Illumination control method based on smart park

Abstract

The invention relates to a lighting control method based on a smart park. The method comprises the following steps: S1, managing equipment information; s2, according to longitude and latitude coordinates of the central area of the park, determining a starting position of a code to realize a minimum unit of street lamp control; s3, paying attention to indexes according to crowd dense areas, sidewalk roads, entrance and exit areas and the like; s4, realizing illumination strategy configuration management work; s5, mode management configuration is achieved; s6, realizing a region-based popularity index calculation method; and S7, realizing intelligent illumination control based on the popularity index through an illumination control scheduling service. Gridding of street lamp control areas is realized on the basis of an electronic map, and intelligent control of brightness of street lamps and street lamp switches in each area is realized through a popularity index algorithm of the areas. The technical implementation scheme comprises the steps of park gridding management, regional popularity index calculation and illumination control scheduling.

Description

Technical Field

[0001] The present invention relates to the field of smart park and smart scenic spot application technology, and in particular to a lighting control method based on a smart park. Background Art

[0002] At night, there are usually people walking or exercising in the park, and street lamps are built along the sidewalks in the park to illuminate them; however, the number of people coming to the park varies at different time periods at night, especially late at night, when there are fewer people and many sections of the road are deserted, and the park's street lamps are still fully turned on for lighting, wasting electricity; and the number of people entering each park entrance is different, resulting in a different number of people on the road section connected to each entrance. When there is no one on the road section connected to some entrances, the fully turned on street lamps waste electricity.

[0003] With the development of urbanization, high-rise buildings in cities have destroyed the original green vegetation, so we must artificially create urban parks to improve the urban ecological environment and provide a good environment for citizens to rest and entertain.

[0004] The construction of large-scale parks or trails and the subsequent maintenance and management of urban parks are currently the most concerned issues for every city. Under the background of "dual carbon", how technological innovation can enable energy conservation and emission reduction is also a problem that every city needs to solve. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a lighting control method based on a smart park to achieve intelligent control of the brightness of street lamps in each area and the intelligent control of street lamp switches. The technical implementation scheme includes park grid management, regional popularity index calculation, and lighting control scheduling.

[0006] To achieve the above object, the present invention adopts the following technical solution: a lighting control method based on a smart park, comprising the following steps:

[0007] Step S1: Manage the device information, realize the management of the basic information of the device, connect the street light equipment of different manufacturers, and realize the switch operation of the light through the manufacturer's API interface; connect the camera of different manufacturers to realize the flow of people monitoring;

[0008] Step S2: Determine the starting position of the coding according to the longitude and latitude coordinates of the central area of ​​the park, determine the side length of the unit grid organized in the coding area according to the sparsity of the position of interest in the coding area or according to the user's settings; slice the park into square grids according to the side length and the park coverage area, and set the area code for each grid; rely on the longitude and latitude of the actual installation location of the street lamp and the grid inclusion relationship, bind the area code and the device code to realize the minimum unit of street lamp control;

[0009] Step S3: according to the focus indicators such as crowded areas, pedestrian roads, entrance and exit areas, etc., the areas are grouped into different groups, and cameras are deployed according to the actual situation, and the camera device codes are bound to the group code relationship;

[0010] Step S4: Implement lighting strategy configuration management, support on-time switching and brightness control of grouped devices based on weekly rules, allow users to configure the planned on-off time period for each day of the week, configure the brightness of lighting for each period of the day, and support lighting strategy configuration based on holidays;

[0011] Step S5: Implement mode management configuration, support switching and scheduling between intelligent lighting control based on popularity index, user-defined lighting strategy control, and user manual light switch control, and support lighting control scheduling parameter configuration;

[0012] Step S6: Implement a method for calculating the popularity index based on a region; use computer vision technology and SVM to detect the head and shoulders of pedestrians, and use the hue-saturation histogram to quickly detect and track a single pedestrian in the region of interest, count the number of pedestrians in the region and obtain the time a single pedestrian stays in the defined region, and calculate the popularity index of the region based on the number of people in a unit time and the area of ​​the region of interest;

[0013] Step S7: Implement intelligent lighting control based on popularity index through lighting control scheduling service.

[0014] In a preferred embodiment, step 6 specifically comprises: calculating the regional popularity index, constructing a positive and negative sample data set of pedestrian images, and scaling all positive and negative samples to sample images with a width of 32 pixels and a height of 32 pixels, extracting the HOG feature vector of the sample image based on the contour information, and using a mixed Gaussian model to extract the motion foreground, denoted as FG, as a mask for the pedestrian head and shoulder detection area, setting label vectors for all scaled positive and negative sample images, with the positive sample being 1 and the negative sample being -1, and inputting the HOG feature vector and the corresponding label vector of each sample into the SVM for training to obtain an SVM classifier based on the pedestrian head and shoulder contour HOG.

[0015] In a preferred embodiment, the trained SVM classifier is used to detect the head and shoulders of pedestrians in the region of interest. Specifically, the FG extracted previously is used as the mask of the region of interest and the trained classifier is used to detect the image in the motion foreground region, the head and shoulders of pedestrians are identified, and marked with the minimum envelope rectangle. The envelope rectangle that satisfies the minimum envelope rectangle length = width and is smaller than W pixels is added to the queue of new detection of pedestrian head and shoulder targets NewMans = {R i|i=1,2,...,n}; track the pedestrian head and shoulder targets, and calculate the Euclidean distance between the newly detected pedestrian head and shoulder targets and the center of the pedestrian head and shoulder objects in the original queue:

[0016] Original queue OldMans = {R j |j=1,2,...,m}

[0017] Euclidean distance

[0018] Among them, the point (x i ,y i ) is the pedestrian rectangle R i The center of the point (x j ,y j ) is the pedestrian rectangle R j If the Euclidean distance is less than D, where D is the average width of randomly selected pedestrian head and shoulder samples, the newly detected pedestrian head and shoulder target replaces the original queue OldMans = {R j |j=1,2,...,m}, and add the object to the tracking queue Tracks={R q |q=1,2,...,Q}; If the newly detected pedestrian head and shoulder target does not find a replacement object in the original queue, it will be added to the original queue and its time constraint will be set to 0; If the object in the original queue has not been updated in this frame, the time constraint of the object will be increased by 1, and it will be replaced and updated with the object in the tracking queue; If the time constraint of an object in the queue is greater than 15, that is, it has not been updated for 15 consecutive frames, it is assumed that the object has disappeared or is not in the area of ​​interest, and it will be kicked out of the detection queue and tracking queue, the people counter will be increased by 1, and the time the object exists in the detection queue, that is, the personal stay time TS, will be recorded. i .

[0019] In a preferred embodiment, meanShift tracking is performed based on the hue constraint, the hue of the pedestrian's head within the envelope rectangle of the detected pedestrian's head and shoulders target is calculated, the hue of the background part in the envelope rectangle is eliminated, and a histogram is calculated. The histogram reverse projection of the detected pedestrian's head and shoulders target is calculated based on the hue histogram, and then the object in the tracking queue is tracked according to meanShift, and the position information of the object in the tracking queue is updated;

[0020] The popularity index of a specific area is calculated as follows: the user sets the statistical time and the detection area, and the individual stay time and number of people are calculated by the tracking algorithm, and the popularity index of the area is derived:

[0021]

[0022] Among them, λ is the popularity index, l is the number of people, TSi is the individual’s stay time, S is the area of ​​the detection area, t is the statistical time, and η is a constant.

[0023] In a preferred embodiment, in step 7, under the popularity index-based intelligent control model, the user is allowed to set different brightness for the groups according to the five levels of the popularity index. The intelligent scheduling system will automatically perform intelligent control on the lighting of different areas according to the user configuration, the popularity index data value of the area, and the sunrise and sunset time.

[0024] Compared with the prior art, the present invention has the following beneficial effects: a lighting control method based on a smart park, which realizes the gridding of street light control areas based on electronic maps, and realizes intelligent control of the brightness of street lights in each area and the switch of street lights through the regional popularity index algorithm. The technical implementation scheme includes park grid management, regional popularity index calculation, and lighting control scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of functional modules of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0027] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0028] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0029] refer to Figure 1 A lighting control method based on smart parks, which realizes gridding of control areas through electronic maps, and realizes intelligent control of street lamp brightness and switches in each area based on the popularity index of the area. The technical implementation scheme includes park grid management, regional popularity index calculation, and lighting control scheduling.

[0030] The working process of this embodiment is as follows:

[0031] Step S1: Manage the device information, realize the management of the basic information of the device, connect the street light equipment of different manufacturers, and realize the switch operation through the manufacturer's API interface. Connect the cameras of different manufacturers to realize the flow of people monitoring.

[0032] Step S2: The system determines the starting position of the code according to the longitude and latitude coordinates of the central area of ​​the park, and determines the side length of the unit grid organized in the coding area according to the sparsity of the location of interest in the coding area or according to the user's settings. The park is sliced ​​into square grids based on the side length and the park coverage area, and a regional code is set for each grid. Relying on the longitude and latitude of the actual installation location of the street lamp and the grid inclusion relationship, the regional code and the device code are bound to realize the minimum unit of street lamp control.

[0033] Step S3: According to the focus indicators such as crowded areas, pedestrian roads, entrance and exit areas, etc., the areas are combined into different groups (larger logical areas), and cameras are deployed according to the actual situation, and the camera device codes are bound to the group coding relationship.

[0034] Step S4: Implement lighting strategy configuration management, support on-time switching and brightness control of grouped devices based on weekly rules, allow users to configure the planned switching time periods for each day of the week, configure the brightness of the lighting for each period of the day, and support lighting strategy configuration based on holidays.

[0035] Step S5: Implement mode management configuration, support switching and scheduling between intelligent lighting control based on popularity index, user-defined lighting strategy control, and user manual light switch control, and support lighting control scheduling parameter configuration.

[0036] Step S6: Implement a method for calculating the popularity index based on the region. Use computer vision technology and SVM to detect the head and shoulders of pedestrians, and use the hue-saturation histogram to quickly detect and track a single pedestrian in the region of interest. Count the number of pedestrians in the region and obtain the time a single pedestrian stays in the defined area. Combine the number of people per unit time and the area of ​​the region of interest to calculate the popularity index of the region.

[0037] Step S7: Implement intelligent lighting control based on popularity index through lighting control scheduling service.

[0038] Specifically, the park grid management, in order to facilitate the management of the lighting control system, combines the park's geographical location with the longitude and latitude coordinates of the electronic map, and divides the park into small areas based on the longitude and latitude coordinates. The system first determines the starting position of the area code based on the longitude and latitude coordinates of the central area of ​​the park, and determines the side length of the unit grid organized in the coding area based on the sparsity of the location of interest in the coding area or according to the user's settings. According to the grid side length combined with the park coverage area, the park is sliced ​​into square grids (areas), and the area code is set for each grid. Relying on the longitude and latitude of the actual installation location of the street lamp and the grid inclusion relationship, the area code and the device code are bound to realize the minimum unit of street lamp control. Secondly, based on the focus indicators such as crowded areas, pedestrian roads, and park entrance and exit areas, the areas are combined into different groups (larger logical areas), and cameras are deployed according to actual conditions, and the camera device code is bound to the group code.

[0039] Regional popularity index calculation, construct positive and negative sample data sets of pedestrian images, and scale all positive and negative samples to sample images with a width of 32 pixels and a height of 32 pixels. Extract the HOG feature vector of the sample image based on the contour information, and use the mixed Gaussian model to extract the motion foreground, denoted as FG, as the mask of the pedestrian head and shoulder detection area. Set label vectors for all scaled positive and negative sample images, with positive samples as 1 and negative samples as -1. Input the HOG feature vector and corresponding label vector of each sample into SVM for training, and obtain an SVM classifier based on the pedestrian head and shoulder contour HOG.

[0040] The trained SVM classifier is used to detect the head and shoulders of pedestrians in the region of interest. Specifically, the FG extracted previously is used as the mask of the region of interest and the trained classifier is used to detect the image in the motion foreground area, the head and shoulders of pedestrians are identified, and marked with the minimum envelope rectangle. The envelope rectangle that satisfies the minimum envelope rectangle length = width and is smaller than W pixels is added to the queue of new detection of pedestrian head and shoulder targets NewMans = {R i |i=1,2,...,n}. Track the pedestrian head and shoulder targets and calculate the Euclidean distance between the newly detected pedestrian head and shoulder targets and the center of the pedestrian head and shoulder objects in the original queue:

[0041] Original queue OldMans = {R j |j=1,2,...,m}

[0042] Euclidean distance

[0043] Among them, the point (x i ,y i ) is the pedestrian rectangle R i The center of the point (x j ,yj ) is the pedestrian rectangle R j If the Euclidean distance is less than D, where D is the average width of randomly selected pedestrian head and shoulder samples, the newly detected pedestrian head and shoulder target replaces the original queue OldMans = {R j |j=1,2,...,m}, and add the object to the tracking queue Tracks={R q |q=1,2,...,Q}; If the newly detected pedestrian head and shoulder target does not find a replacement object in the original queue, it will be added to the original queue and its time constraint will be set to 0; If the object in the original queue has not been updated in this frame, the time constraint of the object will be increased by 1, and it will be replaced and updated with the object in the tracking queue; If the time constraint of an object in the queue is greater than 15, that is, it has not been updated for 15 consecutive frames, it is assumed that the object has disappeared or is not in the area of ​​interest, and it will be kicked out of the detection queue and tracking queue, the people counter will be increased by 1, and the time the object exists in the detection queue, that is, the personal stay time TS, will be recorded. i .

[0044] MeanShift tracking is performed based on hue constraints. The hue of the pedestrian's head within the envelope rectangle of the detected pedestrian's head and shoulders is calculated. The background hue in the envelope rectangle is eliminated and a histogram is calculated. The histogram of the detected pedestrian's head and shoulders is reversely projected based on the hue histogram. Then, the objects in the tracking queue are tracked according to meanShift, and the position information of the objects in the tracking queue is updated.

[0045] The popularity index of a specific area is calculated as follows: the user sets the statistical time and the detection area, and the individual stay time and number of people are calculated by the tracking algorithm, and the popularity index of the area is derived:

[0046]

[0047] Among them, λ is the popularity index, l is the number of people, TS i is the individual’s stay time, S is the area of ​​the detection area, t is the statistical time, and η is a constant.

[0048] Lighting control and scheduling realizes intelligent lighting control based on popularity index, automatic control of user-defined lighting strategies, and switching and scheduling between user manual light on / off control. Under the intelligent control model based on popularity index, users are allowed to set different brightness for groups according to the 5 levels of popularity index (lighting will be turned off when the brightness is zero). The intelligent scheduling system will automatically perform intelligent control on lighting in different areas based on user configuration, regional popularity index data value, and sunrise and sunset time (calculated based on the latitude and longitude of the device).

Claims

1. A lighting control method based on a smart park, characterized in that: The following steps are involved: Step S1: Manage the device information, realize the management of the basic information of the device, connect the street light equipment of different manufacturers, and realize the switch operation of the light through the manufacturer's API interface; connect the camera of different manufacturers to realize the flow of people monitoring; Step S2: determining the starting position of the coding according to the longitude and latitude coordinates of the central area of ​​the park, determining the side length of the unit grid organized in the coding area in a grid manner according to the sparsity of the position of interest in the coding area or according to the user's setting; slicing the park into square grids according to the side length combined with the park coverage area, and setting the area coding for each grid; Relying on the latitude and longitude of the actual installation location of the street lamp and the grid inclusion relationship, the area code and the device code are bound to realize the minimum unit of street lamp control; Step S3: according to the focus indicators such as crowded areas, pedestrian roads, entrance and exit areas, etc., the areas are grouped into different groups, and cameras are deployed according to the actual situation, and the camera device codes are bound to the group code relationship; Step S4: Implement lighting strategy configuration management, support on-time switching and brightness control of grouped devices based on weekly rules, allow users to configure the planned on-off time period for each day of the week, configure the brightness of lighting for each period of the day, and support lighting strategy configuration based on holidays; Step S5: Implement mode management configuration, support switching and scheduling between intelligent lighting control based on popularity index, user-defined lighting strategy control, and user manual light switch control, and support lighting control scheduling parameter configuration; Step S6: Implement a method for calculating the popularity index based on a region; use computer vision technology and SVM to detect the head and shoulders of pedestrians, and use the hue-saturation histogram to quickly detect and track a single pedestrian in the region of interest, count the number of pedestrians in the region and obtain the time a single pedestrian stays in the defined region, and calculate the popularity index of the region based on the number of people in a unit time and the area of ​​the region of interest; Step S7: Implement intelligent lighting control based on popularity index through lighting control scheduling service.

2. The lighting control method based on smart park according to claim 1 is characterized in that: The step 6 specifically includes: calculating the regional popularity index, constructing a positive and negative sample data set of pedestrian images, and scaling all positive and negative samples to sample images with a width of 32 pixels and a height of 32 pixels, extracting the HOG feature vector of the sample image based on the contour information, and using a mixed Gaussian model to extract the motion foreground, denoted as FG, as a mask for the pedestrian head and shoulder detection area, setting label vectors for all scaled positive and negative sample images, with the positive sample being 1 and the negative sample being -1, and inputting the HOG feature vector and the corresponding label vector of each sample into the SVM for training to obtain an SVM classifier based on the pedestrian head and shoulder contour HOG.

3. The lighting control method based on smart park according to claim 2 is characterized in that: The trained SVM classifier is used to detect the head and shoulders of pedestrians in the region of interest. Specifically, the FG extracted previously is used as the mask of the region of interest and the trained classifier is used to detect the image in the motion foreground area, the head and shoulders of pedestrians are identified, and marked with the minimum envelope rectangle. The envelope rectangle that satisfies the minimum envelope rectangle length = width and is smaller than W pixels is added to the queue of new detection of pedestrian head and shoulder targets NewMans = {R i |i=1,2,...,n}; track the pedestrian head and shoulder targets, and calculate the Euclidean distance between the newly detected pedestrian head and shoulder targets and the center of the pedestrian head and shoulder objects in the original queue: Original queue OldMans = {R j |j=1,2,...,m} Euclidean distance Among them, the point (x i ,y i ) is the pedestrian rectangular box R i The center of the point (x j ,y j ) is the pedestrian rectangular box R j If the Euclidean distance is less than D, where D is the average width of randomly selected pedestrian head and shoulder samples, the newly detected pedestrian head and shoulder target replaces the original queue OldMans = {R j |j=1,2,...,m}, and add the object to the tracking queue Tracks={R q |q=1,2,...,Q}; If the newly detected pedestrian head and shoulder target does not find a replacement object in the original queue, it will be added to the original queue and its time constraint will be set to 0; If the object in the original queue has not been updated in this frame, the time constraint of the object will be increased by 1, and it will be replaced and updated with the object in the tracking queue; If the time constraint of an object in the queue is greater than 15, that is, it has not been updated for 15 consecutive frames, it is assumed that the object has disappeared or is not in the area of ​​interest, and it will be kicked out of the detection queue and tracking queue, the people counter will be increased by 1, and the time the object exists in the detection queue, that is, the personal stay time TS, will be recorded. i .

4. The lighting control method based on smart park according to claim 3 is characterized in that: Based on the hue constraint, meanShift tracking is performed to calculate the hue of the pedestrian's head within the envelope rectangle of the detected pedestrian's head and shoulders, remove the hue of the background part in the envelope rectangle, and calculate the histogram. The histogram reverse projection of the detected pedestrian's head and shoulders is calculated based on the hue histogram, and then the object in the tracking queue is tracked according to meanShift, and the position information of the object in the tracking queue is updated; The popularity index of a specific area is calculated as follows: the user sets the statistical time and the detection area, and the individual stay time and number of people are calculated by the tracking algorithm, and the popularity index of the area is derived: Among them, λ is the popularity index, l is the number of people, TS i is the individual’s stay time, S is the area of ​​the detection area, t is the statistical time, and η is a constant.

5. The lighting control method based on smart park according to claim 1 is characterized in that: In step 7, under the popularity index-based intelligent control model, the user is allowed to set different brightness for the groups according to the five levels of the popularity index. The intelligent scheduling system will automatically perform intelligent control on the lighting of different areas according to the user configuration, the popularity index data value of the area, and the sunrise and sunset time.

Images