Classroom real-time light environment detection system based on cloud-side cooperation

Abstract

The invention relates to a classroom real-time light environment detection system based on cloud edge collaboration, and belongs to the technical field of light environment detection. The system comprises a model simulation module, a video preprocessing module, a target detection module and an illumination adjustment module. The model simulation module is used for obtaining a classroom plane graph, obtaining a classroom simulation planning model through a simulation planning model, and obtaining a classroom environment video frame of the classroom environment simulation model through a binocular camera. The video preprocessing module is used for obtaining a classroom environment enhanced image through a video preprocessing model according to the classroom environment video frame; the target detection module is used for obtaining a detection target through a target detection model according to the classroom environment enhancement image; the illumination adjustment module is used for collecting the real-time illumination intensity of the detection target through a photoresistor sensor, and obtaining a comfortable classroom environment through a self-adaptive illumination adjustment model according to the real-time illumination intensity. Classroom real-time light environment detection based on cloud edge collaboration is realized.

Description

Technical Field

[0001] This invention belongs to the field of light environment detection technology, specifically relating to a real-time classroom light environment detection system based on cloud-edge collaboration. Background Technology

[0002] With the development of educational informatization, classroom lighting environment management has gradually become an important aspect of improving teaching quality and learning experience. The classroom lighting environment has a significant impact on students' learning outcomes and teachers' teaching quality. However, traditional lighting environment management methods mainly rely on manual adjustment and fixed settings, lacking flexibility and intelligence, and failing to meet the needs of different teaching scenarios.

[0003] The introduction of big data technology has provided a new perspective and solutions for optimizing classroom lighting environments. In recent years, research on classroom lighting environment management and optimization based on big data has gradually emerged, achieving automatic adjustment and optimization of the classroom lighting environment through data acquisition technology, data analysis technology, and intelligent control strategies. However, big data-based classroom lighting environment management and optimization faces challenges such as complex model construction and insufficient control precision. Therefore, how to utilize deep learning and intelligent control methods to build accurate and efficient optimization models to achieve personalized, dynamic, and intelligent adjustment of the classroom lighting environment has become an urgent problem to be solved. Summary of the Invention

[0004] To address the aforementioned problems in the existing technology, this invention provides a real-time classroom lighting environment detection system based on cloud-edge collaboration. The objective of this invention can be achieved through the following technical solutions: A real-time classroom lighting environment detection system based on cloud-edge collaboration includes a model simulation module, a video preprocessing module, a target detection module, and a lighting adjustment module. The model simulation module is used to acquire the classroom floor plan and obtain the classroom simulation planning model through the simulation planning model, and to acquire the classroom environment video frames of the classroom environment simulation model through a binocular camera; The video preprocessing module is used to obtain an enhanced classroom environment image based on the classroom environment video frames through a video preprocessing model; The target detection module is used to obtain the target to be detected based on the enhanced classroom environment image through a target detection model; The illumination adjustment module is used to collect the real-time illumination intensity of the target through a photoresistor sensor, and obtain a comfortable classroom environment based on the real-time illumination intensity through an adaptive illumination adjustment model.

[0005] Preferably, obtaining the classroom floor plan and then using the simulation planning model to obtain the classroom simulation planning model includes: A two-dimensional model of the classroom is constructed based on the classroom floor plan. Obtain classroom space parameters and add them to the classroom 2D model to obtain the classroom simulation planning model.

[0006] Preferably, the classroom space parameters include interior surface material parameters and lighting parameters.

[0007] Preferably, obtaining the classroom environment enhanced image from the classroom environment video frames using a video preprocessing model includes: A smoothed image of the classroom environment is obtained by applying a mean filter template to the video frames of the classroom environment. The high-frequency components are obtained by removing the smoothed image of the classroom environment from the classroom environment image; The enhanced classroom environment image is obtained by fusing the high-frequency components and the smoothed classroom environment image. The fusion calculation expression is: , Among them, I enhanced I enhances the image of the classroom environment. smooth To enhance images of the classroom environment, I high For the high-frequency components, λ1, λ2, and γ are weighting parameters, ▽ 2 For a second-order Laplace operator, ▽ 4 Here, is the fourth-order Laplace operator, and tanh denotes the hyperbolic tangent function.

[0008] Preferably, obtaining the detected target from the enhanced classroom environment image using a target detection model includes: Based on the enhanced classroom environment image, candidate target point regions are obtained through a target region search model; Based on the candidate target point regions, a local target mapping map is obtained through a local discriminant model; Based on the location target mapping map, a background-suppressed target region is obtained by using a curvature filtering method to suppress the background. The background suppression enhancement target region is obtained by Hadamard operation based on the background suppression target region and the high-frequency components. The detected target is obtained by threshold segmentation based on the background suppression and enhancement target region.

[0009] Preferably, obtaining candidate target point regions based on the enhanced classroom environment image using a target region search model includes: The enhanced classroom environment image is extracted to obtain pixels, and the grayscale value of each pixel is calculated based on its grayscale value. The minimum distance is calculated based on the pixel and its grayscale value using the distance to neighboring high-density points. The joint feature factor of local density peaks is calculated by multiplication based on the gray value of the pixel and the minimum distance. The local density maximum point is obtained by sorting the local density peaks in descending order based on the joint characteristic factors. A target region to be tested is constructed with the local density maximum point as the center, and the gray level probability value is obtained by calculating the gray level probability of the target region to be tested. The candidate target point regions are obtained by sorting the grayscale probability values ​​in descending order.

[0010] Preferably, obtaining the localization target mapping map based on the candidate target point region using a local discriminant model includes: The mean gray value of the central sub-block, the gradient value of the central sub-block, the mean gray value of each layer sub-block, and the gradient value of each layer sub-block are obtained based on the candidate target point region. The difference in grayscale mean and the difference in gradient are calculated by subtraction based on the mean grayscale value of the central sub-block, the gradient value of the central sub-block, the mean grayscale value of each layer sub-block, and the gradient value of each layer sub-block. The gradient product of the diagonal sub-blocks is calculated by multiplication based on the gradient difference, and the gradient product of the diagonal sub-blocks carries a direction label; The average gradient of each sub-block is calculated by averaging the gradient values ​​of each sub-block; the gradient increment is calculated by subtracting the gradients of each sub-block; and the gradient increment is calculated by subtracting the average gradient of each sub-block. The location target mapping map is obtained by region positioning judgment based on the gradient product of the diagonal sub-blocks, the mean difference of gray levels, the gradient difference, the gradient increment, and the mean increment of gradients. The area location determination includes: If the gradient product of the diagonal sub-blocks in the first and second layers of the candidate target point region has four different directional labels, the region is marked as 0. If not, the region is marked as 0. If the mean difference of grayscale and the gradient difference are less than or equal to 0, the region is marked as 0. If not, the region is marked as 0. If the gradient increment is greater than 0, the localization target mapping map is obtained. If not, the region is marked as 0.

[0011] Preferably, obtaining a comfortable classroom environment based on the real-time light intensity using an adaptive light adjustment model includes: Obtain the minimum illuminance and the average illuminance, and calculate the illuminance uniformity based on the minimum illuminance and the average illuminance using uniformity calculation; The optimal classroom real-time light is obtained through the real-time light optimal target value model based on the illuminance uniformity and the real-time light intensity. The maximum supplementary lighting amount is obtained by dynamically calculating the supplementary lighting amount of the classroom real-time light based on the real-time light intensity, the maximum supplementary lighting amount, and the optimal classroom real-time light. The dynamic calculation expression for the supplemental light amount is: , Among them, D lsp The real-time supplemental lighting amount for the classroom, time lsp Let time be the real-time illumination intensity. sitlsp For the optimal real-time classroom lighting, Lsp max This refers to the maximum supplemental lighting amount.

[0012] Preferably, obtaining the optimal classroom real-time light based on the illuminance uniformity and the real-time illuminance intensity using a real-time light optimal target value model includes: The preset state, action, and reward are defined as follows: the state is the illuminance uniformity and the real-time light intensity; the action is the adjustment of light brightness. The reward expression is: , Among them, R t For the aforementioned reward, L optimal For the optimal real-time classroom lighting under the current conditions, L t Let λ be the real-time illumination intensity, and P(L) be the energy consumption weighting coefficient. lamp ) represents the power consumption of the lights, L lamp Indicates the current light brightness; Based on the state, the action, and the reward, the optimal classroom real-time light is obtained through the DQN model.

[0013] Preferably, the classroom real-time lighting environment detection system is based on a cloud-edge collaborative framework, specifically including the step of obtaining a reasonable classroom planning model based on the classroom floor plan through a simulation planning model, which is located on the cloud server; acquiring the classroom environment video, and obtaining the detection target by performing target detection through the classroom environment video through the edge server; the edge server collecting the real-time light intensity through the photoresistor sensor based on the detection target; the cloud server distributing the reasonable classroom planning model to the edge server; and the edge server performing adaptive lighting adjustment based on the real-time light intensity and the reasonable classroom planning model to obtain the optimal real-time classroom light.

[0014] The beneficial effects of this invention are as follows: (1) By combining cloud servers, edge servers and terminal devices, a cloud-edge collaborative intelligent light environment optimization framework is constructed. The cloud uses DIALux evo software to model the classroom light environment, the edge performs real-time video analysis and target detection, and the terminal devices collect light intensity and perform intelligent adjustment to achieve efficient and accurate light management.

[0015] (2) Candidate target points are extracted by improving the density peak search algorithm and refined by combining it with the local discrimination model to further optimize the target positioning accuracy. At the same time, Hadamard operation is used for background suppression to improve the robustness of target detection in complex environments. Real-time illumination intensity of the detected target points is collected to provide accurate data support for illumination adjustment.

[0016] (3) Using the DQN model, with illuminance uniformity and real-time light intensity as states and light brightness adjustment as actions, a reward function is constructed by combining energy consumption weight and optimal light target to achieve adaptive light adjustment, improve light comfort while optimizing energy consumption, and realize intelligent energy-saving management. Attached Figure Description

[0017] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart illustrating a real-time classroom lighting environment detection system based on cloud-edge collaboration according to the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects of the present invention in achieving the intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0020] Please see Figure 1 A real-time classroom lighting environment detection system based on cloud-edge collaboration includes a model simulation module, a video preprocessing module, a target detection module, and a lighting adjustment module. The model simulation module is used to acquire the classroom floor plan and obtain the classroom simulation planning model through the simulation planning model, and to acquire the classroom environment video frames of the classroom environment simulation model through a binocular camera; The video preprocessing module is used to obtain an enhanced classroom environment image based on the classroom environment video frames through a video preprocessing model; The target detection module is used to obtain the target to be detected based on the enhanced classroom environment image through a target detection model; The illumination adjustment module is used to collect the real-time illumination intensity of the target through a photoresistor sensor, and obtain a comfortable classroom environment based on the real-time illumination intensity through an adaptive illumination adjustment model.

[0021] Specifically, obtaining the classroom floor plan and then using the simulation planning model to obtain the classroom simulation planning model includes: S101: A two-dimensional model of the classroom is constructed based on the classroom floor plan using a two-dimensional model; S102: Obtain classroom space parameters and add the classroom space parameters to the classroom two-dimensional model to obtain the classroom simulation planning model.

[0022] Specifically, in step S101, the two-dimensional model is constructed using DIALux evo software; in step S103, the 3D rendering is achieved using Blender.

[0023] In this embodiment, the classroom space parameters include interior surface material parameters and lighting parameters. The floor of the small classroom is made of gray floor tiles with a reflectance of 0.2; the ceiling and walls are painted white with a reflectance of 0.7. In the lecture hall, the walls are painted white with a reflectance of 0.7; the ceiling is made of light-colored material with a reflectance of 0.7; the floor is made of light-colored floor tiles with a reflectance of 0.3. The lighting fixtures include T5 fluorescent lamps in the small classroom, arranged at equal intervals in a 3x4 pattern; and T8 grid fluorescent lamps in the lecture hall, evenly arranged in an embedded manner on the ceiling, except for two groups of two-tube grid lamps above the blackboard and two-tube grid fluorescent lamps per group on the ceiling.

[0024] Specifically, obtaining the enhanced classroom environment image from the classroom environment video frames using a video preprocessing model includes: S201: Obtain a classroom environment image by frame-segmentation processing based on the classroom environment video frames; S202: Obtain a smoothed classroom environment image based on the classroom environment image using a mean filter template; The filtering operator of the mean filtering template is: , in, The filtering operator; S203: Remove the high-frequency component from the smoothed classroom environment image; The high-frequency component is represented as: , Wherein, H represents the high-frequency component, and I represents the classroom environment image. This represents a smoothed image of the classroom environment, where max represents the function that takes the maximum value. The high-frequency component is obtained by taking the maximum value between the smoothed image of the classroom environment and 0. S204: The enhanced classroom environment image is obtained by fusing the high-frequency components and the smoothed classroom environment image; The fusion calculation expression is: , Among them, I enhanced I enhances the image of the classroom environment. smooth To enhance images of the classroom environment, I high For the high-frequency components, λ1, λ2, and γ are weighting parameters, ▽ 2 For a second-order Laplace operator, ▽ 4 Here, is the fourth-order Laplace operator, and tanh denotes the hyperbolic tangent function.

[0025] In this embodiment, a 13×13 mean filter template is used to smooth the image. The middle of the filter operator is a 5×5 zero matrix, the outer side is filled with 2 rows and 2 columns with a fill value of -2, and the outermost side is filled with 2 rows and 2 columns with a fill value of -1.

[0026] Specifically, obtaining the detected target from the enhanced classroom environment image using the target detection model includes: S301: Based on the enhanced classroom environment image, candidate target point regions are obtained through a target region search model; S302: Obtain the target mapping map based on the candidate target point region using a local discriminant model; S303: Based on the positioning target mapping map, a background-suppressed target region is obtained by performing background suppression using a curvature filtering method; S304: Obtain the background suppression enhancement target region by Hadamard operation based on the background suppression target region and the high-frequency component; S305: The detected target is obtained by threshold segmentation based on the background suppression and enhancement target region; The threshold segmentation expression is: , Where Thr is the target being detected, μ is the background mean, θ is a constant, and σ is the background standard deviation.

[0027] Specifically, obtaining candidate target point regions based on the enhanced classroom environment image using a target region search model includes: S301-1: Extract the classroom environment enhancement image to obtain pixels, and calculate the gray value of the pixels based on the gray value; The expression for calculating the grayscale value is: , Where, ρ mLet I(x,y) be the grayscale value of the pixel, and let I(x,y) represent the pixel. S301-2: The minimum distance is calculated based on the pixel and its grayscale value using the distance to neighboring high-density points; The expression for calculating the distance between neighboring high-density points is: , Where δ(i,j) represents the minimum distance, min represents the minimum value function, ρ1 and ρ2 are the gray values ​​of the pixels, and d (i,j) x represents the distance to neighboring high-density points. i Let x represent the x-coordinate of the i-th pixel. j The x-coordinate of the j-th pixel is represented by y. i The y-coordinate represents the ordinate of the i-th pixel. j This represents the ordinate of the j-th pixel; S301-3: The joint feature factor of local density peaks is calculated by multiplication based on the gray value of the pixel and the minimum distance; The multiplication calculation expression is as follows: , Wherein, ω represents the joint feature factor of the local density peak, ρ represents the gray value of the pixel, and δ represents the minimum distance; S301-4: The local density maximum point is obtained by sorting the local density peaks in descending order according to the joint characteristic factors of the local density peaks; S301-5: Construct the target region to be tested with the local density maximum point as the center, and obtain the gray level probability value by calculating the gray level probability of the target region to be tested; The expression for calculating the grayscale probability is: , Among them, R k Let s(i,j) be the grayscale probability value, M represent the number of rows of the target region to be tested, N represent the number of columns of the target region to be tested, i represent the target region in the i-th row, j represent the target region in the j-th column, and s(i,j) be the grayscale value. k Let u be the pixel value of the target to be measured, and logP be the pixel value of the target to be measured. s(i,j) This represents the proportion of pixels with grayscale value s(i,j) in the region; S301-6: The candidate target point regions are obtained by sorting the grayscale probability values ​​in descending order.

[0028] In this embodiment, the local density peak joint feature factors are sorted in descending order, and the top 20 pixels are selected as the local density maximum points. A target region to be tested is established centered on these local density maximum points. The local grayscale probability values ​​of the target region are calculated, and these local grayscale probability values ​​are sorted in descending order. A larger local grayscale probability value indicates a more uniform grayscale distribution within the region, while a smaller local grayscale probability value indicates a more uneven grayscale distribution within the region. The top 12 pixels with the highest local grayscale probability values ​​are selected as candidate target points.

[0029] Specifically, the local discrimination model includes: S302-1: Obtain the mean gray value of the central sub-block, the gradient value of the central sub-block, the mean gray value of each layer sub-block, and the gradient value of each layer sub-block based on the candidate target point region; S302-2: The difference in grayscale mean and the difference in gradient are calculated by subtraction based on the mean grayscale value of the central sub-block, the gradient value of the central sub-block, the mean grayscale value of each layer sub-block, and the gradient value of each layer sub-block; The expression for the grayscale mean difference is: , Among them, MD Bi M represents the difference in grayscale mean. T M represents the average grayscale value of the central sub-block. Bi This represents the average grayscale value of each sub-block in the layer; The gradient difference expression is: , Among them, GD Bi G represents the gradient difference. T G represents the gradient value of the central sub-block. Bi This represents the gradient value of each sub-block in the layer; S302-3: The gradient product of the diagonal sub-blocks is calculated by multiplication based on the gradient difference, and the gradient product of the diagonal sub-blocks carries a direction label; The expression for the gradient product of the diagonal sub-blocks is: , Among them, FG Bi GD represents the gradient product of the diagonal sub-blocks. Bi GD represents the gradient difference of the i-th sub-block in layer B. B(b-i) Let represent the gradient difference of the bi-th sub-block in layer B, where b represents the number of sub-blocks in layer B. S302-4: Calculate the average gradient of each sub-block based on the average gradient value of each sub-block; calculate the gradient increment by subtracting the gradients of each sub-block; calculate the gradient increment by subtracting the average gradient of each sub-block. S302-5: The positioning target mapping map is obtained by region positioning judgment based on the gradient product of the diagonal sub-blocks, the gray-scale mean difference, the gradient difference, the gradient increment, and the gradient mean increment; The area location determination includes: If the gradient product of the diagonal sub-blocks in the first and second layers of the candidate target point region has four different directional labels, the region is marked as 0. If not, the region is marked as 0. If the mean difference of grayscale and the gradient difference are less than or equal to 0, the region is marked as 0. If not, the region is marked as 0. If the gradient increment is greater than 0, the localization target mapping map is obtained. If not, the region is marked as 0.

[0030] Specifically, obtaining a comfortable classroom environment based on the real-time light intensity using an adaptive lighting adjustment model includes: S401: Obtain the minimum illuminance and the average illuminance, and calculate the illuminance uniformity based on the minimum illuminance and the average illuminance using uniformity calculation; The uniformity calculation expression is as follows: , Where U is the illuminance uniformity, and E min E is the minimum illuminance. av The average illuminance; S402: Based on the illuminance uniformity and the real-time illuminance intensity, the optimal classroom real-time light is obtained through the real-time light optimal target value model; S403: Obtain the maximum supplementary light amount. Based on the real-time light intensity, the maximum supplementary light amount, and the optimal real-time classroom light, dynamically calculate the real-time classroom light supplementary light amount. The dynamic calculation expression for the supplemental light amount is: , Among them, D lsp The real-time supplemental lighting amount for the classroom, time lsp Let time be the real-time illumination intensity. sitlsp For the optimal real-time classroom lighting, Lsp max This refers to the maximum supplemental lighting amount.

[0031] Specifically, the real-time optical optimal target value model includes: S402-1: Preset state, action, reward, wherein the state is the illuminance uniformity and the real-time light intensity, and the action is the light brightness adjustment; The reward expression is: , Among them, R t For the aforementioned reward, Loptimal For the optimal real-time classroom lighting under the current conditions, L t Let λ be the real-time illumination intensity, and P(L) be the energy consumption weighting coefficient. lamp ) represents the power consumption of the lights, L lamp Indicates the current light brightness; S402-1: Based on the state, the action, and the reward, the optimal classroom real-time light is obtained through the DQN model.

[0032] Specifically, the classroom real-time lighting environment detection system is based on a cloud-edge collaborative framework. Specifically, the steps of obtaining a rationally planned classroom model based on the classroom floor plan through a simulation planning model are located on the cloud server; acquiring the classroom environment video; performing target detection based on the classroom environment video through an edge server; the edge server collecting real-time light intensity based on the detected targets using a photoresistor sensor; the cloud server distributing the rationally planned classroom model to the edge server; and the edge server performing adaptive lighting adjustment based on the real-time light intensity and the rationally planned classroom model.

[0033] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0034] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0035] The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. The computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0036] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A real-time classroom lighting environment detection system based on cloud-edge collaboration, characterized in that, It includes a model simulation module, a video preprocessing module, a target detection module, and an illumination adjustment module; The model simulation module is used to acquire the classroom floor plan and obtain the classroom simulation planning model through the simulation planning model, and to acquire the classroom environment video frames of the classroom environment simulation model through a binocular camera; The video preprocessing module is used to obtain an enhanced classroom environment image based on the classroom environment video frames through a video preprocessing model; The target detection module is used to obtain the target to be detected based on the enhanced classroom environment image through a target detection model; The illumination adjustment module is used to collect the real-time illumination intensity of the target through a photoresistor sensor, and obtain a comfortable classroom environment based on the real-time illumination intensity through an adaptive illumination adjustment model.

2. The classroom real-time light environment detection system based on cloud-edge collaboration according to claim 1, characterized in that, The process of obtaining the classroom floor plan and then using the simulation planning model to create the classroom simulation planning model includes: A two-dimensional model of the classroom is constructed based on the classroom floor plan. Obtain classroom space parameters and add them to the classroom 2D model to obtain the classroom simulation planning model.

3. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 2, characterized in that, The classroom space parameters include interior surface material parameters and lighting parameters.

4. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 1, characterized in that, The step of obtaining the enhanced classroom environment image based on the classroom environment video frames through a video preprocessing model includes: A smoothed image of the classroom environment is obtained by applying a mean filter template to the video frames of the classroom environment. The high-frequency components are obtained by removing the smoothed image of the classroom environment from the classroom environment image; The enhanced classroom environment image is obtained by fusing the high-frequency components and the smoothed classroom environment image. The fusion calculation expression is: ， Among them, I enhanced I enhances the image of the classroom environment. smooth To enhance images of the classroom environment, I high For the high-frequency components, λ1, λ2, and γ are weighting parameters, ▽ 2 For a second-order Laplace operator, ▽ 4 Here, is the fourth-order Laplace operator, and tanh denotes the hyperbolic tangent function.

5. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 4, characterized in that, The targets obtained from the enhanced classroom environment image through the target detection model include: Based on the enhanced classroom environment image, candidate target point regions are obtained through a target region search model; Based on the candidate target point regions, a local target mapping map is obtained through a local discriminant model; Based on the location target mapping map, a background-suppressed target region is obtained by using a curvature filtering method to suppress the background. The background suppression enhancement target region is obtained by Hadamard operation based on the background suppression target region and the high-frequency components. The detected target is obtained by threshold segmentation based on the background suppression and enhancement target region.

6. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 5, characterized in that, The step of obtaining candidate target point regions based on the enhanced classroom environment image using a target region search model includes: The enhanced classroom environment image is extracted to obtain pixels, and the grayscale value of each pixel is calculated based on its grayscale value. The minimum distance is calculated based on the pixel and its grayscale value using the distance to neighboring high-density points. The joint feature factor of local density peaks is calculated by multiplication based on the gray value of the pixel and the minimum distance. The local density maximum point is obtained by sorting the local density peaks in descending order based on the joint characteristic factors. A target region to be tested is constructed with the local density maximum point as the center, and the gray level probability value is obtained by calculating the gray level probability of the target region to be tested. The candidate target point regions are obtained by sorting the grayscale probability values ​​in descending order.

7. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 5, characterized in that, The step of obtaining the local target mapping map based on the candidate target point region through a local discriminant model includes: The mean gray value of the central sub-block, the gradient value of the central sub-block, the mean gray value of each layer sub-block, and the gradient value of each layer sub-block are obtained based on the candidate target point region. The difference in grayscale mean and the difference in gradient are calculated by subtraction based on the mean grayscale value of the central sub-block, the gradient value of the central sub-block, the mean grayscale value of each layer sub-block, and the gradient value of each layer sub-block. The gradient product of the diagonal sub-blocks is calculated by multiplication based on the gradient difference, and the gradient product of the diagonal sub-blocks carries a direction label; The average gradient of each sub-block is calculated by averaging the gradient values ​​of each sub-block; the gradient increment is calculated by subtracting the gradients of each sub-block; and the gradient increment is calculated by subtracting the average gradient of each sub-block. The location target mapping map is obtained by region positioning judgment based on the gradient product of the diagonal sub-blocks, the mean difference of gray levels, the gradient difference, the gradient increment, and the mean increment of gradients. The area location determination includes: If the gradient product of the diagonal sub-blocks in the first and second layers of the candidate target point region has four different directional labels, the region is marked as 0. If not, the region is marked as 0. If the mean difference of grayscale and the gradient difference are less than or equal to 0, the region is marked as 0. If not, the region is marked as 0. If the gradient increment is greater than 0, the localization target mapping map is obtained. If not, the region is marked as 0.

8. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 1, characterized in that, The process of obtaining a comfortable classroom environment based on the real-time light intensity using an adaptive light adjustment model includes: Obtain the minimum illuminance and the average illuminance, and calculate the illuminance uniformity based on the minimum illuminance and the average illuminance using uniformity calculation; The optimal classroom real-time light is obtained through the real-time light optimal target value model based on the illuminance uniformity and the real-time light intensity. The maximum supplementary lighting amount is obtained by dynamically calculating the supplementary lighting amount of the classroom real-time light based on the real-time light intensity, the maximum supplementary lighting amount, and the optimal classroom real-time light. The dynamic calculation expression for the supplemental light amount is: ， Among them, D lsp The real-time supplemental lighting amount for the classroom, time lsp Let time be the real-time illumination intensity. sitlsp For the optimal real-time classroom lighting, Lsp max This refers to the maximum supplemental lighting amount.

9. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 8, characterized in that, The process of obtaining the optimal classroom real-time light based on the illuminance uniformity and the real-time light intensity using the real-time light optimal target value model includes: The preset state, action, and reward are defined as follows: the state is the illuminance uniformity and the real-time light intensity; the action is the adjustment of light brightness. The reward expression is: ， Among them, R t For the aforementioned reward, L optimal For the optimal real-time classroom lighting under the current conditions, L t Let λ be the real-time illumination intensity, and P(L) be the energy consumption weighting coefficient. lamp ) represents the power consumption of the lights, L lamp Indicates the current light brightness; Based on the state, the action, and the reward, the optimal classroom real-time light is obtained through the DQN model.

10. The real-time classroom light environment detection system based on cloud-edge collaboration according to claim 1, characterized in that, The real-time classroom lighting environment detection system is based on a cloud-edge collaborative framework. Specifically, the steps of obtaining a reasonable classroom model through a simulation planning model based on the classroom floor plan are located on the cloud server; acquiring the classroom environment video; obtaining the detection target through target detection via an edge server based on the classroom environment video; and collecting the real-time light intensity through a photoresistor sensor based on the detection target on the edge server. The cloud server distributes the rationally planned classroom model to the edge server; The edge server performs adaptive lighting adjustment based on the real-time light intensity and the rationally planned classroom model to obtain the optimal real-time classroom light.

Images