Intelligent lighting lamp regulation and control system for environment monitoring image processing

Abstract

The invention discloses an intelligent lighting lamp regulation and control system for environment monitoring image processing, and relates to the field of image processing, and the system is characterized in that a mode construction module designs a mode of a lighting lamp according to a rule topology network, and obtains a currently selected mode; the data acquisition module is used for acquiring comprehensive environment data and performing splicing processing on the comprehensive environment data to obtain a comprehensive panoramic image set; the data aggregation processing module is used for carrying out adaptive segmentation on the comprehensive panoramic image set, and carrying out feature aggregation on segmented images by using a K-U clustering method to obtain a feature data set; and the regulation and control module analyzes the environment of the sub-region based on the feature data set, generates a feature scheme, and regulates and controls the illuminating lamp by using PID control according to the feature scheme, thereby realizing intelligent regulation and control of the illuminating lamp based on environment monitoring image processing.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and more specifically, to an intelligent control system for lighting lamps for environment monitoring image processing. Background Art

[0002] In some public scenes, traditional lighting systems can only provide fixed lighting modes and are difficult to switch dynamically according to the requirements of different scenes (such as energy saving, high efficiency, and free adjustment). Traditional lighting control systems may rely on a single sensor (such as light or personnel detection sensors) to collect environmental information, resulting in a single source of information and insufficient accuracy and comprehensiveness of regulation. As a result, traditional lighting systems often adopt a unified brightness control method, resulting in energy waste.

[0003] In addition, when capturing images of lighting environments, especially when capturing images continuously and uninterruptedly, multi-camera monitoring equipment will generate a large amount of image data, making it difficult to conduct in-depth mining during image processing. Moreover, when analyzing images, due to the characteristics of light source diffusion and attenuation, as well as the superposition and interlacing of multiple light sources, the acquired image colors and brightness vary, which greatly increases the difficulty of image analysis and makes light source control even more difficult.

[0004] Therefore, based on the above problems, it is necessary to design intelligent control of lighting lamps based on environmental monitoring image processing. Summary of the invention

[0005] In view of the problems existing in the prior art, the purpose of the present invention is to provide an intelligent control system for lighting lamps based on environmental monitoring image processing to achieve intelligent control of lighting lamps.

[0006] To achieve the above object, the present invention provides the following technical solution: the lighting intelligent control system for environmental monitoring image processing includes: Mode construction module: designs the mode of the lighting lamp according to the rule expansion network and obtains the currently selected mode; Data collection module: used to collect comprehensive environmental data, and splice the comprehensive environmental data to obtain a comprehensive panoramic atlas; Data aggregation processing module: used to adaptively segment the comprehensive panoramic atlas, perform feature aggregation on the segmented images using the KU clustering method, and obtain feature data sets;

[0007] Control module: Analyze the environment of the sub-area based on the feature data set, generate a feature plan, and use PID control to control the lighting according to the feature plan.

[0008] Preferably, the method for constructing the rule network includes: The rule-based network includes primary nodes and secondary nodes, which are connected by directed edges. The first-level node is simulated by the mode of the lighting lamp, and the lighting lamp mode is set to normal mode, energy-saving mode, free mode and follow mode; The secondary node is simulated by the rule name, rule description, trigger condition and processing method. The rule name is set to be the same as the mode of the lighting lamp. The rule description is used to process the control of the lighting lamp in different modes. The trigger condition is the mode selection. The processing method is the method of processing the environmental monitoring image based on the mode selection. The rule of the normal mode is that all lights are illuminated with fixed brightness and color. The energy-saving mode is to adjust the brightness and color of lights in unmanned and occupied areas according to the flow of people. The rule of the free mode is described as manual adjustment. The follow mode is to turn off the lights in unmanned areas according to the flow of people, and adjust the brightness and color of lights in occupied areas. The direction of the directed edge is from the primary node to the secondary node.

[0009] Preferably, the method for collecting comprehensive environmental data comprises: Get the complete space that needs lighting control, and divide the complete space into R_K sub-areas; The environment is imaged through the monitoring equipment installed in each sub-area, and the acquisition frequency within the cycle time is set to K_U times; The images collected from each sub-area at the same time point form an image data set, and the image data sets within the collection period form an image data set; Use one-hot encoding to convert sub-regions into numerical values, and use the numerical codes corresponding to each sub-region as labels to mark the image dataset; The image datasets of all labeled sub-regions are aggregated to form comprehensive environment data.

[0010] Preferably, the method of performing splicing processing on the comprehensive environmental data to obtain a comprehensive panoramic atlas includes: Step A1: extracting image data sets of each sub-area in the comprehensive environmental data, scaling each image of each image data group in the image data set to the same size, and converting them into gray images; Step A2: Use SIFT algorithm to detect key feature points in each image, and generate description vectors for each key feature point; Step A3: For each key feature point in any image, calculate its Euclidean distance with the key feature points in other images, search for key feature points whose Euclidean distance is less than a preset segmentation threshold, record them as corresponding feature points, and use the key feature points and corresponding feature points as correct matching point pairs; Step A4: Based on the correct matching point pairs, a direct linear transformation algorithm or a least squares method is used to calculate the homography matrix between the images, and one image is selected as a reference image. The other images are transformed into the coordinate system of the reference image through the homography matrix, and the transformed images are fused to generate the final panoramic image. Step A5: Collect the panoramic images of each sub-area to form a comprehensive panoramic image set.

[0011] Preferably, the method of using the SIFT algorithm to detect key feature points in each image and generating a description vector for each key feature point includes: Set the image parameter range to , discretized into u_l parameters at fixed intervals within the parameter range , one parameter controls a Gaussian filter, the image is represented in scale space by the Gaussian filter, the scale map after Gaussian blur at different scales is obtained, the difference of the scale map at adjacent parameters is calculated, and the DoG pyramid is obtained; Set Image , then the blurred image in the scale space representation of the image is ,and ,in, Indicated by the parameter The Gaussian filter controlled by , is the pixel coordinate on the image, represents the horizontal axis, Vertical axis.

[0012] In each layer of the DoG pyramid and between adjacent layers, the size of each pixel is compared with its surrounding pixels to obtain local extreme points, which are recorded as candidate feature points; Calculate the principal curvature and secondary curvature of each candidate feature point. , Indicates that the image is The second derivative in the direction, Indicates that the image is Second derivative in direction, secondary curvature , Represents the mixed second-order derivative of the image in the x and y directions, sets the limit threshold, and if the main curvature and minor curvature of the candidate feature point are both greater than the limit threshold, it is retained; otherwise, it is discarded; Collect all the retained candidate feature points and record them as the key feature points of the image. For each key feature point, take it as the center, select a 16x16 neighborhood window, and divide it into 4x4 4x4 sub-regions. In each 4x4 sub-region, calculate the sum of the gradient amplitudes in 8 directions. The description vector consists of the sum of the gradient amplitudes in 8 directional intervals, which is a 128-dimensional feature vector.

[0013] Preferably, the method of scaling each image in each image data group in the image data set to the same size comprises: Step B1: Set the size of the panorama , identifies the number of images in the image dataset , get the average image size as , compare the size of each image in the image data set with the average image size, and use the image with the closest size as the reference image; Step B2: In the reference image, a reference block is randomly selected using a reference frame, and based on the image selected by the reference block, the same parts as other images in the image data set are identified and circled using a similar frame; Step B3: the image is enlarged or reduced in the same proportion as the similar frame until the size of the similar frame is the same as the reference frame. At this time, the enlarged image has the same size as the reference image. Step B4: Repeat steps B2 and B3 until all images in the image data set have the same size as the reference image.

[0014] Preferably, the method for adaptively segmenting the comprehensive panoramic atlas comprises: Step C1: For each panorama in the comprehensive panorama set, extract the maximum pixel value and minimum pixel value , set the interval threshold ,in, and represents the proportion coefficient, Represents the set of all pixels in the image. Indicates the position in the image The pixel value at ; Step C2: randomly select a pixel value at a pixel point as a reference point, take the reference point as the center of the circle, calculate the pixel difference between it and the adjacent pixel points, include the pixel points whose pixel difference is less than or equal to the interval threshold into the cutting range, and continue to calculate until a pixel point whose pixel difference is greater than the interval threshold appears, and then stop, collect all the pixel points included in the cutting range as a small block; Step C3: Repeat step C2 until all pixels in the panorama are completely segmented, and then divide the panorama into Q_W small blocks to form a block set; Step C4: Repeat step C3 until all panoramas in the comprehensive panorama set are completely segmented.

[0015] Preferably, the method of performing feature aggregation on the segmented image using the KU clustering method to obtain a feature data set includes: For each panorama in the comprehensive panorama set, a set of tiles is formed by segmenting each panorama; Step D1: preset the number of clusters K_B, randomly select K_B small blocks, and calculate the pixel average value of each small block as the cluster center; Step D2: Calculate the distance between the pixel average value of each small block and the cluster center, and assign the small block to the cluster center with the nearest distance, dividing the small blocks into K_B groups; Step D3: Calculate the pixel average of all small blocks in each group, update it as the new cluster center, and repeat step D2 until all cluster centers no longer change; Step D4: Mark the small tiles in each group as a cluster, and collect all the clusters as feature data sets; Step D5: Repeat steps D1 to D4 to obtain a feature dataset of all panoramas in the comprehensive panorama collection.

[0016] Preferably, the method of analyzing the environment of the sub-region based on the feature data set to generate a feature solution includes: Build a machine learning model based on a deep neural network design. Set the model input as the feature data set and lighting layout of all sub-areas in the complete space, and the output as the feature solution. Use the sample set to train the model until the number of iterations is reached, and then obtain a trained model. Input the feature data set and lighting layout of all sub-areas in the current complete space into the trained model to obtain the current feature solution; The sample set consists of feature data sets and lighting layouts of all sub-areas in the complete space, as well as the corresponding feature schemes; The characteristic scheme is presented by expanding the lighting in each sub-area horizontally and expanding it vertically by brightness and color to form a two-dimensional table, and then forming R_K two-dimensional tables to constitute the characteristic scheme.

[0017] Preferably, the method for regulating the lighting lamp using PID control according to the characteristic scheme includes: The brightness of the lighting lamp at various powers is extracted, and the brightness and color of each sub-area are adjusted using PID control according to the characteristic scheme.

[0018] Compared with the prior art, the present invention has the following beneficial effects: By designing four modes (normal, energy-saving, free, and follow), combined with the node design and trigger conditions of rule-based network expansion, the system can flexibly adapt to different scene requirements. In energy-saving mode and follow mode, the brightness and color are automatically adjusted based on dynamic detection of personnel flow, saving energy while improving user experience.

[0019] By dividing sub-areas and collecting images of multiple sub-areas with monitoring equipment, and combining image stitching technology to generate a comprehensive panoramic atlas, the environmental status of the entire space can be fully perceived. Sub-areas are marked using one-hot coding to achieve regionalized data management. The coverage and comprehensiveness of environmental information collection are improved, providing data support for subsequent intelligent control. The problem of single information source and insufficient spatial coverage in traditional systems is solved.

[0020] The SIFT algorithm is used to extract key feature points. Combined with image stitching technology, it can efficiently and accurately extract key features in environmental images and generate high-quality panoramic atlases. Image segmentation and KU clustering methods are used to refine the feature information in the panoramic image and extract accurate feature data sets. The ability to perceive changes in environmental conditions is improved, especially the data processing capabilities for complex spaces (such as multi-regional and multi-dynamic scenes). It provides a more accurate decision-making basis for the regulation of lighting.

[0021] By building a deep neural network model, a real-time feature solution is generated based on the environmental feature dataset and lighting layout design to achieve intelligent control of lighting. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of the structure of an intelligent lighting control system for environmental monitoring image processing proposed by the present invention; Figure 2 It is a schematic diagram of the method of the lighting intelligent control system applied to environmental monitoring image processing in the present invention. DETAILED DESCRIPTION

[0023] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0024] Embodiment 1 Reference Figure 1 and Figure 2 Embodiment 1 further illustrates the intelligent lighting control system for environment monitoring image processing proposed by the present invention.

[0025] The present invention is applicable to the following scenarios: Smart building lighting: intelligent lighting management for large buildings such as office buildings, shopping malls, and hotels.

[0026] Public facilities: Dynamic lighting adjustment in subway stations, airports, parking lots, etc.

[0027] Home smart lighting: Combine free mode and follow mode to achieve personalized and energy-saving home lighting.

[0028] Industrial scenarios: Regional lighting management in factory workshops, warehouses, and other scenarios.

[0029] Through the implementation of the above technical solutions, the system can achieve efficient, intelligent and energy-saving lighting solutions in a variety of scenarios, including: Mode construction module: designs the mode of the lighting lamp according to the rule expansion network and obtains the currently selected mode; The construction method of rule extension network includes: The rule-based network includes primary nodes and secondary nodes, which are connected by directed edges. The first-level node is simulated by the mode of the lighting lamp, and the lighting lamp mode is set to normal mode, energy-saving mode, free mode and follow mode; The secondary node is simulated by the rule name, rule description, trigger condition and processing method. The rule name is set to be the same as the mode of the lighting lamp. The rule description is used to process the control of the lighting lamp in different modes. The trigger condition is the mode selection. The processing method is the method of processing the environmental monitoring image based on the mode selection. The rule of the normal mode is that all lights are illuminated with fixed brightness and color. The energy-saving mode is to adjust the brightness and color of lights in unmanned and occupied areas according to the flow of people. The rule of the free mode is described as manual adjustment. The follow mode is to turn off the lights in unmanned areas according to the flow of people, and adjust the brightness and color of lights in occupied areas. The direction of the directed edge is from the primary node to the secondary node; For example, if the energy-saving mode is selected, the first-level node in the rule network will be started according to the selected mode, and the second-level node will be started through the directed edge. In the second-level node, the rule name is energy-saving mode, and the rule description is to distinguish and adjust the brightness and color of the lights in unmanned and occupied areas according to the flow of people. That is, for the entire environment, or in the house, the lights in places where no human activities are identified are turned off, and the brightness and color of the lights in the environment with human activities are adjusted. The color is the light color, and the processing method is to selectively process the environmental monitoring image according to the selected energy-saving mode.

[0030] The rules of the free mode are described as human adjustment, specifically, people can freely adjust the light and brightness according to their own behavior, which is not limited to the brightness and color of the lighting in the other three modes.

[0031] Through energy-saving mode and follow mode, the system can dynamically sense the flow of people and regional status, reduce the brightness or turn off the lights in unmanned areas, and reduce unnecessary energy waste. In manned areas, the brightness and color are adjusted according to the environmental status to ensure the lighting effect while saving energy to the greatest extent. The overall energy consumption is significantly reduced and the energy utilization efficiency of the system is improved. In large-scale scenarios (such as shopping malls, office buildings, and communities), it can effectively reduce operating costs.

[0032] Data collection module: used to collect comprehensive environmental data, and splice the comprehensive environmental data to obtain a comprehensive panoramic atlas; Methods for collecting comprehensive environmental data include: Obtain a complete space that requires lighting control, and divide the complete space into R_K sub-areas; for example, the complete space may be an entire house, and the entire house may be divided into multiple sub-areas such as a work area, a rest area, a training area, and a dining area according to functions or other customized division requirements.

[0033] The environment is imaged through the monitoring equipment installed in each sub-area, and the acquisition frequency within the cycle time is set to K_U times; The images collected from each sub-area at the same time point form an image data set, and the image data sets within the collection period form an image data set; Use one-hot encoding to convert sub-regions into numerical values, and use the numerical codes corresponding to each sub-region as labels to mark the image dataset; The image datasets of all labeled sub-regions are aggregated to form comprehensive environment data.

[0034] Assume that there are 4 monitoring devices (such as cameras) installed in the work area, and data is collected at a frequency of K_U=5 times within a cycle time of 30 seconds. In the present invention, it is assumed that all monitoring devices collect data once at the same time point, that is, the timestamps of the images are the same. In this way, 5 image data groups are obtained. The 5 image data groups within 30 seconds are aggregated to form comprehensive environmental data. Therefore, when real-time monitoring of the environment is performed, it is only necessary to continuously collect images in a cycle time to continuously monitor the entire house.

[0035] The method of stitching the comprehensive environmental data to obtain a comprehensive panoramic atlas includes: Step A1: extracting image data sets of each sub-area in the comprehensive environmental data, scaling each image of each image data group in the image data set to the same size, and converting them into gray images; If there are obvious color differences between images, color correction can be performed first, such as using histogram equalization or color transfer algorithm, and then the color image is converted into a grayscale image, that is, the color image is separated into three channels: red, green, and blue. The grayscale value of each pixel is calculated according to the selected conversion method (such as weighted average method), and the calculated grayscale values ​​are combined into a new grayscale image.

[0036] Step A2: Use SIFT algorithm to detect key feature points in each image, and generate description vectors for each key feature point; Step A3: For each key feature point in any image, calculate its Euclidean distance with the key feature points in other images, search for key feature points whose Euclidean distance is less than a preset segmentation threshold, record them as corresponding feature points, and use the key feature points and corresponding feature points as correct matching point pairs; The preset segmentation threshold can be set based on experimental data analysis or historical data experience analysis.

[0037] Step A4: Based on the correct matching point pairs, a direct linear transformation algorithm or a least square method is used to calculate the homography matrix between the images. One image is selected as a reference image. The other images are transformed into the coordinate system of the reference image through the homography matrix. Bilinear interpolation or Lanczos interpolation can be used to resample the transformed image to fill the pixel vacancies caused by the transformation. The transformed images are fused to generate the final panoramic image. Image fusion can use weighted average fusion or feathering. Weighted average fusion is to weighted average the pixel values ​​of the overlapping area to eliminate the stitching seams. Commonly used weight functions include linear weights, multi-band fusion weights, etc. Feathering is to make a smooth transition at the edge of the overlapping area to eliminate obvious stitching marks.

[0038] Step A5: Collect the panoramic images of each sub-area to form a comprehensive panoramic image set.

[0039] The method of using SIFT algorithm to detect key feature points in each image and generating description vectors for each key feature point includes: Set the image parameter range to , discretized into u_l parameters at fixed intervals within the parameter range , one parameter controls a Gaussian filter, the image is represented in scale space by the Gaussian filter, the scale map after Gaussian blur at different scales is obtained, the difference of the scale map at adjacent parameters is calculated, and the DoG pyramid is obtained; Set Image , then the blurred image in the scale space representation of the image is ,and ,in, Indicated by the parameter The Gaussian filter controlled by , is the pixel coordinate on the image, represents the horizontal axis, Vertical axis.

[0040] The image is rescaled by convolving the original image with multiple Gaussian filters of different parameters (scales), where the scale of the Gaussian filter is controlled by the parameter, The larger the value of , the wider the filter.

[0041] In each layer and adjacent layers of the DoG pyramid, the size of each pixel is compared with the surrounding pixels (including pixels of the same scale and adjacent scales), and local extreme points (maximum and minimum) are obtained and recorded as candidate feature points; The local maximum and local minimum of each pixel point requires checking whether the determinant of the Hessian matrix of each pixel point is positive or negative. If the determinant is positive, the point is a local maximum; if the determinant is negative, the point is a local minimum.

[0042] Since there may be some unstable points among the detected candidate feature points, these points need to be filtered. Specifically, the principal curvature and minor curvature of each point need to be calculated, and unstable points need to be filtered out based on these values.

[0043] Calculate the principal curvature and secondary curvature of each candidate feature point. , the principal curvature is the average of the second-order derivatives in the x and y directions, which reflects the average rate of change of the image in two orthogonal directions. Indicates that the image is The second-order derivative in the direction, and The rate of change in direction is related to Indicates that the image is The second-order derivative in the direction, and The rate of change in direction is related to the secondary curvature. , the secondary curvature is half the square root of the square of the difference between the second-order derivatives in the x and y directions and the square of the mixed second-order derivative, which reflects the difference in the rate of change of the image in two orthogonal directions. It represents the mixed second-order derivative of the image in the x and y directions, which is related to the interaction of the rate of change in the x and y directions. Set the limit threshold. If the main curvature and minor curvature of the candidate feature point are both greater than the limit threshold, it is retained; otherwise, it is discarded; Collect all the retained candidate feature points and record them as the key feature points of the image. For each key feature point, take it as the center, select a 16x16 neighborhood window, and divide it into 4x4 4x4 sub-regions. In each 4x4 sub-region, calculate the sum of the gradient amplitudes in 8 directions. The description vector consists of the sum of the gradient amplitudes in 8 directional intervals, that is, a 4x4x8=128-dimensional feature vector.

[0044] For each point, the description vector represents the gradient information of the area around the point. Specifically, it is necessary to calculate the gradient magnitude and direction of the area around the key point. The gradient magnitude is expressed as , the gradient direction is expressed as ,in, and Respectively Partial derivatives in the x and y directions. It is divided into 8 direction intervals, each of which has a range of 45°. For each interval, the sum of the gradient amplitudes within the interval is calculated, that is, ,in, , the description vector is composed of the sum of the gradient amplitudes in these 8 direction intervals, which is the description vector generated by the SIFT algorithm.

[0045] This solution uses the SIFT algorithm to extract image feature points and generate a 128-dimensional description vector, which effectively enhances the anti-interference ability of feature point matching. Using image stitching technology, the sub-region images are accurately fused through the homography matrix to generate a high-quality panoramic atlas to ensure the accuracy of data processing. The adaptive segmentation method is used to flexibly segment the panorama, which can adapt to different image features and avoid the errors that may be caused by fixed segmentation methods. The system still has high processing accuracy and robustness in complex environments (such as light changes and dynamic scenes). Accurate panoramas and feature data provide a reliable basis for subsequent analysis and regulation.

[0046] The method for scaling each image of each image data group in the image data set to the same size comprises: Step B1: Set the size of the panorama , which can be pre-set based on experience or actual requirements, or can be pre-set based on historical or experimental data analysis to identify the number of images in the image data set , get the average image size as , compare the size of each image in the image data set with the average image size, and use the image with the closest size as the reference image; Step B2: In the reference image, a reference block is randomly selected using a reference frame (circular or square frame), and based on the image selected by the reference block, the same parts as other images in the image data set are identified and circled using a similar frame; the similar frame has the same shape as the reference frame, but can be of different sizes; Step B3: the image is enlarged or reduced in the same proportion as the similar frame until the size of the similar frame is the same as the reference frame. At this time, the enlarged image has the same size as the reference image. Step B4: Repeat steps B2 and B3 until all images in the image data set have the same size as the reference image.

[0047] For example, assuming that there are 5 images in the image data set, one image is selected as the reference image. If a 1×1 square frame is used as the reference frame, the reference frame randomly circles a piece of image content on the reference image as the reference block. Based on the content in the reference frame, the places in other images that are the same as the content in the reference frame are identified and then circled. At this time, the frame formed by the circle is the similarity frame. When the similarity frame is scaled to the same size as the reference frame, the image in the similarity frame will be scaled together, thereby making the content size on the two images the same.

[0048] Data aggregation processing module: used to adaptively segment the comprehensive panoramic atlas, perform feature aggregation on the segmented images using the KU clustering method, and obtain feature data sets; Methods for adaptive segmentation of comprehensive panorama atlases include: Step C1: For each panorama in the comprehensive panorama set, extract the maximum pixel value and minimum pixel value , set the interval threshold ,in, and represents the proportion coefficient, Represents the set of all pixels in the image. Indicates the position in the image The pixel value at ; Step C2: randomly select a pixel value at a pixel point as a reference point, take the reference point as the center of the circle, calculate the pixel difference between it and the adjacent pixel points, include the pixel points whose pixel difference is less than or equal to the interval threshold into the cutting range, and continue to calculate until a pixel point whose pixel difference is greater than the interval threshold appears, and then stop, collect all the pixel points included in the cutting range as a small block; Step C3: Repeat step C2 until all pixels in the panorama are completely segmented, and then divide the panorama into Q_W small blocks to form a block set; Step C4: Repeat step C3 until all panoramas in the comprehensive panorama set are completely segmented.

[0049] The KU clustering method is used to perform feature aggregation on the segmented image. The method for obtaining the feature data set includes: For each panorama in the comprehensive panorama set, a set of tiles is formed by segmenting each panorama; Step D1: preset the number of clusters K_B, randomly select K_B small blocks, and calculate the pixel average value of each small block as the cluster center; Step D2: Calculate the distance between the pixel average value of each small block and the cluster center, and assign the small block to the cluster center with the nearest distance, dividing the small blocks into K_B groups; Step D3: Calculate the pixel average of all small blocks in each group, update it as the new cluster center, and repeat step D2 until all cluster centers no longer change; Step D4: Mark the small tiles in each group as a cluster, and collect all the clusters as feature data sets; Step D5: Repeat steps D1 to D4 to obtain a feature dataset of all panoramas in the comprehensive panorama collection.

[0050] Control module: Analyze the environment of the sub-area based on the feature data set, generate a feature solution, and use PID control to control the lighting according to the feature solution; The method of analyzing the environment of the sub-region based on the feature data set and generating the feature solution includes: Build a machine learning model based on a deep neural network design. Set the model input as the feature data set and lighting layout of all sub-areas in the complete space, and the output as the feature solution. Use the sample set to train the model until the number of iterations is reached, and then obtain a trained model. Input the feature data set and lighting layout of all sub-areas in the current complete space into the trained model to obtain the current feature solution; The sample set consists of feature data sets and lighting layouts of all sub-areas in the complete space, as well as the corresponding feature schemes; The characteristic scheme is presented by expanding the lighting lamps of each sub-area horizontally and expanding them vertically by brightness and color to form a two-dimensional table, and then forming R_K two-dimensional tables to form the characteristic scheme; Lighting layout refers to the distribution of lights in each sub-area, and each light can also be manually labeled. The brightness and color of the lights in the sub-area in normal mode, energy-saving mode and follow-up mode in unmanned and manned environments can also be pre-set to form a standard specification table; Methods for regulating lighting lamps using PID control according to characteristic schemes include: The brightness of the lighting lamp at various powers is extracted, and the brightness and color of each sub-area are adjusted using PID control according to the characteristic scheme.

[0051] By introducing a deep neural network model, the system can generate dynamic control schemes based on multi-dimensional feature data sets, improving the intelligent level of control. The PID control algorithm is used to adjust the brightness and color of each sub-area lighting in real time, which can effectively reduce the adjustment delay and improve the control accuracy. The system's response speed to environmental changes is improved, and it can quickly adapt to dynamically changing scenes. It realizes regionalized refined control, avoids the "one-size-fits-all" control method of traditional systems, and further improves the flexibility of the system.

[0052] For example, first the staff selects the mode of the house, and then circulates in the rule network according to the selected mode to intelligently control the lighting of the entire house.

[0053] When the normal mode is selected, the brightness and color of the lights in each sub-area under the normal mode are preset. The preset here can be directly set by the program or it can be said to be the default. For example, in the normal mode, the brightness and color of the lights in the rest area are set to 30 and orange respectively, then the lights will be directly adjusted to this value according to the preset.

[0054] When the energy-saving mode is selected, the sub-area where the person is located is first identified based on the image of the monitoring device, and the sub-area where the person is located is distinguished from other sub-areas. For example, in the energy-saving mode, the brightness and color of the lighting are set to 50 and bright white when there are people in the rest area, and 10 and orange when there are no people. The lighting at the corresponding position is adjusted and controlled according to the real-time brightness.

[0055] Generally speaking, a sub-area will be equipped with multiple lighting fixtures, each of which will control the brightness and color at a position. Therefore, by identifying image blocks with the same pixels, the positions with the same brightness are grouped together, and then the brightness and color are adjusted by the lighting fixtures at that position, thereby unifying the color and brightness in the space.

[0056] Embodiment 2 Reference Figure 1 and Figure 2 , Example 2 further illustrates an intelligent lighting control system for environmental monitoring image processing proposed by the present invention.

[0057] A lighting intelligent control method for environmental monitoring image processing comprises the following steps: Step S1: Expand the network according to the rules to obtain the currently selected mode of the lighting lamp; Step S2: Collect comprehensive environmental data and perform splicing processing to obtain a comprehensive panoramic atlas; Step S3: adaptively segmenting each panoramic image in the comprehensive panoramic image set into a plurality of small image blocks to form an image block set; Step S4: Use the KU clustering method to perform feature aggregation on each small tile in each tile set, and aggregate small tiles with similar pixel values ​​together to form a feature data set; Step S5: Analyze the environment of the sub-area based on the feature data set and generate a feature solution; Step S6: Use PID control to regulate the lighting lamp according to the characteristic scheme.

[0058] An intelligent control system for lighting lamps using environmental monitoring image processing, and an intelligent control method for lighting lamps using environmental monitoring image processing include: Mode construction module: designs the mode of the lighting lamp according to the rule expansion network and obtains the currently selected mode; Data collection module: used to collect comprehensive environmental data, and splice the comprehensive environmental data to obtain a comprehensive panoramic atlas; Data aggregation processing module: used to adaptively segment the comprehensive panoramic atlas, perform feature aggregation on the segmented images using the KU clustering method, and obtain feature data sets; Control module: Analyze the environment of the sub-area based on the feature data set, generate a feature solution, and use PID control to control the lighting according to the feature solution; The modules are connected to each other via wired and / or wireless means.

[0059] The technical advantages of the present invention are as follows Multi-mode flexible control: supports four modes: normal, energy-saving, free, and follow, to meet the needs of multiple scenarios.

[0060] Comprehensive information collection: Through image monitoring equipment and panoramic stitching technology, complete space environmental perception is achieved.

[0061] Refined image processing: SIFT algorithm is used to extract feature points, combined with image stitching and adaptive segmentation to ensure the accuracy of data processing.

[0062] Intelligent control: Generate feature solutions based on deep neural networks and combine them with PID control to achieve precise dynamic adjustment.

[0063] Energy saving: In energy-saving mode and follow mode, energy consumption is significantly reduced and energy utilization efficiency is improved.

[0064] High robustness and flexibility: The system has strong adaptability to complex scene changes, flexible regulation and rapid response.

[0065] Modularity and scalability: facilitates system maintenance and optimization, and is suitable for scenarios of different scales and requirements.

[0066] In addition, according to the implementation of the present application, the process described in the attached figure of a lighting intelligent control system for environmental monitoring image processing can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided by the present application. Of course, the architecture shown in the attached figure of a lighting intelligent control system for environmental monitoring image processing is only exemplary, and when implementing different devices, adaptive selection or adjustment can be performed according to actual needs.

[0067] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0068] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent lighting control system for environmental monitoring image processing, characterized in that: The lighting intelligent control system for environmental monitoring image processing includes: Mode construction module: designs the mode of the lighting lamp according to the rule expansion network and obtains the currently selected mode; Data collection module: used to collect comprehensive environmental data, and splice the comprehensive environmental data to obtain a comprehensive panoramic atlas; Data aggregation processing module: used to adaptively segment the comprehensive panoramic atlas, perform feature aggregation on the segmented images using the KU clustering method, and obtain feature data sets; Control module: Analyze the environment of the sub-area based on the feature data set, generate a feature plan, and use PID control to control the lighting according to the feature plan.

2. The lighting intelligent control system based on environmental monitoring image processing according to claim 1 is characterized in that: The method for constructing the rule network includes: The rule-based network includes primary nodes and secondary nodes, which are connected by directed edges. The first-level node is simulated by the mode of the lighting lamp, and the lighting lamp mode is set to normal mode, energy-saving mode, free mode and follow mode; The secondary node is simulated by the rule name, rule description, trigger condition and processing method. The rule name is set to be the same as the mode of the lighting lamp. The rule description is used to process the control of the lighting lamp in different modes. The trigger condition is the mode selection. The processing method is the method of processing the environmental monitoring image based on the mode selection. The rule of the normal mode is that all lights are illuminated with fixed brightness and color. The energy-saving mode is to adjust the brightness and color of lights in unmanned and occupied areas according to the flow of people. The rule of the free mode is described as manual adjustment. The follow mode is to turn off the lights in unmanned areas according to the flow of people, and adjust the brightness and color of lights in occupied areas. The direction of the directed edge is from the primary node to the secondary node.

3. The lighting intelligent control system based on environmental monitoring image processing according to claim 2 is characterized in that: The method for collecting comprehensive environmental data comprises: Get the complete space that needs lighting control, and divide the complete space into R_K sub-areas; The environment is imaged through the monitoring equipment installed in each sub-area, and the acquisition frequency within the cycle time is set to K_U times; The images collected from each sub-area at the same time point form an image data set, and the image data sets within the collection period form an image data set; Use one-hot encoding to convert sub-regions into numerical values, and use the numerical codes corresponding to each sub-region as labels to mark the image dataset; The image datasets of all labeled sub-regions are aggregated to form comprehensive environment data.

4. The lighting intelligent control system based on environmental monitoring image processing according to claim 3 is characterized in that: The method of performing splicing processing on the comprehensive environmental data to obtain a comprehensive panoramic atlas comprises: Step A1: extracting image data sets of each sub-area in the comprehensive environmental data, scaling each image of each image data group in the image data set to the same size, and converting them into gray images; Step A2: Use SIFT algorithm to detect key feature points in each image, and generate description vectors for each key feature point; Step A3: For each key feature point in any image, calculate its Euclidean distance with the key feature points in other images, search for key feature points whose Euclidean distance is less than a preset segmentation threshold, record them as corresponding feature points, and use the key feature points and corresponding feature points as correct matching point pairs; Step A4: Based on the correct matching point pairs, a direct linear transformation algorithm or a least squares method is used to calculate the homography matrix between the images, and one image is selected as a reference image. The other images are transformed into the coordinate system of the reference image through the homography matrix, and the transformed images are fused to generate the final panoramic image. Step A5: Collect the panoramic images of each sub-area to form a comprehensive panoramic image set.

5. The lighting intelligent control system based on environmental monitoring image processing according to claim 4 is characterized in that: The method of using the SIFT algorithm to detect key feature points in each image and generating a description vector for each key feature point includes: Set the image parameter range to , discretized into u_l parameters at fixed intervals within the parameter range , one parameter controls a Gaussian filter, the image is represented in scale space by the Gaussian filter, the scale map after Gaussian blur at different scales is obtained, the difference of the scale map at adjacent parameters is calculated, and the DoG pyramid is obtained; Set Image , then the blurred image in the scale space representation of the image is ,and ,in, Indicated by the parameter The Gaussian filter controlled by , is the pixel coordinate on the image, represents the horizontal axis, ordinate; In each layer of the DoG pyramid and between adjacent layers, the size of each pixel is compared with its surrounding pixels to obtain local extreme points, which are recorded as candidate feature points; Calculate the principal curvature and secondary curvature of each candidate feature point. , Indicates that the image is The second derivative in the direction, Indicates that the image is Second derivative in direction, secondary curvature , Represents the mixed second-order derivative of the image in the x and y directions; Set a limit threshold. If the main curvature and minor curvature of the candidate feature point are both greater than the limit threshold, it will be retained, otherwise, it will be discarded; Collect all the retained candidate feature points and record them as the key feature points of the image. For each key feature point, take it as the center, select a 16x16 neighborhood window, and divide it into 4x4 4x4 sub-regions. In each 4x4 sub-region, calculate the sum of the gradient amplitudes in 8 directions. The description vector consists of the sum of the gradient amplitudes in 8 directional intervals, which is a 128-dimensional feature vector.

6. The lighting intelligent control system based on environmental monitoring image processing according to claim 5 is characterized in that: The method for scaling each image of each image data group in the image data set to the same size comprises: Step B1: Set the size of the panorama , identifies the number of images in the image dataset , get the average image size as , compare the size of each image in the image data set with the average image size, and use the image with the closest size as the reference image; Step B2: In the reference image, a reference block is randomly selected using a reference frame, and based on the image selected by the reference block, the same parts as other images in the image data set are identified and circled using a similar frame; Step B3: the image is enlarged or reduced in the same proportion as the similar frame until the size of the similar frame is the same as the reference frame. At this time, the enlarged image has the same size as the reference image. Step B4: Repeat steps B2 and B3 until all images in the image data set have the same size as the reference image.

7. The lighting intelligent control system based on environmental monitoring image processing according to claim 6 is characterized in that: The method for adaptively segmenting a comprehensive panoramic atlas comprises: Step C1: For each panorama in the comprehensive panorama set, extract the maximum pixel value and minimum pixel value , set the interval threshold ,in, and represents the proportion coefficient, Represents the set of all pixels in the image. Indicates the position in the image The pixel value at ; Step C2: randomly select a pixel value at a pixel point as a reference point, take the reference point as the center of the circle, calculate the pixel difference between it and the adjacent pixel points, include the pixel points whose pixel difference is less than or equal to the interval threshold into the cutting range, and continue to calculate until a pixel point whose pixel difference is greater than the interval threshold appears, and then stop, collect all the pixel points included in the cutting range as a small block; Step C3: Repeat step C2 until all pixels in the panorama are completely segmented, and then divide the panorama into Q_W small blocks to form a block set; Step C4: Repeat step C3 until all panoramas in the comprehensive panorama set are completely segmented.

8. The lighting intelligent control system based on environmental monitoring image processing according to claim 7 is characterized in that: The method of using the KU clustering method to perform feature aggregation on the segmented image to obtain a feature data set includes: For each panorama in the comprehensive panorama set, a set of tiles is formed by segmenting each panorama; Step D1: preset the number of clusters K_B, randomly select K_B small blocks, and calculate the pixel average value of each small block as the cluster center; Step D2: Calculate the distance between the pixel average value of each small block and the cluster center, and assign the small block to the cluster center with the nearest distance, dividing the small blocks into K_B groups; Step D3: Calculate the pixel average of all small blocks in each group, update it as the new cluster center, and repeat step D2 until all cluster centers no longer change; Step D4: Mark the small tiles in each group as a cluster, and collect all the clusters as feature data sets; Step D5: Repeat steps D1 to D4 to obtain a feature dataset of all panoramas in the comprehensive panorama collection.

9. The lighting intelligent control system based on environmental monitoring image processing according to claim 8 is characterized in that: The method of analyzing the environment of the sub-region based on the feature data set to generate a feature solution includes: Build a machine learning model based on a deep neural network design. Set the model input as the feature data set and lighting layout of all sub-areas in the complete space, and the output as the feature solution. Use the sample set to train the model until the number of iterations is reached, and then obtain a trained model. Input the feature data set and lighting layout of all sub-areas in the current complete space into the trained model to obtain the current feature solution; The sample set consists of feature data sets and lighting layouts of all sub-areas in the complete space, as well as the corresponding feature schemes; The characteristic scheme is presented by expanding the lighting in each sub-area horizontally and expanding it vertically by brightness and color to form a two-dimensional table, and then forming R_K two-dimensional tables to constitute the characteristic scheme.

10. The lighting intelligent control system based on environmental monitoring image processing according to claim 9 is characterized in that: The method for regulating the lighting lamp using PID control according to the characteristic scheme includes: The brightness of the lighting lamp at various powers is extracted, and the brightness and color of each sub-area are adjusted using PID control according to the characteristic scheme.

Images