A landscape lighting control method and system based on crowd gathering conditions

Abstract

The present invention relates to the field of intelligent lighting control technology, specifically a landscape lighting control method and system based on crowd gathering conditions. The present invention includes: first, the present invention proposes a landscape lighting analysis model, which accurately grasps the lighting needs of different areas by analyzing the distribution area of ​​the crowd and the density of the crowd; in addition, the road information and vegetation information are also considered as factors affecting the lighting effect of the landscape lights, and the control parameters of the landscape lights are optimized; secondly, the present invention predicts the movement state of the crowd and realizes the secondary adjustment of the control parameters; the method establishes a transfer matrix by analyzing the changing characteristics of the crowd, and predicts the movement state in combination with the adjusted lighting information, which is helpful for intelligent control decision-making; finally, a comprehensive calculation is performed based on the lighting quality, energy consumption and user satisfaction to accurately evaluate the effectiveness of the landscape lighting control method based on the crowd gathering conditions proposed by the present invention.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent lighting control, and in particular to a landscape lighting control method and system based on crowd gathering conditions. Background Art

[0002] Landscape lights are a type of lighting fixtures specifically used to illuminate and beautify outdoor environments. Their design and installation are intended to enhance visual effects and safety at night or in dim conditions. Landscape lights have many characteristics, including: aesthetics, which can enhance the visual appeal and decorative effects of the environment through different light colors, brightness and projection angles; enhancing the environment, which can emphasize or highlight specific landscape elements through the projection and design of lights; energy saving and environmental protection. Modern landscape lights mostly use energy-saving light sources, such as LED lights, which have a higher energy efficiency ratio and a longer service life, helping to reduce energy consumption and environmental impact.

[0003] Landscape lights are used in many occasions because of their many good properties, such as urban public spaces, buildings, parks, and tourist attractions. Most of the landscape light application scenarios mentioned above are in places with dense traffic, and the control of landscape lights is a very difficult problem because the movement and changes of traffic are usually dynamic and unpredictable. In high-traffic places, the density and distribution of the crowd will change with time and activity, which makes it more complicated and challenging to predict and adjust the control parameters of landscape lights. If the traditional landscape light control method is used, the brightness of the landscape light is set to a fixed value or adjusted in a fixed mode within a predetermined time period; this method has limitations and is difficult to cope with dynamic changes in traffic and diverse usage needs. In contrast, the intelligent control method based on real-time data and traffic prediction can adjust the control parameters of landscape lights more flexibly and accurately to improve energy efficiency, save energy, and optimize user experience.

[0004] In the prior art, there are some excellent methods for controlling landscape lights in crowded conditions in an intelligent manner; for example, a Chinese patent with prior art application number CN202110537008.1 discloses a landscape lighting design method and system based on crowd gathering; the method includes: obtaining lighting partition information according to the lighting design range and lighting design scheme information; obtaining first design element information according to the lighting design scheme information; inputting the lighting partition information and the first design element information into the first element analysis model to obtain the partition element ratio; obtaining partition range weight information according to the lighting position information and the lighting design range; obtaining the first partition weight according to the lighting partition information and the partition range weight information; obtaining historical pedestrian flow data according to the lighting position information; obtaining a pedestrian flow trend chart and a pedestrian flow setting threshold according to the historical pedestrian flow data; obtaining a first partition threshold according to the partition element ratio, the first partition weight, and the pedestrian flow setting threshold; obtaining a first partition control instruction according to the first partition threshold; however, the above method only uses the pedestrian flow information in the area as the control factor, but ignores the influence of regional environmental factors on the lighting effect of the landscape lamp, which will result in inaccurate adjustment of the control parameters of the landscape lamp and reduce the user experience.

[0005] To this end, the present invention proposes a landscape lighting control method and system based on crowd gathering conditions. Summary of the invention

[0006] The purpose of the present invention is to propose a landscape lighting control method and system based on crowd gathering, which can realize effective control of landscape lights under crowd gathering conditions; the method and system mainly include: first, using a camera to obtain video within the illumination distance of the landscape light, and preprocessing the video to obtain images within a continuous time period; secondly, performing landscape lighting analysis on the image; the analysis process adopts a landscape lighting analysis model, which includes: an input layer, a recognition layer, a marking layer, an analysis layer, a lighting parameter calculation layer and an output layer; wherein the analysis layer is used to analyze the crowd, road surface and vegetation appearing in the video; calculate the distribution position and density of the crowd; identify the material, flatness and presence of water accumulation of the road surface; calculate the type, distribution and plant density of vegetation; thereby accurately grasp the impact of landscape lights on lighting; the lighting parameter calculation layer can calculate the control parameters of the landscape lights that meet the current scene based on the analysis results of the analysis layer; finally, in order to realize more intelligent control of landscape lights The present invention proposes an optimization method for secondary adjustment of landscape light control parameters, which timely adjusts the control parameters of landscape lights by predicting the movement state of people flow in the future; the method obtains the transfer probability by calculating the movement state of people flow in the video, and makes the next prediction of people flow in combination with the lighting conditions after adjusting the landscape light control parameters; the method can realize the advance control of landscape lights, making it more intelligent; in the whole landscape light control process, the present invention proposes an evaluation method for evaluating the effect of landscape light control parameters after adjustment; the method comprehensively considers the lighting quality, energy consumption and user satisfaction; the lighting quality mainly includes the brightness and uniformity of light in each lighting area; the energy consumption is mainly compared with the comparison between before adjustment and after adjustment; the user satisfaction is obtained through the form of questionnaire and the feedback result is obtained through quantitative data. This comprehensive evaluation method is helpful to evaluate the comprehensiveness of the landscape light control effect of the present invention, which is beneficial for relevant personnel to formulate a better plan according to the evaluation results.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A landscape lighting control method based on crowd gathering conditions, comprising:

[0009] Initially setting the landscape light according to current time information; the current time information includes: current season and current time; wherein the initial setting sets the landscape light according to preset information, and only includes the brightness and lighting range of the landscape light;

[0010] Furthermore, the scene video is acquired through a camera; wherein the video range acquired by the camera is based on the landscape lamp as the center point and the maximum lighting distance as the radius;

[0011] Furthermore, the scene video is preprocessed to obtain a first image set within a continuous time period; wherein the preprocessing includes: extracting video frames within a continuous time period of the scene video to obtain a video frame set; denoising images in the video frame set to obtain a denoised image set; enhancing images in the denoised image set to obtain an enhanced image set; color correcting images in the enhanced image set to obtain a corrected image set; unifying the image sizes in the corrected image set to obtain the first image set.

[0012] Furthermore, a landscape lighting analysis model is used for analysis; wherein the lighting analysis model comprises: an input layer, a recognition layer, a marking layer, an analysis layer, a lighting parameter calculation layer and an output layer;

[0013] The images in the first image set are input from the input layer to the recognition layer to obtain a first recognition result, a second recognition result and a third recognition result; wherein the first recognition result represents a crowd; the third recognition result represents a road surface; and the third recognition result represents a plant;

[0014] Furthermore, the marking layer marks the first recognition result, the second recognition result and the third recognition result to obtain a set of marked images;

[0015] Furthermore, the set of marked images is input into the analysis layer to obtain crowd analysis results, road surface analysis results and vegetation analysis results; wherein the analysis layer includes: a crowd analysis module, a road surface analysis module and a vegetation analysis module:

[0016] The specific analysis process of the crowd analysis module includes: obtaining crowd information in the marked image set; the crowd information is expressed as: CI = {(D i ,N i ,L i ,Av i )|1≤i≤M}; where M represents the total number of people; D i It is represented as the straight-line distance from crowd i to the landscape light; L i Represented as the spatial distribution position of population i; N i It is represented by the total number of people in group i; Av i It is expressed as the average volume of each person in the population i; the average volume is expressed as: H n W is the height of the nth person; n It is expressed as the width of the nth person;

[0017] Calculate the crowd density in the first image set; wherein the calculation formula of the crowd density is: cd iIt is expressed as the crowd density of crowd i; Where A represents the monitoring area;

[0018] Obtain the population analysis result; the population analysis result is expressed as: PAR = {Max (D i ),L i ,CD|1≤i≤M};

[0019] The specific analysis process of the road surface analysis module includes: extracting road surface features of the images in the marked image set, including: texture features, color features and local features; classifying according to the road surface features to obtain road surface types; segmenting the images in the marked image set; extracting features for each segmented area to obtain concave-convex features, including: grayscale gradient, contour and color; quantifying the concave-convex degree of each segmented area according to the concave-convex features; using an object detection algorithm for each segmented area to identify the shape and boundary of water accumulation on the road surface, and judging whether there is water accumulation on the road surface; obtaining the position and shape of water accumulation when water accumulation occurs;

[0020] The specific process of the vegetation analysis module includes: using a neural network to identify plants based on the vegetation tags in the tag image set to obtain plant categories; and obtaining the distribution position, growth status, height distribution and leaf coverage of the corresponding plants based on the plant categories.

[0021] Furthermore, the crowd analysis results, road surface analysis results and vegetation analysis results are input into the lighting parameter calculation layer to obtain the control parameters of the landscape lights; wherein the lighting parameter calculation layer is used to calculate the lighting distance and lighting intensity of the landscape lights;

[0022] Wherein, the lighting distance is compensated for errors according to the initial lighting distance of the landscape lamp and the farthest crowd distance in the crowd analysis result;

[0023] The calculation formula of the lighting intensity is: Wherein, I represents the initial lighting intensity of the landscape lamp; LP i Expressed as the location factor of population i; CR k Expressed as the leaf coverage of vegetation k; RL b Indicates the concave and convex position of the road surface b; UD b It is expressed as the degree of roughness of the road surface b; PO b It is expressed as the degree of water accumulation on road surface b; ω 1i Represented as the weight parameter of population i; ω 2k Represented as the weight parameter of vegetation k; ω 3b The weight parameter representing the concave and convex position of the road surface b; ω 4b The weight parameter of the unevenness of the road surface b; ω5b It is expressed as a weight parameter of the degree of water accumulation on road surface b.

[0024] Furthermore, the regulation parameter is outputted via the output layer as a first landscape light control parameter;

[0025] Further, predicting the movement state of the crowd, and analyzing the landscape lighting analysis model according to the prediction results;

[0026] wherein, according to the analysis layer, a plurality of analysis results of the crowd in a continuous time period are obtained by analyzing the crowd in the first image set;

[0027] Extracting crowd change characteristics of the plurality of crowd analysis results; wherein the crowd change characteristics include: time characteristics, crowd location characteristics, crowd density characteristics, crowd movement speed characteristics and crowd flow direction characteristics;

[0028] Establishing a crowd flow transfer matrix according to the crowd change characteristics;

[0029] The one-step transition probability is calculated according to the crowd flow transfer matrix; wherein the calculation formula of the one-step transition probability is:

[0030]

[0031] Among them, P A→B (t) represents the probability of one-step transition from point A to point B at time t; cd t,A It is expressed as the flow density of people at point A at time t; cd t,B is the flow density of people at point B at time t; cd t,C is the crowd density at time t; v A→B Expressed as the speed of movement from point A to point B;

[0032] Obtaining the lighting information of each crowd area after the landscape lights are adjusted; wherein the lighting information includes: lighting intensity and lighting distribution;

[0033] The one-step transition probability is combined with the lighting information to predict the movement state of the crowd; wherein the function of predicting the movement state of the crowd is:

[0034]

[0035] Among them, ΔS i (t+1) is the change in the crowd density at point i at time t+1; P i→j It is expressed as the probability of one-step transition from point i to point j at time t; L j (t) represents the lighting information at time t and point j; α represents the parameter for adjusting the lighting to affect the movement of people; d ijIt is expressed as the distance between point i and point j.

[0036] Further, the first landscape light control parameter is adjusted to obtain a second landscape light control parameter;

[0037] Furthermore, the second landscape lamp control parameter is evaluated to obtain a comprehensive evaluation value of the lighting effect of the landscape lamp; wherein the lighting evaluation of the adjusted landscape lamp includes:

[0038] Determine the lighting evaluation index of the landscape lamp; wherein the lighting evaluation index includes: lighting quality index, energy consumption index and user satisfaction;

[0039] Furthermore, the lighting quality indicators include: average illumination, clarity and illumination uniformity; the energy consumption indicators include: energy saving rate and energy consumption rate; the user satisfaction can be conducted by questionnaire survey;

[0040] Further, collecting various indicator data of the landscape lights after adjustment within the evaluation period;

[0041] Further, the evaluation values ​​of each indicator are calculated;

[0042] Furthermore, according to the evaluation values ​​of each of the indicators, the comprehensive evaluation value of the lighting effect of the landscape lamp is calculated; the calculation formula of the comprehensive evaluation value of the lighting effect is: CE = c 1 *LQAV+c 2 *ECAV+c 3 *USEV; LQAV stands for lighting quality assessment value; ECAV stands for energy consumption assessment value; USEV stands for user satisfaction assessment value; c 1 Expressed as the weight coefficient for lighting quality assessment; c 2 Expressed as the weight coefficient of energy consumption assessment; c 3 Expressed as the weight coefficient of user satisfaction.

[0043] A landscape lighting control system based on crowd gathering conditions, the system comprising: a monitoring unit, a collection unit, a data processing unit, a control parameter adjustment unit, a control parameter adaptive adjustment unit, an evaluation unit, a music unit and a remote management unit;

[0044] Wherein, the monitoring unit is used to monitor the scene within the lighting distance of the landscape lamp;

[0045] The acquisition unit is used to acquire video within the monitoring range of the monitoring unit;

[0046] A data processing unit, used for preprocessing video data; wherein the data processing unit comprises:

[0047] Extract video frames within a continuous time period of a scene video to obtain a video frame set; denoise the images in the video frame set to obtain a denoised image set; enhance the images in the denoised image set to obtain an enhanced image set; perform color correction on the images in the enhanced image set to obtain a corrected image set; unify the image sizes in the corrected image set to obtain a first image set.

[0048] The control parameter adjustment unit is used to adjust the lighting parameters of the landscape lamp; wherein the control parameter adjustment unit adopts a landscape lamp lighting analysis model; wherein the landscape lamp lighting analysis model includes: an input layer, a recognition layer, a marking layer, an analysis layer, a lighting parameter calculation layer and an output layer;

[0049] The analysis layer includes: crowd analysis module, road surface analysis module and vegetation analysis module:

[0050] The specific analysis process of the crowd analysis module includes: obtaining crowd information in the marked image set; the crowd information is expressed as: CI = {(D i ,N i ,L i ,Av i )|1≤i≤M}; where M represents the total number of people; D i It is represented as the straight-line distance from crowd i to the landscape light; L i Represented as the spatial distribution position of population i; N i It is represented by the total number of people in group i; Av i It is expressed as the average volume of each person in the population i; the average volume is expressed as: H n W is the height of the nth person; n It is expressed as the width of the nth person;

[0051] Calculate the crowd density in the first image set; wherein the calculation formula of the crowd density is: cd i It is expressed as the crowd density of crowd i; Where A represents the monitoring area;

[0052] Obtain the population analysis result; the population analysis result is expressed as: PAR = {Max (D i ),L i ,CD|1≤i≤M};

[0053] The specific analysis process of the road surface analysis module includes: extracting road surface features of the images in the marked image set, including texture features, color features and local features; classifying according to the road surface features to obtain road surface types;

[0054] Performing road area segmentation on the images in the marked image set; performing feature extraction on each segmented area to obtain concave and convex features, including grayscale gradient, contour and color;

[0055] Quantifying the degree of concavity and convexity of each segmented area according to the concavity and convexity features; using an object detection algorithm for each segmented area to identify the shape and boundary of water accumulation on the road surface, and determine whether there is water accumulation on the road surface; when water accumulation occurs, obtaining the position and shape of the water accumulation;

[0056] The specific process of the vegetation analysis module includes: using a neural network to identify plants based on the vegetation tags in the tag image set to obtain plant categories; and obtaining the distribution position, growth status, height distribution and leaf coverage of the corresponding plants based on the plant categories.

[0057] Wherein, the lighting parameter calculation layer is used to calculate the lighting distance and lighting intensity of the landscape lamp;

[0058] Wherein, the lighting distance is compensated for errors according to the initial lighting distance of the landscape lamp and the farthest crowd distance in the crowd analysis result;

[0059] The calculation formula of the lighting intensity is: Wherein, I represents the initial lighting intensity of the landscape lamp; LP i Expressed as the location factor of population i; CR k Expressed as the leaf coverage of vegetation k; RL b Indicates the concave and convex position of the road surface b; UD b It is expressed as the degree of roughness of the road surface b; PO b It is expressed as the degree of water accumulation on road surface b; ω 1i Represented as the weight parameter of population i; ω 2k Represented as the weight parameter of vegetation k; ω 3b The weight parameter representing the concave and convex position of the road surface b; ω 4b The weight parameter of the unevenness of the road surface b; ω 5b It is expressed as a weight parameter of the degree of water accumulation on road surface b.

[0060] The control parameter adaptive adjustment unit is used to perform secondary optimization on the lighting parameters of the landscape lamp;

[0061] The evaluation unit is used to evaluate the effect of the adjusted lighting parameters of the landscape lamp;

[0062] The music unit is used for the landscape lamp to play music;

[0063] The remote management unit is used to remotely manage the landscape lights.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] 1. The present invention adopts a landscape lighting analysis model, which includes: an input layer, an identification layer, a marking layer, an analysis layer, a lighting parameter calculation layer and an output layer; wherein the analysis layer is used to analyze the crowd, road surface and vegetation appearing in the video; calculate the distribution position and density of the crowd; identify the material, flatness and presence of water accumulation of the road surface; calculate the type, distribution and plant density of vegetation; thereby accurately grasping the impact of landscape lighting; the lighting parameter calculation layer can calculate the control parameters of the landscape lights in the current scene according to the analysis results of the analysis layer, thereby improving the user experience.

[0066] 2. In order to realize a more intelligent control method for landscape lights, the present invention proposes an optimization method for secondary adjustment of landscape light control parameters. The method timely adjusts the control parameters of landscape lights by predicting the movement state of people flow in the future; the method obtains the transfer probability by calculating the movement state of people flow in the video, and makes the next prediction of people flow in combination with the lighting conditions after adjusting the landscape light control parameters; the method can realize the advance control of landscape lights, making it more intelligent and enhancing the user experience.

[0067] 3. The present invention proposes an evaluation method for evaluating the effect of landscape lamp control parameters after adjustment; the method takes into account lighting quality, energy consumption and user satisfaction; lighting quality mainly includes the brightness of each lighting area and the uniformity of light; energy consumption is mainly compared between before and after adjustment; user satisfaction is obtained through questionnaires and quantitative data to obtain feedback results. This comprehensive evaluation method helps to evaluate the comprehensiveness of the landscape lamp control effect of the present invention and helps to improve the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A flow chart of a landscape lighting control method based on crowd gathering conditions provided by an embodiment of the present invention;

[0069] Figure 2 A structural diagram of a landscape lighting control system based on crowd gathering conditions provided by an embodiment of the present invention;

[0070] Figure 3 A schematic diagram of a scenario provided by an embodiment of the present invention.

[0071] In the figure: 1. surveillance camera; 2. lampshade; 3. lamp body; 4. lamp post. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] As a kind of lamp specially used for lighting and beautifying outdoor environment, landscape lamp is designed to enhance visual effect and safety at night or in dim conditions. There are many characteristics in the use of landscape lamp, such as beautifying the environment through different light colors; enhancing the visual appeal and decorative effect of the environment by changing the brightness and distribution of light; modern landscape lamps mostly use energy-saving lamp sources, which have high energy efficiency and long service life, and help reduce energy consumption and environmental impact.

[0074] Based on the above many good characteristics, landscape lights are used in many occasions, such as tourist attractions, communities and parks. In the traditional method of controlling landscape lights, the lighting time and brightness of landscape lights are basically set in advance; for traditional control methods, most technologies now tend to develop in the direction of intelligence. For example, the method of the prior art analyzes the density of pedestrian flow in the area and adaptively adjusts the control parameters of the landscape lights; however, most methods ignore the impact of environmental factors on the lighting effect of landscape lights, so it is impossible to accurately regulate the control parameters of landscape lights; for this reason, the present invention proposes a landscape lighting control method and system based on the gathering of pedestrian flow; the method and system can perform adaptive control of landscape lights according to the changes in pedestrian flow; in order to illustrate the feasibility of the present invention, the following will be described in detail from two embodiments.

[0075] Embodiment 1:

[0076] In the implementation of this application, the implementation scene is selected in the park. The park is one of the must-have places for residents to carry out night activities and is also a place with dense traffic. Therefore, it is necessary to design a good landscape light control solution. In the implementation process, this embodiment combines the proposed method with the system to achieve a good lighting effect in the park scene. First, in Figure 1 The specific steps of the method proposed in the present invention are given in the following, including: S10. Obtaining current time information; S20. Obtaining scene video; S30. Preprocessing the video; S40. Performing lighting analysis; S50. Predicting the movement trend of people flow and optimizing the control parameters of landscape lights; S60 evaluating the lighting effect of landscape lights. Figure 2 The system structure proposed by the present invention is given in the text, which includes: a monitoring unit, a collection unit, a data processing unit, a control parameter adjustment unit, a control parameter adaptive adjustment unit, an evaluation unit, a music unit and a remote management unit; Figure 1 The method steps can be Figure 2 The units in the above code implement the corresponding functions. The following will explain the process of combining the two:

[0077] Further, by Figure 1 Step S10 acquires the current time information to perform initial settings on the landscape lights; the initial settings only include the brightness and lighting range of the landscape lights; wherein the current time information includes: current season information and current time point information;

[0078] In the initial setting of landscape lights, it is necessary to determine whether it is spring, summer, autumn or winter according to the seasonal changes in the area; this can determine the time when it gets dark in the corresponding season; the initial setting can be pre-set by relevant staff based on historical data;

[0079] Furthermore, the monitoring is performed by the monitoring unit of the system; the monitoring unit is composed of a monitoring camera installed on the upper part of the landscape lamp; the monitoring camera can capture the park area with the landscape lamp as the center point and the maximum straight-line illumination distance as the radius; wherein the maximum straight-line illumination distance of the landscape lamp is 2.5 meters; for details, refer to Figure 3 Schematic diagram of the landscape lamp in the figure, the landscape lamp is composed of a lampshade 2, a lamp body 3 and a lamp post 4, and a surveillance camera 1 is installed on the lampshade 2 for monitoring.

[0080] Furthermore, the acquisition unit of the system acquires the video within the monitoring range through the monitoring camera, corresponding to the above step S20; wherein the acquisition unit continuously acquires the monitoring video within 5 minutes in units of time;

[0081] Furthermore, the preprocessing of the collected surveillance video is implemented by the data processing unit of the system, corresponding to the above-mentioned step S30; the specific data processing process includes: extracting video frames within a continuous time period of the scene video to obtain a video frame set; denoising the images in the video frame set to obtain a denoised image set; enhancing the images in the denoised image set to obtain an enhanced image set; color correcting the images in the enhanced image set to obtain a corrected image set; unifying the image sizes in the corrected image set to obtain the first image set.

[0082] In the embodiment of the present application, the video of the park area is preprocessed, mainly including: extracting video frames, image denoising, color correction, image enhancement and unifying image size; wherein, the video frames can display the actual effect and coverage of landscape lighting in real time. By analyzing these video frames, the brightness and color temperature of the landscape lights can be adjusted in time; removing noise from the image can reduce interference, making the identification and analysis of the landscape lighting effect more accurate, so that the lighting needs of each area can be determined more accurately; image enhancement can enhance the contrast and clarity of the image, making the lighting effect of the landscape lights more clearly visible; color correction can ensure that the lighting effects of different parts or different time periods remain consistent; unifying the size of the images of the landscape lighting area helps to improve the consistency and comparability of the data, simplify the data processing process, provide accurate spatial analysis, and facilitate long-term monitoring and optimization of lighting design.

[0083] Further, the control parameter adjustment unit of the system is used to adjust the control parameters of the landscape lights, corresponding to the above step S40;

[0084] In the process of adjusting the landscape lights, it is necessary to analyze the content in the image using the landscape light analysis model, including: crowd analysis, road surface analysis and vegetation analysis; the specific process is:

[0085] Input the image input layer in the first image set; the input layer can filter the images in the first image set; mainly remove the duplicate images in the first image set, the first image set includes 9000 images in total, and 4500 images are left after filtering, and the positions of people in the remaining images are spatially changed;

[0086] Further, the filtered image set is input into the recognition layer to obtain a first recognition result, a second recognition result and a third recognition result; the first recognition result refers to a crowd; the second recognition result refers to a road; the third recognition result refers to vegetation; wherein the recognition layer uses the Faster-RCNN model to identify the object type and location;

[0087] Furthermore, the marking layer performs marking according to the recognition result; the marking content includes: target type and target position;

[0088] Further, the analysis is performed at the analysis layer according to the marking result of the marking layer; wherein the analysis content includes: a crowd analysis module, a road surface analysis module and a vegetation analysis module;

[0089] The crowd analysis module is used to analyze the density of people within the monitoring range. The specific analysis process includes: obtaining crowd information according to the marked content; the crowd information is expressed as: CI = {(D i ,N i,L i ,Av i )|1≤i≤M}; where M represents the total number of people; D i It is represented as the straight-line distance from crowd i to the landscape light; L i Represented as the spatial distribution position of population i; N i It is represented by the total number of people in group i; Av i It is expressed as the average volume of each person in the population i; the average volume is expressed as: H n W is the height of the nth person; n It is expressed as the width of the nth person. The height and width of a person are obtained by roughly calculating the length and width of the rectangular detection box marked in the crowd recognition process according to the proportion.

[0090] Calculate the crowd density in the first image set; wherein the calculation formula of the crowd density is: cd i It is expressed as the crowd density of crowd i; Where A represents the monitoring area; where the farthest illumination distance of the landscape lamp is 2.5 meters, then the farthest illumination distance is the radius of the circle, and the area of ​​the camera to obtain the video is A = 2.5*2.5*π≈20m 2 ;

[0091] Obtain the population analysis result; the population analysis result is expressed as: PAR = {Max (D i ),L i ,CD|1≤i≤M}; Refer to Table 1, where the results of the crowd analysis in the monitored area are given;

[0092] Table 1. Population analysis results

[0093] Position number Crowd distance / m Crowd location description Crowd density index 1 0.5 Entrance 4 2 1.4 Main walking path 3 3 2.0 Lakeside 2 4 2.17 lawn 3

[0094] Table 1 shows the crowd analysis within the maximum illumination range of the landscape lamp. It can be concluded from the table that there are crowds at 4 locations within the monitoring range, namely the entrance, the main walkway, the lakeside and the lawn. The crowd density indexes are 4, 3, 2 and 3 respectively. In the embodiment of the present application, the crowd density index range is set to 1 to 5, where 1 is the lowest and 5 is the highest.

[0095] The road surface analysis module is used to analyze the unevenness of the road surface and the impact of water accumulation on the propagation of landscape light sources; the specific analysis process includes:

[0096] Acquire the road surface portion in the image according to the marking information;

[0097] Further, the pavement features of the pavement portion are extracted; wherein the pavement features include: texture features, color features and local features;

[0098] Further, the road surface is classified according to the extracted features; this classification is mainly used for the construction material of the road surface, such as asphalt concrete, gravel, lawn or sand and gravel;

[0099] Furthermore, the marked road surface is segmented according to the classification result; feature extraction is performed on each segmented area to obtain concave and convex features, including grayscale gradient, contour and color;

[0100] quantifying the degree of concavity and convexity of each segmented region according to the concavity and convexity feature;

[0101] Using an object detection algorithm for each segmented area to identify the shape and boundary of water accumulation on the road surface, and determine whether there is water accumulation on the road surface; when water accumulation occurs, obtaining the location and shape of the water accumulation;

[0102] Furthermore, according to the material of the road surface, the degree of unevenness of the road surface area, and the water accumulation, the influence of the road surfaces in different areas on the lighting of the landscape lights can be obtained;

[0103] The vegetation analysis module is used to analyze the absorption of light by vegetation and the impact on lighting conditions. The specific process includes:

[0104] According to the vegetation marks in the marked image set, a neural network is used to identify the plants to obtain the plant categories; wherein the neural network used for vegetation identification is a plant identification network constructed with a convolutional network as a framework;

[0105] The distribution position, growth status, height distribution and leaf coverage of the corresponding plants are obtained according to the plant category.

[0106] In the parameter control of landscape lights, the embodiment of the present application adopts a landscape light lighting analysis model; the model is mainly used for crowd analysis, road analysis and vegetation analysis; in the process of crowd analysis, the distribution area of ​​the crowd and the density of the flow of people are understood, so that the lighting range of the landscape lights and the lighting needs of different areas can be well grasped; in order to accurately adjust the landscape light lighting control parameters, the road surface factors and vegetation factors are also taken into consideration in this part; different road surface materials have different reflectivity and reflection characteristics; in addition, the width and curvature of the road surface adjust the parameters of the landscape lights; vegetation of different types and densities will affect the penetration and reflection of light. For example, dense leaves will block most of the light, resulting in shadows under the ground and uneven lighting effects. Therefore, it is also necessary to adjust the landscape light control parameters to optimize the lighting effect of the vegetation-covered area.

[0107] Furthermore, the results obtained by the analysis layer are input into the lighting parameter calculation layer to obtain the control parameters of the landscape lights; wherein the control parameters mainly include the lighting distance and lighting intensity of the landscape lights;

[0108] The lighting distance is compensated for the error according to the initial lighting distance of the landscape lamp and the farthest crowd distance in the crowd analysis result; according to Table 1, the farthest crowd distance is 2.17m, so the lighting distance of the landscape lamp should be adjusted to at least 2.17m;

[0109] The calculation formula of the lighting intensity is: Wherein, I represents the initial lighting intensity of the landscape lamp; LP i Expressed as the location factor of population i; CR k Expressed as the leaf coverage of vegetation k; RL b Indicates the concave and convex position of the road surface b; UD b It is expressed as the degree of roughness of the road surface b; PO b It is expressed as the degree of water accumulation on road surface b; ω 1i Represented as the weight parameter of population i; ω 2k Represented as the weight parameter of vegetation k; ω 3b The weight parameter representing the concave and convex position of the road surface b; ω 4b The weight parameter of the unevenness of the road surface b; ω 5b It is expressed as the weight parameter of the water accumulation degree of road surface b;

[0110] The distribution intensity of lights is adjusted according to the density of people in different areas; the adjustment standard can be whether the surveillance camera can clearly capture the crowd;

[0111] In the embodiment of the present application, the adjustment control parameters of the landscape lights are calculated through the lighting parameter calculation layer; including: lighting distance, lighting intensity and light distribution; the lighting distance is achieved through the farthest analysis of the crowd, so that the light coverage range of the landscape lights needs to reach a certain required range, thereby ensuring that sufficient lighting can be provided at the far end of the crowd activity area, which helps to reduce accidents and increase the safety of night activities; the calculation of lighting intensity is comprehensively calculated through the results of crowd distribution and density analysis, road surface analysis and vegetation analysis, which helps to design a lighting solution that is environmentally adaptable and sustainable; by dynamically adjusting the control parameters, the lighting system can operate flexibly under different environmental conditions and usage requirements, thereby reducing energy consumption and carbon footprint; the light distribution is mainly achieved through image analysis of the clarity of lighting in the crowd area, thereby ensuring good visibility at night.

[0112] Furthermore, the calculation result is outputted from the output layer as the first landscape light control parameter; the specific parameter adjustment is shown in Table 2;

[0113] Table 2. First landscape light control parameter description

[0114]

[0115] From the table, the initial lighting distance of the landscape lights can be adjusted from 1.5m to 2.17m; and the lighting intensity is adjusted for different areas with dense crowds; for example, the initial lighting intensity at the entrance is 2000 lumens / m 2 Adjust to 2200 lumens / m 2 The main walkway will have an initial lighting intensity of 1500 lumens / m 2 Adjust to 1400 lumens / m 2 ; The initial lighting intensity at the lakeside is 1000 lumens / m 2 Adjust to 900 lumens / m 2 ; The initial lighting intensity at the lawn is 1200 lumens / m 2 Adjust to 1100 lumens / m 2 .

[0116] Furthermore, the adaptive parameter adjustment of the landscape lights is implemented by the control parameter adaptive adjustment unit of the system, corresponding to the above step S50; the adaptive adjustment mainly predicts the change of the flow of people to adjust the control parameters of the landscape lights; in this way, the landscape lights can be adjusted in time, solving the problem of insufficient user experience caused by delayed adjustment of the landscape lights; the specific adaptive adjustment process includes:

[0117] Predicting the flow status of the crowd, and analyzing the landscape lighting analysis model according to the prediction results; wherein, analyzing the crowd in the first image set according to the analysis layer, obtaining a plurality of crowd analysis results in a continuous time period;

[0118] Further, extracting crowd change characteristics of the plurality of crowd analysis results; wherein the crowd change characteristics include: time characteristics, crowd location characteristics, crowd density characteristics, crowd movement speed characteristics and crowd flow direction characteristics;

[0119] Furthermore, a crowd flow transfer matrix is ​​established according to the crowd change characteristics;

[0120] The one-step transition probability is calculated according to the crowd flow transfer matrix; wherein the calculation formula of the one-step transition probability is:

[0121]

[0122] Among them, P A→B (t) represents the probability of one-step transition from point A to point B at time t; cd t,A It is expressed as the flow density of people at point A at time t; cdt,B is the flow density of people at point B at time t; cd t,C is the crowd density at time t; v A→B Expressed as the speed of movement from point A to point B;

[0123] In the embodiment of the present application, the one-step transition probability matrix analyzed based on the video data collected within 5 minutes is:

[0124]

[0125] The one-step transition probability is combined with the lighting information to predict the movement state of the crowd; wherein the function of predicting the movement state of the crowd is:

[0126] Among them, the function of predicting the movement state of the flow of people is:

[0127]

[0128] Among them, ΔS i (t+1) is the change in the crowd density at point i at time t+1; P i→j It is expressed as the probability of one-step transition from point i to point j at time t; L j (t) represents the lighting information at time t and point j; α represents the parameter for adjusting the lighting to affect the movement of people; d ij It is expressed as the distance between point i and point j;

[0129] In order to control landscape lights more intelligently and improve user experience, the present invention makes a secondary adjustment to the control parameters of landscape lights according to the predicted state of crowd movement; the prediction method establishes a transfer matrix by analyzing the changing characteristics of the crowd in the video, and predicts the movement state based on the matrix combined with the adjusted light information of the landscape lights; through accurate crowd flow prediction, the landscape lights can provide sufficient lighting in crowded areas according to actual needs, and reduce the light intensity or turn off some lamps in sparsely populated areas, thereby saving energy and reducing operating costs; at the same time, improving user experience.

[0130] Further, the first landscape light control parameter is adjusted to obtain a second landscape light control parameter;

[0131] Furthermore, the evaluation unit of the system evaluates the lighting effect of the adjusted landscape lamp, corresponding to the above step S60; the specific evaluation process includes:

[0132] Determine the lighting evaluation index of the landscape lamp; wherein the lighting evaluation index includes: lighting quality index, energy consumption index and user satisfaction;

[0133] Furthermore, the lighting quality indicators include: average illumination, clarity and illumination uniformity; the energy consumption indicators include: energy saving rate and energy consumption rate; the user satisfaction can be conducted by questionnaire survey;

[0134] Further, the various index data of the landscape lights after adjustment within the evaluation period are collected; wherein, in the embodiment of the present application, the index data specifically refers to the day as the unit, and the data within one month is collected; it is known that a total of 30 days of data are collected in this evaluation;

[0135] Further, the evaluation values ​​of each indicator are calculated;

[0136] The calculation formula for the evaluation value of the average illumination in the lighting quality index is: Among them, E_avg i Expressed as the average illuminance evaluation value on the i-th day; E i,n A is the measurement value of the nth measurement point on the i-th day; i,n It is expressed as the area of ​​the nth measurement point on the i-th day;

[0137] The clarity refers to the area still maintaining a good lighting effect under the minimum lighting illumination; the calculation formula is: E_min i Expressed as the minimum illumination on the i-th day; L_min i,n It is expressed as the minimum reflectance of the nth detection point on the i-th day; E i It is represented as the adjusted lighting intensity on the i-th day; k is represented as the beam distribution coefficient;

[0138] The calculation formula for the evaluation value of the illumination uniformity is: Among them, U i It is expressed as the uniformity of illumination on the i-th day;

[0139] The evaluation value of the lighting quality index is:

[0140] The calculation formula of energy saving rate in the energy consumption index is: Among them, S i Expressed as the energy saving rate on the i-th day; E_adj i E_uadj is the total energy consumption of the landscape lamp after adjustment on the i-th day; i It is represented by the total energy consumption of the landscape lamp without adjustment on the i-th day;

[0141] The calculation formula of the energy consumption rate is: Among them, C_ra i Expressed as the energy consumption rate on the i-th day; A_tatol iIt is represented as the total area of ​​the irradiated area on the i-th day; T is represented as the evaluation time; in the embodiment of the present application, the evaluation time is set to 5 hours, that is, from 19:00 to 24:00;

[0142] The calculation formula for the evaluation value of the energy consumption index is:

[0143] Wherein, the user satisfaction can be quantified by actual feedback from a questionnaire within 30 days on the user's satisfaction with the lighting effect;

[0144] Furthermore, according to the evaluation values ​​of each of the indicators, the comprehensive evaluation value of the lighting effect of the landscape lamp is calculated; the calculation formula of the comprehensive evaluation value of the lighting effect is: CE = c 1 *LQAV+c 2 *ECAV+c 3 *USEV; LQAV stands for lighting quality assessment value; ECAV stands for energy consumption assessment value; USEV stands for user satisfaction assessment value; c 1 Expressed as the weight coefficient for lighting quality assessment; c 2 Expressed as the weight coefficient of energy consumption assessment; c 3 Expressed as the weight coefficient of user satisfaction.

[0145] In order to prove that the present invention has a good effect on the control of landscape lights, an evaluation method is also proposed in this process; this method takes into account the lighting quality, energy consumption and user satisfaction; the lighting quality mainly includes the brightness of each lighting area and the uniformity of light; the energy consumption is mainly compared between before and after adjustment; the user satisfaction is obtained through the form of a questionnaire and the feedback results are obtained through quantitative data; this comprehensive evaluation method not only helps to evaluate the comprehensiveness and actual effect of the present invention on the control of landscape lights, but also can guide subsequent improvement and optimization measures.

[0146] Furthermore, the remote management unit of the system realizes control and management of the landscape lights;

[0147] The system also includes a music unit for playing music. The music type played can be based on pre-set music types, mainly in line with the theme, including: popular songs, events or related festivals.

[0148] In the embodiment of the present application, the method of the present invention is combined with the system to realize the control of the landscape lights when people gather in the park; the present invention mainly starts from the following main aspects: first, the monitoring video within the lighting range of the landscape lights is obtained, and the image is obtained by preprocessing; wherein, a landscape light lighting analysis model is used to analyze the crowd, road and vegetation in the image; in the process of crowd analysis, the distribution area of ​​the crowd and the density of the crowd are understood, so that the lighting range of the landscape lights and the lighting needs of different areas can be well grasped; different road surface materials have different reflectivity and reflection characteristics; in addition, the width and curvature of the road surface adjust the parameters of the landscape lights; vegetation of different types and densities will affect the penetration and reflection of the light; finally, in order to control the landscape lights more intelligently and improve the user experience, the control parameters of the landscape lights are adjusted according to the predicted state of the movement of the crowd. Secondary adjustment; the prediction method establishes a transfer matrix by analyzing the changing characteristics of the crowd in the video, and predicts the movement state based on the matrix combined with the light information after the landscape lamp is adjusted; through accurate crowd flow prediction, the landscape lamp can provide sufficient lighting in crowded areas according to actual needs, and reduce the light intensity or turn off some lamps in sparsely populated areas; in addition, in order to prove that the present invention has a good effect on the control of landscape lamps, an evaluation method is also proposed in this process; this method takes into account the lighting quality, energy consumption and user satisfaction; the lighting quality mainly includes the brightness and uniformity of light in each lighting area; the energy consumption is mainly compared between the unadjusted and adjusted conditions; the user satisfaction is obtained through quantitative data in the form of a questionnaire to obtain feedback results; this comprehensive evaluation method helps to evaluate the comprehensiveness and actual effect of the landscape lamp control effect of the present invention.

[0149] Embodiment 2:

[0150] In Example 1, the method of the present invention is combined with the system to realize the control of landscape lights in a park under the condition of crowd gathering; in order to further illustrate the feasibility of the method proposed by the present invention, a technical description will be further made in the embodiment of the present application, and the implementation scene can be a scenic spot; the specific process is basically the same as that of Example 1, mainly including:

[0151] Performing initial settings on the landscape lights according to current time information;

[0152] Further, a scene video is obtained through a camera;

[0153] Further, the scene video is preprocessed to obtain a first image set within a continuous time period;

[0154] Furthermore, the landscape lighting analysis model is used to analyze and obtain the control parameters of the landscape lighting;

[0155] Further, predicting the movement state of the crowd, and analyzing the landscape lighting analysis model according to the prediction results;

[0156] wherein, according to the analysis layer, a plurality of analysis results of the crowd in a continuous time period are obtained by analyzing the crowd in the first image set;

[0157] Extracting crowd change characteristics of the plurality of crowd analysis results; wherein the crowd change characteristics include: time characteristics, crowd location characteristics, crowd density characteristics, crowd movement speed characteristics and crowd flow direction characteristics;

[0158] Establishing a crowd flow transfer matrix according to the crowd change characteristics;

[0159] The one-step transition probability is calculated according to the crowd flow transfer matrix; wherein the calculation formula of the one-step transition probability is:

[0160]

[0161] Among them, P A→B (t) represents the probability of one-step transition from point A to point B at time t; cd t,A It is expressed as the flow density of people at point A at time t; cd t,B is the flow density of people at point B at time t; cd t,C is the crowd density at time t; v A→B Expressed as the speed of movement from point A to point B;

[0162] Obtaining the lighting information of each crowd area after the landscape lights are adjusted; wherein the lighting information includes: lighting intensity and lighting distribution;

[0163] The one-step transition probability is combined with the lighting information to predict the movement state of the crowd; wherein the function of predicting the movement state of the crowd is:

[0164]

[0165] Among them, ΔS i (t+1) is the change in the crowd density at point i at time t+1; P i→j It is expressed as the probability of one-step transition from point i to point j at time t; L j (t) represents the lighting information at time t and point j; α represents the parameter for adjusting the lighting to affect the movement of people; d ij It is expressed as the distance between point i and point j.

[0166] Further, the first landscape light control parameter is adjusted to obtain a second landscape light control parameter;

[0167] Furthermore, the second landscape lamp control parameter is evaluated to obtain a comprehensive evaluation value of the lighting effect of the landscape lamp; the specific process includes:

[0168] Determine the lighting evaluation index of the landscape lamp; wherein the lighting evaluation index includes: lighting quality index, energy consumption index and user satisfaction;

[0169] The lighting quality indicators include: average illumination, clarity and illumination uniformity; the energy consumption indicators include: energy saving rate and energy consumption rate; the user satisfaction can be conducted through questionnaire survey;

[0170] Further, collecting various indicator data of the landscape lights after adjustment within the evaluation period;

[0171] Further, the evaluation values ​​of each indicator are calculated;

[0172] Furthermore, according to the evaluation values ​​of each of the indicators, the comprehensive evaluation value of the lighting effect of the landscape lamp is calculated; the calculation formula of the comprehensive evaluation value of the lighting effect is: CE = c 1 *LQAV+c 2 *ECAV+c 3 *USEV; LQAV stands for lighting quality assessment value; ECAV stands for energy consumption assessment value; USEV stands for user satisfaction assessment value; c 1 Expressed as the weight coefficient for lighting quality assessment; c 2 Expressed as the weight coefficient of energy consumption assessment; c 3 Expressed as the weight coefficient of user satisfaction.

[0173] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A landscape lighting control method based on crowd gathering, characterized in that: include: Performing initial settings on the landscape lights according to current time information; Obtaining a scene video; wherein the range of the scene video is centered on the landscape lamp and the maximum lighting distance is the radius; Preprocessing the scene video to obtain a first image set within a continuous time period; The landscape lighting analysis model is used for analysis, and the specific process includes: The images in the first image set are input from the input layer to the recognition layer to obtain a first recognition result, a second recognition result and a third recognition result; wherein the first recognition result represents a crowd; the third recognition result represents a road surface; and the third recognition result represents a plant; The first recognition result, the second recognition result and the third recognition result are marked by a marking layer to obtain a marked image set; Inputting the marked image set into the analysis layer to obtain crowd analysis results, road surface analysis results and vegetation analysis results; Inputting the crowd analysis result, the road surface analysis result and the vegetation analysis result into the lighting parameter calculation layer to obtain the control parameters of the landscape lamp; The control parameters are outputted via the output layer as the first landscape light control parameters; Predicting the movement state of a crowd, and adjusting the first landscape light control parameter using the landscape light illumination analysis model according to the prediction result to obtain the second landscape light control parameter; The second landscape lamp control parameter is evaluated to obtain a comprehensive evaluation value of the lighting effect of the landscape lamp.

2. A landscape lighting control method based on crowd gathering conditions according to claim 1, characterized in that: The preprocessing includes: extracting video frames within a continuous time period of the scene video to obtain a video frame set; denoising images in the video frame set to obtain a denoised image set; enhancing images in the denoised image set to obtain an enhanced image set; color correcting images in the enhanced image set to obtain a corrected image set; unifying the image sizes in the corrected image set to obtain the first image set.

3. A landscape lighting control method based on crowd gathering conditions according to claim 1, characterized in that: The analysis layer includes: crowd analysis module, road surface analysis module and vegetation analysis module: The specific analysis process of the crowd analysis module includes: obtaining crowd information in the marked image set; the crowd information is expressed as: CI = {(D i ,N i ,L i ,Av i )|1≤i≤M}; where M represents the total number of people; D i It is represented as the straight-line distance from crowd i to the landscape light; L i Represented as the spatial distribution position of population i; N i It is represented by the total number of people in group i; Av i It is expressed as the average volume of each person in the population i; the average volume is expressed as: H n W is the height of the nth person; n It is expressed as the width of the nth person; Calculate the crowd density in the first image set; wherein the calculation formula of the crowd density is: cd i It is expressed as the crowd density of crowd i; Where A represents the monitoring area; Obtain the population analysis result; the population analysis result is expressed as: PAR = {Max (D i ),L i ,CD|1≤i≤M}; The specific analysis process of the road surface analysis module includes: extracting road surface features of the images in the marked image set, including texture features, color features and local features; classifying according to the road surface features to obtain road surface types; Performing road area segmentation on the images in the marked image set; performing feature extraction on each segmented area to obtain concave and convex features, including grayscale gradient, contour and color; quantifying the degree of concavity and convexity of each segmented region according to the concavity and convexity feature; Using an object detection algorithm for each segmented area to identify the shape and boundary of water accumulation on the road surface, and determine whether there is water accumulation on the road surface; when water accumulation occurs, obtaining the location and shape of the water accumulation; The specific process of the vegetation analysis module includes: Using a neural network to identify plants based on the vegetation tags in the tag image set to obtain plant categories; The distribution position, growth status, height distribution and leaf coverage of the corresponding plants are obtained according to the plant category.

4. The method for controlling landscape lighting based on crowd gathering according to claim 1, characterized in that: The lighting parameter calculation layer is used to calculate the lighting distance and lighting intensity of the landscape lamp; Wherein, the lighting distance is compensated for errors according to the initial lighting distance of the landscape lamp and the farthest crowd distance in the crowd analysis result; The calculation formula of the lighting intensity is: Wherein, I represents the initial lighting intensity of the landscape lamp; LP i Expressed as the location factor of population i; CR k Expressed as the leaf coverage of vegetation k; RL b Indicates the concave and convex position of the road surface b; UD b It is expressed as the degree of roughness of the road surface b; PO b It is expressed as the degree of water accumulation on road surface b; ω 1i Represented as the weight parameter of population i; ω 2k Represented as the weight parameter of vegetation k; ω 3b The weight parameter representing the concave and convex position of the road surface b; ω 4b The weight parameter of the unevenness of the road surface b; ω 5b It is expressed as a weight parameter of the degree of water accumulation on road surface b.

5. The method for controlling landscape lighting based on crowd gathering according to claim 1, characterized in that: The crowd flow movement state prediction includes: Analyzing the crowd in the first image set according to the analysis layer, obtaining a plurality of crowd analysis results in a continuous time period; Extracting crowd change characteristics of the plurality of crowd analysis results; wherein the crowd change characteristics include: time characteristics, crowd location characteristics, crowd density characteristics, crowd movement speed characteristics and crowd flow direction characteristics; Establishing a crowd flow transfer matrix according to the crowd change characteristics; The one-step transition probability is calculated according to the crowd flow transfer matrix; wherein the calculation formula of the one-step transition probability is: Among them, P A→B (t) represents the probability of one-step transition from point A to point B at time t; cd t,A It is expressed as the flow density of people at point A at time t; cd t,B is the flow density of people at point B at time t; cd t,C is the crowd density at time t; v A→B Expressed as the speed of movement from point A to point B; Obtaining the lighting information of each crowd area after the landscape lights are adjusted; wherein the lighting information includes: lighting intensity and lighting distribution; The one-step transition probability is combined with the lighting information to predict the movement state of the crowd; wherein the function of predicting the movement state of the crowd is: Among them, ΔS i (t+1) is the change in crowd density at point i at time t+1; P i→j It is expressed as the probability of one-step transition from point i to point j at time t; L j (t) represents the lighting information at time t and point j; α represents the parameter for adjusting the lighting to affect the movement of people; d ij It is expressed as the distance between point i and point j.

6. The method for controlling landscape lighting based on crowd gathering according to claim 1, characterized in that: The parameter adjustment evaluation of the landscape lamp includes: Determine the lighting evaluation index of the landscape lamp; wherein the lighting evaluation index includes: lighting quality index, energy consumption index and user satisfaction; The lighting quality indicators include: average illumination, clarity and illumination uniformity; the energy consumption indicators include: energy saving rate and energy consumption rate; the user satisfaction can be conducted through questionnaire survey; Collect various indicator data after the landscape lights are adjusted during the evaluation period; Calculate the evaluation value of each indicator; According to the evaluation values ​​of each indicator, the comprehensive evaluation value of the lighting effect of the landscape lamp is calculated; the calculation formula of the comprehensive evaluation value of the lighting effect is: CE=c1*LQAV+c2*ECAV+c3*USEV; wherein, LQAV represents the lighting quality evaluation value; ECAV represents the energy consumption evaluation value; USEV represents the user satisfaction evaluation value; c1 represents the weight coefficient of the lighting quality evaluation; c2 represents the weight coefficient of the energy consumption evaluation; c3 represents the weight coefficient of the user satisfaction.

7. A landscape lighting control system based on crowd gathering, characterized in that: The system includes: a monitoring unit, a collection unit, a data processing unit, a control parameter adjustment unit, a control parameter adaptive adjustment unit, an evaluation unit, a music unit and a remote management unit; wherein the monitoring unit is used to monitor the scene within the lighting distance of the landscape lamp; the collection unit is used to collect the video within the monitoring range of the monitoring unit; the data processing unit is used to pre-process the video data; the control parameter adjustment unit is used to adjust the lighting parameters of the landscape lamp; the control parameter adaptive adjustment unit is used to perform secondary optimization on the lighting parameters of the landscape lamp; the evaluation unit is used to evaluate the effect of the adjusted lighting parameters of the landscape lamp; the music unit is used for the landscape lamp to play music; the remote management unit is used to remotely manage the landscape lamp; The control parameter adjustment unit adopts a landscape lighting analysis model; wherein the landscape lighting analysis model includes: an input layer, a recognition layer, a marking layer, an analysis layer, a lighting parameter calculation layer and an output layer; The analysis layer includes: crowd analysis module, road surface analysis module and vegetation analysis module: The specific analysis process of the crowd analysis module includes: Obtain crowd information in the marked image set; the crowd information is expressed as: CI = {(D i ,N i ,L i ,Av i )|1≤i≤M}; where M represents the total number of people; D i It is represented as the straight-line distance from crowd i to the landscape light; L i Represented as the spatial distribution position of population i; N i It is represented by the total number of people in group i; Av i It is expressed as the average volume of each person in the population i; the average volume is expressed as: H n W is the height of the nth person; n It is expressed as the width of the nth person; Calculate the crowd density in the first image set; wherein the calculation formula of the crowd density is: cd i It is expressed as the crowd density of crowd i; Where A represents the monitoring area; Obtain the population analysis result; the population analysis result is expressed as: PAR = {Max (D i ),L i ,CD|1≤i≤M}; The specific analysis process of the road surface analysis module includes: Extracting road surface features of the images in the marked image set, including texture features, color features and local features; classifying according to the road surface features to obtain road surface types; Performing road area segmentation on the images in the marked image set; performing feature extraction on each segmented area to obtain concave and convex features, including grayscale gradient, contour and color; quantifying the degree of concavity and convexity of each segmented region according to the concavity and convexity feature; Using an object detection algorithm for each segmented area to identify the shape and boundary of water accumulation on the road surface, and determine whether there is water accumulation on the road surface; when water accumulation occurs, obtaining the location and shape of the water accumulation; The specific process of the vegetation analysis module includes: using a neural network to identify plants according to the vegetation tags in the tag image set to obtain plant categories; obtaining the distribution position, growth status, height distribution and leaf coverage of the corresponding plants according to the plant categories; The lighting parameter calculation layer is used to calculate the lighting distance and lighting intensity of the landscape lamp; Wherein, the lighting distance is compensated for errors according to the initial lighting distance of the landscape lamp and the farthest crowd distance in the crowd analysis result; The calculation formula of the lighting intensity is: Wherein, I represents the initial lighting intensity of the landscape lamp; LP i Expressed as the location factor of population i; CR k Expressed as the leaf coverage of vegetation k; RL b Indicates the concave and convex position of the road surface b; UD b It is expressed as the degree of roughness of the road surface b; PO b It is expressed as the degree of water accumulation on road surface b; ω 1i Represented as the weight parameter of population i; ω 2k Represented as the weight parameter of vegetation k; ω 3b The weight parameter representing the concave and convex position of the road surface b; ω 4b The weight parameter of the unevenness of the road surface b; ω 5b It is expressed as a weight parameter of the degree of water accumulation on road surface b.

8. A landscape lighting control system based on crowd gathering conditions according to claim 7, characterized in that: The data processing unit comprises: Extract video frames within a continuous time period of a scene video to obtain a video frame set; denoise the images in the video frame set to obtain a denoised image set; enhance the images in the denoised image set to obtain an enhanced image set; perform color correction on the images in the enhanced image set to obtain a corrected image set; unify the image sizes in the corrected image set to obtain a first image set.

Images