen that in this fisheye image, the image center point corresponds to the sky zenith, the radial direction of the image is the distance from the pixel to the center corresponding to the change of elevation angle, the closer to the image boundary is to the horizon, the image circumferential angle direction corresponds to the horizontal azimuth angle, and the 720-column azimuth grid of the full sky model is matched.

[0070] S6. Based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky, the vertical illuminance of the human eye is calculated by constructing a projection factor. In some embodiments, the vertical illuminance of the human eye is calculated by constructing a projection factor based on the binary fisheye sky visibility mask and the absolute sky field of the entire sky, including: S61, constructs the glare perception field of the human eye based on the field of view corresponding to the main viewing direction; To simulate the actual reception characteristics of ambient light by the human eye, this application defines the glare receptive field of the human eye, assuming the azimuth angle of the line of sight is the primary viewing direction, the vertical field of view is [0°, 29°], and the horizontal field of view is 116°, i.e., 58° to the left and right of the line of sight center. This range defines the spatial area that can enter the human eye and significantly affect vision, thus providing a clear geometric boundary for subsequent radiation integration. In single-view calculations, the corresponding field of view can be directly constructed based on the current primary viewing direction to complete subsequent illuminance and glare calculations. When multi-view or large-scale time-series calculations are required, a standard viewpoint can be introduced as a reference viewpoint for pre-calculation to improve computational efficiency. The standard viewpoint can be the azimuth angle of the primary viewing direction. In such cases, a field mask and corresponding grid index are pre-generated for the reference viewpoint; for any actual observation direction, the set of sky grid cells in the corresponding field of view is quickly obtained by azimuth index translation.

[0071] S62, construct the vertical viewing plane normal vector and incident light projection within the glare sensing field, and calculate the projection factor; After determining the glare receptive field, a vertical receiving plane for the human eye, i.e., the vertical viewing plane, is constructed. Its normal vector is used to describe the spatial orientation of light reception. Let the azimuth angle of the main viewing direction be... Then the normal vector perpendicular to the view plane It can be represented as: (42) For any sky grid cell within the field of view, the spatial direction vector of the incident light direction It can be represented as: (43) The projection factor is obtained by calculating the cosine of the angle between the spatial direction vector of the incident light and the normal vector perpendicular to the viewing plane through the vector dot product. The formula is: (44) From a standard perspective, the direction cosine can be further simplified to: (45) In equation (45), This represents the angle between the incident direction of the sky grid cell and the normal vector of the viewing plane. This projection factor is used to characterize the effective contribution of light rays from different incident directions to the vertical viewing plane.

[0072] S63, based on the projection factor, perform grid-weighted integration using the binary fisheye sky visibility mask to calculate the vertical illuminance of the human eye.

[0073] This step takes the full-sky absolute sky field reconstructed in S4 and the binary fisheye sky visibility mask generated in S5 as inputs. According to the inverse projection coordinate mapping relationship, the angular direction of each 180×720 sky grid cell is mapped to the fisheye projection image plane and the corresponding occlusion mark is read. The invalid sky radiation in the area blocked by buildings and vegetation is removed by the binary screening of the mask, so as to obtain the effective radiance field that retains only the visible area. Then, a weighted discrete integral is performed on all visible sky grids of the human eye, and the effective radiance, human eye optical projection weight, and grid solid angle are fused to realize flux accumulation. At the same time, the vertical illuminance of the human eye, including the total contribution of diffuse and direct rays, and the vertical illuminance of only the direct solar component are solved. This realizes the geometric mapping and occlusion correction quantization conversion from the anisotropic three-dimensional sky light field to the illuminance of the human eye's two-dimensional receiving plane.

[0074] Specifically, according to the inverse projection coordinate mapping relationship in step S53, the angular coordinate direction corresponding to each sky grid cell is mapped to the fisheye projection image plane to obtain the corresponding pixel coordinates. Then read the sky visibility mask value at that location. If the mask value is 1, retain the original radiance of that sky grid cell. If the mask value is 0, the effective radiance of that sky grid cell is recorded as 0. This yields the effective radiance field considering building and vegetation shading: (46) Then, a weighted integral is performed on all unobstructed sky grid cells within the glare receptive field. Combining their radiance, solid angle, and projection factor, the total illuminance obtained by the human eye at the vertical receiving plane after the superposition of diffuse sky light and direct sunlight is calculated. The unit is The formula is: (47) To characterize the visible direct sunlight component within the field of view, the vertical illuminance corresponding to the visible direct sunlight component is calculated. The formula is: (48) In equations (46), (47), and (48), i is the index number of the discrete sky grid cell. This represents the effective radiance of the i-th sky grid cell after occlusion correction; This represents the original radiance of the i-th unobstructed sky grid cell. This indicates that the sky grid cells are mapped to the fisheye projection plane pixels. The binary sky visibility mask value at the location, which takes the value 0 or 1; It represents the projection factor of the i-th sky grid cell corresponding to the vertical plane of the light incident on the human eye, and characterizes the attenuation weight of the light incident angle on the received flux; This represents the solid angle corresponding to the i-th sky grid cell; This represents the discrete radiance value corresponding to the direct solar component within the i-th sky grid cell.

[0075] S7. Based on the vertical illuminance of the human eye and coupled with the visual mechanism, a dynamic glare risk assessment is performed.

[0076] In some embodiments, the glare includes discomfort glare and disability glare; When assessing the risk of uncomfortable glare, the total illuminance and effective luminance are coupled, based on the probability of sunlight glare. The model calculates the probability of sunlight glare, and then combines it with preset grading thresholds to classify the risk level of uncomfortable glare; This application uses a dual evaluation system to quantitatively analyze glare risk, targeting psychological discomfort glare by employing the probability of sunlight glare. The model performs calculations, The model takes into account both the saturation effect of vertical illuminance to the human eye and the brightness contrast effect of the sky grid cells within the field of view. Represented as: (49) In equation (49), Indicates total illuminance. Indicates the first in the field of view Effective brightness of each sky grid cell; Indicates the first The solid angle corresponding to each sky grid cell represents the spatial area occupied by that sky region within the human field of vision, and can also be written as... ; Indicates the first The position index of each sky grid cell is used to characterize the degree of deviation of the sky grid cell from the center of the line of sight. The smaller the value, the smaller the deviation, and the greater the contribution to glare.

[0077] In calculating the position index, the angular distance between the center of the sky grid cell and the center of the line of sight is first defined as... Define the auxiliary angle as Then we have: (50) (51) Then introduce intermediate quantities The location index is calculated using the following formula: (52) (53) In equations (50)(51)(52)(53), Indicates the first The angular distance between the center of each sky grid cell and the center of the line of sight; This represents the projection factor of the i-th sky grid cell corresponding to the plane perpendicular to the human eye where the light rays are incident; This represents the auxiliary angle parameter used to correct the exponential decay law of position; This represents the zenith angle of the i-th sky grid cell; This represents the azimuth angle of the i-th sky grid cell relative to the center of the line of sight; The natural logarithm of the position index is used to describe the weighted decay of unpleasant glare from light sources at different positions in the field of view. This indicates that the original position index has been obtained after restoration; Indicates lower bound truncation, forced Not less than Used to prevent sky grid cells from being extremely far from the center of the line of sight. Approaching 0, causing fractions Numerical explosion, computational overflow.

[0078] When the elevation angle of the sky grid cell approaches the horizon, A larger value is assigned to suppress boundary numerical anomalies. Based on the calculated... The values, with 0.35, 0.40 and 0.45 as grading thresholds, can classify uncomfortable glare into four levels: imperceptible, perceptible, disturbing and intolerable.

[0079] When assessing the risk of disability glare, the vertical illuminance received at the cornea is converted into retinal irradiance based on the human eye imaging optical model, and the risk level of disability glare is classified in combination with a preset risk threshold.

[0080] To address the physiological aspect of disability glare, this application introduces a human eye imaging optical model to convert the vertical illuminance received at the cornea into irradiance on the retinal surface. First, the retinal imaging diameter is calculated: (54) Then calculate the retinal irradiance. : (55) In equations (54) and (55), Where is the retinal imaging diameter, and f is the intraocular imaging focal length. Angle subtended by the sun's light source; Vertical illumination at the cornea The diameter of the pupil. This refers to the transmissivity of the ocular media.

[0081] Based on the calculated retinal irradiance index, when At that time, it was determined that there was a potential risk of permanent eye damage; when When, it is classified as a high-risk level; when When, it is classified as a medium-risk level; when At that time, it was classified as a low-risk level.

[0082] Based on the results of physical inversion and visual assessment, a high spatiotemporal resolution glare risk dataset covering the target urban road network is generated and output in the form of spatial visualization. Depending on the application requirements, the output results include instantaneous glare distribution maps at typical times and comprehensive glare risk distribution maps at long-term scales (such as throughout the year), enabling accurate identification of high-risk road sections and time periods.

[0083] Please see Figure 3 As shown, another aspect of this application embodiment provides a solar glare assessment system that combines all-sky radiation and street view photographs, the system comprising: The data acquisition module is used to acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network. The data calculation module is used to calculate the solar position parameters and sky state parameters based on the meteorological data; The dynamic optical performance conversion module is used to construct a type-specific dynamic optical performance empirical model, and combine the radiation flux component, the solar position parameter and the sky state parameter to solve the dynamic conversion coefficient from the radiation domain to the luminosity domain, and output the dual-domain physical quantity results. The sky field reconstruction module is used to reconstruct the anisotropic all-sky absolute sky field containing direct sunlight spots by discretizing and meshing based on the results of the dual-domain physical quantities and the Perez all-sky model modified according to sky type. The projection transformation module is used to stitch together multi-view road network street scene images to generate a panoramic image, and to perform pixel-level semantic segmentation and fisheye projection transformation on the panoramic image to generate a binary fisheye sky visibility mask. The vertical illuminance calculation module is used to calculate the vertical illuminance of the human eye by constructing a projection factor based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky. The glare risk assessment module is used to perform dynamic glare risk assessment based on the vertical illuminance of the human eye and coupled with the visual mechanism.

[0084] Please refer to Figure 4-8, which shows the glare risk distribution map for the entire year, taking Guangzhou as an example, and the glare results at two times on March 21st. Specifically, Figure 4aThis view displays a panoramic view with azimuth and elevation angles. The horizontal axis represents the azimuth angle, extending from 0° north through 180° south back to 360° north. The vertical axis represents the elevation angle, from the 0° horizon to the 90° zenith. The view presents a realistic view of the city's buildings against the open sky. Two horizontal dashed lines mark reference lines for different elevation angles, visually showing the distribution of sky obstruction by city buildings from different azimuth angles.

[0085] Figure 4b This diagram represents a hemispherical sky light geometry image, with a hemisphere representing the complete sky field of view. It marks the sun's position, glare source, driving vehicle, and driving direction, visually demonstrating the incident geometric relationship of light rays from the hemispherical sky within the vehicle's field of view, and explaining the spatial physical mechanism by which the human eye receives direct and scattered light from the sky and generates glare in road driving scenarios.

[0086] Figure 4c It is a real-life fisheye panoramic photo that fully records the sky and surrounding buildings within the upward hemisphere's field of view; on the right is a corresponding polar coordinate angle division diagram, which marks the North (N), East (E) directions and zenith angles such as 30° and 60°, realizing the correspondence between the real-life image and the polar coordinate angle system, facilitating quantitative analysis of the field of view by dividing it into angle grids.

[0087] Figure 4d This is a polar coordinate cloud image of the entire sky brightness under clear weather at 13:40 on January 3. The polar coordinate arrangement is matched with the hemispherical sky field of view, and the color gradient represents the magnitude of the sky brightness value. The observation parameters for this period are marked in the lower left corner. Under clear weather, the area where the sun is located shows a clear high brightness concentration area, while the brightness distribution of other sky areas is relatively gentle, which can clearly show the spatial distribution characteristics of the brightness of the clear sky.

[0088] Figure 4e This is a polar coordinate cloud image of the entire sky at 15:00 on January 2nd, under cloudy weather conditions. It uses polar coordinate format, showing that the cloud cover causes a large-scale, gradually varying, and uneven distribution of sky brightness under cloudy conditions. High-brightness areas are more widely distributed as the cloud clusters extend. The corresponding observation parameters for the time period are also labeled, allowing for comparison with images taken under clear weather conditions. Figure 4d A comparative analysis of sky brightness distribution under varying cloud cover conditions.

[0089] Figure 5aThis is a schematic diagram of the cumulative duration of uncomfortable glare throughout the year at the scale of the entire road network of Guangzhou. Dark gray is used as the urban base, and the road network is represented by four levels of color: green, yellow, orange, and red to represent the annual gradient of uncomfortable glare duration. The unit of the legend is h / yr. Green road sections have the lowest duration of uncomfortable glare throughout the year, while red road sections have the highest risk. Overall, it can be seen that the distribution of red high-risk road sections is more concentrated in the high-density built-up areas of the central urban area such as Tianhe, Yuexiu, and Haizhu. Suburban roads are mainly low-risk green and yellow. It can intuitively reflect the spatial distribution characteristics of the city's high-rise building clusters with less obstruction and open road sections that are prone to high probability of subjective glare. The accompanying scale and compass complete the spatial orientation and distance calibration.

[0090] Figure 5b This is a schematic diagram showing the cumulative duration of disability glare throughout the year on a city-wide road network scale in Guangzhou. Figure 5a A consistent base map, road network architecture, and color grading logic are used, but the legend duration threshold is significantly reduced, and focusing on the representation causes physiological disability glare that reduces visual resolution; the distribution range and density of high-value red road segments within the map are significantly less than... Figure 5a Uncomfortable glare was observed in only a few unobstructed main roads and open north-south sections, with orange and red high-risk areas. Most urban roads were predominantly green or low yellow, indicating that disabling glare requires a higher level of unobstructed vision under direct sunlight. Significant visual impairment risk only occurs on roads with specific orientations and low building obstructions. Figure 5a A comparative analysis of spatial differences in glare risk between the two categories was conducted.

[0091] Figure 6a This is a spatial distribution map of the annual average vertical irradiance of the human eye across the entire road network of Guangzhou. A gray background map represents the city's administrative districts, waterways, and complete road network, while a compass and kilometer scale are used for spatial positioning. Five color levels—green, light yellow, yellow, orange, and red—represent the annual average gradient of retinal irradiance. Green road sections have the lowest average irradiance, while red road sections represent the sections with the highest risk of retinal light damage from direct sunlight throughout the year. High-value orange-red road sections are densely distributed in the core urban built-up areas such as Tianhe, Yuexiu, and Haizhu districts. Roads in the suburbs and those with ample vegetation and building cover are predominantly green and light yellow. This visually demonstrates that high-density urban open main roads and low-shading sections accumulate high-intensity retinal irradiance over a long period, resulting in a significantly higher risk of visual impairment due to disabling glare.

[0092] Figure 6b This is a spatial distribution map of the average solar glare probability (DGP) of roads throughout Guangzhou, and... Figure 6a With the same geographical boundaries and road network base, the color grading logic is the same as... Figure 6aA unified probability scale with values ​​ranging from 0% to 70% is used, with green representing a low probability of uncomfortable glare, and red representing road sections with DGP values ​​close to 70%, indicating extremely strong subjective glare discomfort. Orange-red high-risk sections are concentrated on main and secondary roads in the central urban area and east-west roads without tall buildings obstructing the view. Densely built-up areas in older cities are mostly in the green low-probability range. This scale clearly reflects road direction, surrounding building height, and the degree of visual obstruction, directly determining the probability of subjective uncomfortable glare during driving. Figure 6a The spatial distribution differences between the two types of glare risk were compared.

[0093] Figure 7a This is a schematic diagram of glare risk across Guangzhou's entire city in the eastward direction. It uses dark gray as the city's geographical base map, clearly showing the outlines of waterways, administrative districts, and the dense urban road network. It is accompanied by a compass and a 0-15km scale to ensure spatial orientation and distance judgment. A gradient of green, light yellow, orange, and red is used to distinguish different levels of glare accumulation duration. Green represents low-risk road sections, and red represents high-risk road sections. The distribution of high-value warm colors is more concentrated on the main roads in the central urban area and open roads along the river. The side roads in the old city with sufficient building cover are mostly green low-risk areas, which intuitively shows the spatial differences in glare risk of different road networks during the morning rush hour.

[0094] Figure 7b This is a schematic diagram of glare risk across all roads in Guangzhou in the overall westward direction, and... Figure 7a The base map, color scheme, scale, and orientation system are consistent; the overall distribution range and density of high-risk warm-toned road sections are compared to... Figure 7a Significant changes have occurred, with the red high-risk areas expanding in some east-west oriented, unobstructed sections of roads near water. This reflects the shift in glare risk patterns caused by changes in road lighting conditions following the change in the sun's position at sunset. Figure 7a Compare and analyze the differences in glare distribution patterns caused by variations in the sun's position.

[0095] Figure 8a This is a map showing the glare risk of the entire road network in Guangzhou at 07:30 AM on March 21st. Dark colors represent the city's geographical base, water system, and administrative boundaries. Spatial positioning is achieved using a scale and compass. Different colors are used to distinguish multiple levels of glare risk indicators. The overall proportion of blue tones is higher, and the distribution of high-risk warm-toned road sections is relatively sparse. This reflects that under the characteristics of solar altitude and orientation, the glare intensity of most roads in the urban area is relatively low. Only some main roads along the river and without building obstruction show medium- to high-risk sections, presenting the spatial distribution characteristics of glare risk.

[0096] Figure 8b This is a glare risk map of the entire road network in Guangzhou at 16:30 on March 21st. Figure 8aThe base map framework, orientation markers, and color scheme are consistent; the coverage and density of high-risk colors such as red and pink in the image are significantly higher than... Figure 8a The number of high-risk sections on main roads and open riverside roads in the central urban area has increased significantly. This is because changes in solar altitude and sunshine duration have intensified the impact of direct sunlight on roads, increasing the overall risk of glare. Figure 8a Analyze the differences in spatial patterns of glare.

[0097] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0098] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0099] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0100] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the methods described in the embodiments of this application. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0101] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0102] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0103] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0104] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0105] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0106] This application provides a method for assessing solar glare by combining all-sky radiation and street view photographs. The method extracts radiative flux and dew point temperature from meteorological data and acquires multi-view street view images of the road network. It calculates the sun's position and sky state characteristics based on meteorological parameters. A dynamic optical visibility efficiency empirical model is constructed for different sky types, and multiple meteorological parameters are coupled to solve for the dynamic conversion coefficient from the radiative domain to the photometric domain, simultaneously outputting radiative and photometric dual-scale data. Anisotropic all-sky fields containing direct sunlight spots are discretized and reconstructed based on a modified Perez model for different sky types. Street view images are stitched, semantically segmented, and transformed using fisheye projection to generate a binary sky visibility mask matching the human eye's hemispherical field of view. The vertical illuminance of the human eye after building and vegetation shading correction is solved by integrating the sky light field and projection factor. Finally, the method couples with the human eye's visual physiological mechanisms, simultaneously using a DGP model to assess subjective discomfort glare and a human eye imaging optical model to quantify retinal irradiance to characterize disabling glare, achieving a multi-time-period, multi-seasonal, dual-dimensional dynamic quantitative assessment of urban road network glare risk. This scheme takes into account both complex atmospheric optical variations and the shading effect of urban built-up areas, effectively reducing simulation errors of traditional static models and providing quantitative support for urban road lighting environment optimization and traffic safety planning. Furthermore, this application also has the following beneficial effects: 1. This application constructs a full-link physical inversion framework from the atmosphere to the human eye, which can break through the bottleneck of cross-scale light environment simulation. By innovatively unifying and coupling atmospheric radiation transmission process, urban geometric shading and human eye physiological optical mechanism, it realizes a complete physical downscaling mapping from macroscopic meteorological radiation parameters to microscopic human eye received energy, which significantly improves the physical consistency and interpretability of light environment simulation.

[0107] 2. This application introduces a meteorological-driven anisotropic all-sky model, which can significantly improve the inversion accuracy under complex weather conditions. Based on meteorological reanalysis data, the all-sky brightness distribution is dynamically reconstructed, which can effectively characterize the anisotropic scattering characteristics under the influence of aerosols and clouds, overcome the distortion problem of the traditional isotropic assumption under complex weather conditions, and thus significantly improve the accuracy of glare risk assessment under complex weather conditions such as cloudy weather.

[0108] 3. This application utilizes vehicle-mounted street view images and deep learning semantic segmentation methods to accurately extract the visible field of the urban sky, and achieves pixel-level coupling between the visible field of the urban sky and the radiation field through spatial projection, effectively replacing the traditional high-cost 3D modeling method, and significantly improving the scalability and computational efficiency of urban-scale applications while ensuring accuracy.

[0109] 4. This application couples physiological optics with dynamic photometric efficiency conversion mechanism, which can realize multidimensional quantitative assessment of glare risk. By introducing atmospheric precipitable water vapor content to nonlinearly correct the photometric conversion process, and combining it with human eye imaging optical model, the physical radiation quantity is directly converted into physiologically significant indicators such as the probability of uncomfortable glare and retinal irradiance. This realizes the direct mapping from physical quantity to visual perception, and improves the scientificity and application value of the assessment results.

[0110] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0111] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0112] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0113] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0114] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for assessing solar glare by combining all-sky radiation and street view photographs, characterized in that, The method includes: Acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network; Calculate the solar position parameters and sky state parameters based on the meteorological data; A type-specific dynamic optical performance empirical model is constructed, and the dynamic conversion coefficient from the radiation domain to the luminosity domain is solved by combining the radiation flux component, the solar position parameter and the sky state parameter, and the dual-domain physical quantity results are output. Based on the dual-domain physical quantity results, the anisotropic all-sky absolute sky field containing direct solar radiation spots is reconstructed by discretized gridding based on the Perez all-sky model with sky type correction. A panoramic image is generated by stitching together multi-view road network street scene images. The panoramic image is then subjected to pixel-level semantic segmentation and fisheye projection transformation to generate a binary fisheye sky visibility mask. Based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky, the vertical illuminance of the human eye is calculated by constructing a projection factor. Dynamic glare risk assessment is performed based on the vertical illuminance of the human eye and coupled with visual mechanisms.

2. The method according to claim 1, characterized in that, The solar position parameters include solar altitude angle, solar azimuth angle, solar declination angle, and solar hour angle; the sky state parameters include atmospheric mass, sky brightness parameter, sky clarity parameter, and atmospheric precipitable water vapor content, wherein the atmospheric precipitable water vapor content is calculated based on dew point temperature.

3. The method according to claim 1, characterized in that, A type-specific dynamic optical performance empirical model is constructed, and the dynamic conversion coefficient from the radiation domain to the luminosity domain is solved by combining the radiative flux components, the solar position parameters, and the sky state parameters. The dual-domain physical quantity results are output, including: Sky conditions are categorized based on the clarity parameter, and a corresponding sky category index is matched for each sky condition category. Based on the sky category index, match the scattering coefficient and direct radiation coefficient corresponding to the dynamic light performance empirical model; Based on the scattering coefficient and the direct illuminance coefficient, the ratio of scattered light illuminance to direct light illuminance is calculated according to the light performance formula. Based on the ratio of scattered illuminance to direct illuminance, the horizontal scattered illuminance and normal direct illuminance are obtained by inversion of radiance and illuminance. By combining radiometric and photometric parameters, the results of physical quantities in both domains are output.

4. The method according to claim 1, characterized in that, Based on the aforementioned dual-domain physical quantity results, and using the Perez all-sky model modified according to sky type, the anisotropic all-sky absolute sky field containing direct solar radiation spots is reconstructed using discretized gridding, including: Construct the Perez coefficient matrix and match the Perez coefficients corresponding to the brightness distribution parameters based on the sky category index; The brightness distribution parameters are solved differently based on different clearness conditions, and the parameters are corrected under low clearness conditions to obtain a set of corrected parameters. Using zenith angle and azimuth angle as the dividing directions, the sky hemisphere is discretized based on a regular grid to obtain discrete sky grid cells; For each of the sky grid cells, the relative sky brightness distribution is calculated based on the set of correction parameters; The scattering components are determined according to the output mode and the normalization factor is calculated. The relative sky brightness distribution is then normalized to generate a scattered sky field. The solar direct component is assigned to the corresponding sky grid cell of the sun, and the scattered sky field is superimposed to generate the absolute sky field of the whole sky.

5. The method according to claim 4, characterized in that, After generating the absolute sky field for the entire sky, it includes: Construct a global horizontal radiation energy model to verify energy conservation. The energy conservation test includes: Calculate the total radiant energy of the reconstructed sky field on the horizontal plane; The theoretical value of total radiant energy is calculated based on meteorological data; The normalization factor is adjusted based on the deviation between the total radiant energy and the theoretical value of the total radiant energy until the energy conservation requirement is met, and the reconstructed sky field is output.

6. The method according to claim 1, characterized in that, A panoramic image is generated by stitching together multi-view street scene images of the road network. The panoramic image is then subjected to pixel-level semantic segmentation and fisheye projection transformation to generate a binary fisheye sky visibility mask, including: The multi-view road network street scene images are stitched together to generate an equidistant cylindrical projection panoramic image, and a calibration image is obtained after orientation calibration. The calibration image is used to predict the semantic category based on a deep learning semantic segmentation method, and a panoramic binary sky mask is generated after assignment. By establishing a reverse projection coordinate mapping relationship, a fisheye projection transformation is performed on the panoramic binary sky mask to obtain a binary fisheye sky visibility mask.

7. The method according to claim 1, characterized in that, Based on the binary fisheye sky visibility mask and the absolute sky field of the entire sky, the vertical illuminance of the human eye is calculated by constructing a projection factor, including: Based on the field of view corresponding to the main viewing direction, construct the glare perception field of the human eye; Construct the vertical viewing plane normal vector and the incident light projection within the glare receptive field, and calculate the projection factor; Based on the projection factor, and combined with the binary fisheye sky visibility mask, a grid-weighted integral is performed to calculate the vertical illuminance of the human eye.

8. The method according to claim 1, characterized in that, The glare includes both discomfort glare and disability glare; When assessing the risk of uncomfortable glare, the total illuminance and effective luminance are coupled, based on the probability of sunlight glare. The model calculates the probability of sunlight glare, and then combines it with preset grading thresholds to classify the risk level of uncomfortable glare. When assessing the risk of disability glare, the vertical illuminance received at the cornea is converted into retinal irradiance based on the human eye imaging optical model, and the risk level of disability glare is classified in combination with a preset risk threshold.

9. A solar glare assessment system combining all-sky radiation and street view photographs, applied to the solar glare assessment method according to any one of claims 1-8, characterized in that, The system includes: The data acquisition module is used to acquire meteorological data and extract radiation flux components and dew point temperature, as well as acquire multi-view street view images of the road network; The data calculation module is used to calculate the solar position parameters and sky state parameters based on the meteorological data; The dynamic optical performance conversion module is used to construct a type-specific dynamic optical performance empirical model, and combine the radiation flux component, the solar position parameter and the sky state parameter to solve the dynamic conversion coefficient from the radiation domain to the luminosity domain, and output the dual-domain physical quantity results. The sky field reconstruction module is used to reconstruct the anisotropic all-sky absolute sky field containing direct sunlight spots by discretizing and meshing based on the results of the dual-domain physical quantities and the Perez all-sky model modified according to sky type. The projection transformation module is used to stitch together multi-view road network street scene images to generate a panoramic image, and to perform pixel-level semantic segmentation and fisheye projection transformation on the panoramic image to generate a binary fisheye sky visibility mask. The vertical illuminance calculation module is used to calculate the vertical illuminance of the human eye by constructing a projection factor based on the binary fisheye sky visibility mask and the absolute sky field of the whole sky. The glare risk assessment module is used to perform dynamic glare risk assessment based on the vertical illuminance of the human eye and coupled with the visual mechanism.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 8.

Images