Intelligent lighting communication control method and system

Abstract

The invention discloses an intelligent illumination communication control method and system, and belongs to the field of intelligent illumination control, and the method comprises the steps: collecting an area image of illumination covered by an illumination lamp source array; transmitting the regional image, and performing portrait recognition and division to obtain a plurality of user regions; an image of a user area is collected, multiple marked area images are obtained, user type identification is carried out, multiple user type information is obtained, user feature identification is carried out, and multiple user feature identification results are obtained, so that an illumination control scheme of an illumination lamp source array is optimized, an optimal illumination control scheme is obtained, and the illumination lamp source array is controlled. The technical problem that the illumination effect of the illumination lamp source array is poor due to the lack of real-time capture and recognition of user features in the prior art is solved, and the technical effects that the illumination control scheme is optimized by recognizing the feature information of different types of users, illumination lamp source control is performed, and the illumination effect is improved are achieved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent lighting control, and in particular to an intelligent lighting communication control method and system. Background Art

[0002] With the increasing maturity of LED lighting technology, LED-based intelligent lighting control systems have been widely used in office, commercial, public and other scenarios, and have the technical advantages of high efficiency, energy saving, humanization and comfort. Existing LED intelligent lighting systems mostly adopt fixed lighting control strategies, lacking flexible adaptability to complex application environments with changing light requirements and diverse users. Especially in indoor environments with dense lamps, complex application scenarios with multiple categories of users have put forward higher requirements for the intelligent recognition and precise control capabilities of lighting control. When the user category, position and posture change, the existing technology cannot perceive the changes in user characteristics in real time and make corresponding adjustments, which leads to insufficient illumination or glare in some areas, resulting in poor lighting effects of the lighting source array. Summary of the invention

[0003] The present application provides an intelligent lighting communication control method and system, aiming to solve the technical problem in the prior art of lacking real-time capture and recognition of user characteristics, resulting in poor lighting effects of lighting source arrays.

[0004] In view of the above problems, the present application provides an intelligent lighting communication control method and system.

[0005] The first aspect disclosed in the present application provides an intelligent lighting communication control method, the method comprising: collecting images of a target area covered by an array of lighting sources to obtain an area image; transmitting the area image through wireless communication, and performing portrait recognition and division to obtain multiple user areas; collecting images of multiple user areas to obtain multiple marked area images, performing user type recognition on the multiple marked area images to obtain multiple user type information; based on the multiple user type information, mapping and selecting a matching multiple feature recognition channels, performing user feature recognition on the multiple marked area images to obtain multiple user feature recognition results, wherein the user feature recognition results include user behavior and user line of sight direction; according to the multiple user feature recognition results, optimizing the lighting control scheme of the lighting source array to obtain an optimal lighting control scheme; according to the optimal lighting control scheme, controlling the lighting source array through wireless communication.

[0006] Another aspect disclosed in the present application provides an intelligent lighting communication control system, which includes: a regional image acquisition module, which is used to acquire images of a target area covered by a lighting source array to obtain a regional image; a regional image division module, which is used to transmit the regional image through wireless communication, and perform portrait recognition and division to obtain multiple user areas; a user type recognition module, which is used to acquire images of multiple user areas to obtain multiple marked area images, perform user type recognition on the multiple marked area images, and obtain multiple user type information; a user feature recognition module, which is used to map and select multiple matching feature recognition channels based on multiple user type information, perform user feature recognition on the multiple marked area images, and obtain multiple user feature recognition results, wherein the user feature recognition results include user behavior and user line of sight direction; a control scheme optimization module, which is used to optimize the lighting control scheme of the lighting source array according to the multiple user feature recognition results to obtain an optimal lighting control scheme; and a lighting source control module, which is used to control the lighting source array through wireless communication according to the optimal lighting control scheme.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] By collecting images of the area covered by the lighting source array, the regional image is obtained to obtain visual information of the entire area to be illuminated and controlled; the regional image is transmitted in real time through the wireless network to achieve wireless high-speed transmission of remote image data, which is conducive to subsequent image analysis and processing; and the regional image is subjected to portrait recognition and regional segmentation to obtain multiple independent user areas to achieve the distinction and recognition of different user spaces; the image collection of each user area is continued to obtain the respective marked area image to capture the image information of each area; the identification of users of different categories is performed to obtain the user type data of each area for matching the feature recognition channel; according to the user type information, the appropriate An appropriate feature recognition channel is used to perform behavior recognition and line of sight direction judgment on the marked area image of each user area, and obtain user behavior characteristics and line of sight direction characteristic information; according to the user feature recognition results of all user areas, the optimal lighting control scheme is determined through control scheme optimization, so as to accurately control the lighting source array, and complete the technical solution of intelligent adaptive lighting control of the entire area, which solves the technical problem of lack of real-time capture and recognition of user characteristics in the prior art, resulting in poor lighting effect of the lighting source array, and achieves the technical effect of identifying the characteristic information of different categories of users, optimizing the lighting control scheme, controlling the lighting source, and improving the lighting effect.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flow chart of an intelligent lighting communication control method is provided for an embodiment of the present application;

[0011] Figure 2 A schematic diagram of a flow chart for obtaining multiple user type information in an intelligent lighting communication control method is provided for an embodiment of the present application;

[0012] Figure 3 A structural schematic diagram of an intelligent lighting communication control system is provided for an embodiment of the present application.

[0013] Explanation of reference numerals: regional image acquisition module 11 , regional image division module 12 , user type recognition module 13 , user feature recognition module 14 , control scheme optimization module 15 , lighting source control module 16 . DETAILED DESCRIPTION

[0014] The overall idea of ​​the technical solution provided by this application is as follows:

[0015] The embodiments of the present application provide an intelligent lighting communication control method and system, which realizes accurate capture and intelligent analysis of user characteristics of multiple user areas through technical means such as wireless network image acquisition, user feature recognition and closed-loop control optimization, and generates an optimal lighting control solution on this basis, so as to effectively solve the problem that existing lighting cannot flexibly adapt to multiple users in complex scenes.

[0016] Specifically, first, an image of the target area covered by the lighting source array is acquired to obtain a regional image. Then, wireless communication is used to achieve low-latency remote transmission of the regional image as input for intelligent analysis and control. Next, the regional image is processed by portrait recognition and division to obtain multiple user areas. Further, the regional images of each user are obtained according to the multiple user areas, and multiple marked regional images are obtained. Next, different user type information is identified, and targeted feature recognition channels are matched and selected to analyze key information such as user behavior and line of sight, and multiple user feature recognition results are obtained. Subsequently, based on the multiple user feature recognition results, the lighting control scheme is optimized to obtain the optimal lighting control scheme, and the lighting source array is precisely adjusted through wireless communication to complete the intelligent and rapid response to the changing environment, realizing the real-time capture of user feature changes, the organic combination of intelligent analysis and active control, so as to achieve the purpose of flexible adaptation to human needs and improve the lighting effect.

[0017] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0018] Embodiment 1

[0019] like Figure 1 As shown, an embodiment of the present application provides an intelligent lighting communication control method, the method comprising:

[0020] Collect images within the target area covered by the illumination light source array to obtain an area image;

[0021] In the embodiment of the present application, a camera installed on the top or side of the target area is used to collect images of the target area covered by the lighting source array in a top-down or side-view manner. The target area is an indoor area, such as an office, an airport lobby, a shopping mall, etc. The collected regional image includes the entire target area covered by the lighting source array.

[0022] Optionally, in order to prevent the target area from being too large and causing the camera to be unable to perform panoramic monitoring, multiple groups of cameras distributed at multiple locations in the target area can be set up, and the local images captured by each camera can be stitched to finally obtain a regional image of the entire target area covered by the LED lighting source.

[0023] The acquired regional image is the premise for the subsequent portrait recognition and segmentation of users in the target area. The higher the clarity of the regional image, the more accurate the subsequent user portrait recognition and segmentation effect will be. Therefore, a camera with a clarity greater than 500w pixels is used for acquisition. In addition, since the subsequent user feature recognition and lighting control are based on the regional image, the regional image needs to be acquired and updated in real time, and the acquisition frequency can be set to more than 30 frames per second.

[0024] The regional image is transmitted through wireless communication, and portrait recognition and segmentation are performed to obtain multiple user regions;

[0025] Furthermore, this step also includes:

[0026] Acquire a sample area image set according to the monitoring data records in the target area;

[0027] Marking and dividing the portrait area in the sample area image set to obtain a sample user area division result set;

[0028] Using the sample area image set and the sample user area division result set, based on semantic segmentation, construct a user identifier;

[0029] The user identifier is used to perform portrait recognition and region division on the region image to obtain the multiple user regions.

[0030] In a feasible implementation, after acquiring the regional image, the regional image is transmitted from the image acquisition device, i.e., a camera installed in the target area, to the user identifier by wireless communication. The transmission interface may adopt short-range wireless communication methods such as WiFi, Bluetooth, ZigBee, etc., and the transmission process is performed in real time.

[0031] Among them, the user identifier is a pre-built deep learning model for portrait recognition and segmentation, which can quickly and accurately identify the portrait area in the regional image, and subdivide the recognized portrait area to obtain multiple user areas. First, the monitoring data records in the target area are extracted from the database storing the monitoring data, that is, the image data obtained and recorded by the historical monitoring collection in the target area, and the sample area image set is screened to obtain. Among them, the sample area images of the sample area image set need to contain portraits of different numbers of people and different positions, such as portraits distributed in the middle or corner of the area, single portraits or multiple portraits coexisting, etc., so that the user identifier is more adaptable to complex scenes. Secondly, by using manual annotation, each area containing users is marked in the sample area image in the sample area image set to obtain a sample user area, and each sample user area is numbered, such as sample user area 1, sample user area 2, etc. Among them, the sample user area refers to the area within a certain range around each existing user, and the range can be set to within 0.5 meters around each user. By manually annotating the sample area image, multiple sample user areas of each sample area image are obtained, that is, which areas correspond to which users. Subsequently, the corresponding sample user region division results are extracted from each sample region image in the sample region image set by image segmentation to obtain a sample user region division result set. Then, by using the obtained sample region image set and the corresponding sample user region division result set, a user identifier is trained and constructed to realize user identification and region division of the region image of the target region. Specifically, a deep learning model is constructed as a user identifier based on semantic segmentation technology, wherein the semantic segmentation technology uses a convolutional neural network to classify each pixel of the region image and segment and identify different semantic targets in the region image. Then, the sample region image in the sample region image set is used as input, and the corresponding sample user region division results in the sample user region division result set are used as supervision labels. The parameters of the convolutional neural network are trained by the back propagation algorithm to obtain a semantic segmentation model, and its training goal is to minimize the error loss of the sample region image and the sample user region division result. After a large number of training iterations, the semantic segmentation model can extract the significant features of the portrait of the region image, realize accurate prediction and segmentation of the existing user region, and obtain a user identifier that can automatically realize portrait recognition and division of the region image, providing a basis for the processing of the region image.

[0032] After the regional image is transmitted to the user identifier, the user identifier automatically extracts the significant features of the regional image, and predicts the user areas in the regional image based on the knowledge learned during the training process, and gives each predicted user area a unique identifier to obtain the location and range information of all user areas in the regional image. Among them, the number of predicted user areas is the number of multiple user areas, which matches the actual number of people in the target area, and each user area corresponds to one user. The user identifier performs portrait recognition and division on the regional image to obtain multiple user areas, which provides a basis for intelligent lighting optimization control based on user characteristics.

[0033] Collecting images of multiple user areas to obtain multiple marked area images, and performing user type recognition on the multiple marked area images to obtain multiple user type information;

[0034] Further, such as Figure 2 As shown, this step also includes:

[0035] Obtain a set of sample marked area images, and mark the user type in each sample marked area image to obtain a set of sample user types;

[0036] Using the sample marked area image set and the sample user type set, based on a convolutional neural network, a user type identifier is constructed;

[0037] The user type identifier is used to perform user identification on the multiple marked area images to obtain multiple user type information.

[0038] In a feasible implementation, after obtaining multiple user areas, images in each user area are collected to form multiple marked area images. Each marked area image corresponds to a user area and carries unique identification information of the user area. Then, the user type identifier is used to identify the user type of the multiple marked area images to obtain multiple user type information, including adult males, adult females, elderly people, children, etc.

[0039] The user type identifier is a pre-built model for user type identification. First, based on each sample area image in the sample area image set, the corresponding sample marked area image is obtained to form a sample marked area image set. Then, for the sample marked area images in the sample marked area image set, the corresponding user types are manually annotated, such as adult males, adult females, children, etc., to obtain a sample user type set. Subsequently, the obtained sample marked area image set and the corresponding sample user type set are input into the preset convolutional neural network as training data. The convolutional neural network learns the image features of the sample marked area images corresponding to the sample user types through iterative training. After repeated training, the convolutional neural network model associates the image content and the user type category, realizes the classification judgment of the marked area image, and obtains the user type information in each user area. At this point, the user type identifier is obtained, which can automatically judge the user category in the image for the input multiple marked area images, and complete the user type recognition of multiple user areas.

[0040] Subsequently, the obtained multiple marked area images are sequentially input as inputs into the trained user type identifier. The user type identifier analyzes the content features of each input marked area image, determines which type of user contained in the marked area image belongs to, such as adult male, adult female, etc., and sequentially obtains the user category information contained in each input marked area image to form multiple user type information representing the user types in multiple user areas, laying a foundation for the subsequent accurate matching of feature recognition channels according to user types.

[0041] Based on the multiple user type information, mapping and selecting multiple matching feature recognition channels, performing user feature recognition on the multiple marked area images, and obtaining multiple user feature recognition results, wherein the user feature recognition results include user behavior and user sight direction;

[0042] Furthermore, this step includes:

[0043] Based on multiple sample user type information, multiple sample marked area image sets are obtained;

[0044] Marking user behaviors and user sight lines in the plurality of sample marked area images to obtain a plurality of sample user behavior sets and a plurality of sample user sight line direction sets;

[0045] Using the multiple sample marked area image sets, the multiple sample user behavior sets, and the multiple sample user sight direction sets, construct multiple feature recognition channels corresponding to the multiple sample user type information;

[0046] According to the multiple user type information, multiple matching feature recognition channels are mapped and selected, and input recognition is performed on the multiple marked area images to obtain multiple user behaviors and multiple user sight lines as multiple user feature recognition results.

[0047] Furthermore, multiple feature recognition channels are constructed, including:

[0048] Using the multiple sample marked area image sets and the multiple sample user behavior sets, constructing multiple user behavior recognition branches;

[0049] Using the multiple sample marked area image sets and the multiple sample user sight direction sets, constructing multiple user sight direction recognition branches;

[0050] By combining the multiple user behavior recognition branches and the multiple user sight direction recognition branches, multiple feature recognition channels are obtained.

[0051] In a preferred embodiment, in order to obtain sample data required for training feature recognition channels, first, based on multiple sample user type information, sample marked area image sets of corresponding types are obtained respectively to obtain multiple sample marked area image sets, that is, based on multiple sample user type information, such as adult males, adult females, children, etc., sample marked area images of different user types are collected in batches, and classified according to the user type in the image to form a corresponding sample user area image set. Each sample marked area image set contains sufficient image data showing different behaviors and sight directions of users of the same type, such as sample images of adult males reading, walking, talking, etc. Then, based on the manual annotation method, the sample marked area images in the obtained multiple sample marked area image sets are calibrated in turn for user behaviors and user sight directions. For each sample marked area image, the corresponding behavior category of the user in the image is annotated, such as standing, reading, talking, etc., so as to form multiple sample user behavior sets; at the same time, the head direction of the user is annotated to obtain the sight direction of the sample user, so as to form multiple sample user sight direction sets.

[0052] Then, based on multiple sample labeled area image sets and multiple sample user behavior sets, for each type of sample user type information, the sample images and corresponding behavior annotations are respectively constructed, that is, the sample labeled area image set and the sample user behavior set corresponding to a certain sample user type information, and an independent user behavior recognition branch is constructed. Among them, the construction method is implemented by deep learning structures such as convolutional neural networks, with the sample labeled area image as input, the corresponding sample user behavior as the supervisory signal, and the user behavior recognition branch parameters are adjusted by back propagation and other methods, and finally multiple user behavior recognition branches corresponding to multiple sample user type information are obtained, which provides a basis for constructing multiple feature recognition channels. At the same time, based on multiple sample labeled area image sets and multiple sample user sight direction sets, for each type of sample user type information, the sample labeled area image set and the sample user sight direction set are respectively constructed, and an independent user sight direction recognition branch is constructed. Among them, the construction method is implemented by deep learning structures such as convolutional neural networks, with the sample labeled area image as input, the corresponding sample user sight direction as the supervisory signal, and the user sight direction recognition branch parameters are adjusted by back propagation and other methods, and finally multiple user sight direction recognition branches corresponding to multiple sample user type information are obtained, which provides a basis for constructing feature recognition channels. After that, for each type of user information, the corresponding user behavior recognition branch and user sight direction recognition branch are combined to construct a feature recognition channel for identifying the behavior and sight characteristics of this type of user. Finally, based on multiple sample user type information, multiple matching feature recognition channels are obtained. When feature recognition is performed for a certain type of user, the matching feature recognition channel is selected to achieve personalized recognition of user classification.

[0053] Afterwards, according to the user type information in the multiple user areas obtained by identification, the matching feature recognition channels are searched and selected respectively. If an elderly user is identified in a certain user area, the feature recognition channel corresponding to the elderly user type is selected. Then, the marked area image in each user area is input separately or in parallel into the corresponding feature recognition channel to obtain the behavior category and sight direction results of the users in this type of user area. Subsequently, the user behavior and user sight direction of the identified output in each user area are summarized to form a feature recognition result set representing multiple users in the entire target area, and multiple user feature recognition results are obtained, so as to efficiently obtain the behavior and sight information of each user in the target area, and provide support for the formulation of refined intelligent lighting optimization control solutions based on user characteristics.

[0054] Optimizing the lighting control scheme of the lighting source array according to the plurality of user feature recognition results to obtain an optimal lighting control scheme;

[0055] Furthermore, this step includes:

[0056] Constructing a lighting control function according to the user behaviors and the user sight directions in the plurality of user feature recognition results;

[0057] Randomly combining the turned-on lighting sources, lighting directions and lighting brightness in the lighting source array to generate a lighting control optimization library;

[0058] According to the lighting control function, lighting control optimization is performed in the lighting control optimization library, the lighting adaptability of the lighting control scheme is calculated using the lighting control function, the lighting control scheme with the maximum lighting adaptability is output, and the optimal lighting control scheme is obtained.

[0059] Furthermore, the lighting control function is as follows:

[0060]

[0061] Among them, lig is the control fitness, w1 is the first weight, w2 is the second weight, w3 is the third weight, F i is the matching degree between the brightness of the user area where the i-th user is located and the user behavior in the lighting control scheme, K i is the intersection degree between the lighting light direction of the user area where the i-th user is located and the user's sight direction in the lighting control scheme, and P is the lighting power of the lighting control scheme.

[0062] In a feasible implementation, in order to optimize the lighting control scheme based on the user feature recognition results, first, according to the user behavior and the user sight direction in the multiple user feature recognition results, a lighting control function is constructed as follows:

[0063]

[0064] Among them, lig is the control fitness, w1 is the first weight, w2 is the second weight, w3 is the third weight, F i is the matching degree between the brightness of the user area where the i-th user is located and the user behavior in the lighting control scheme, K i is the intersection degree between the lighting light direction of the user area where the i-th user is located and the user's sight direction in the lighting control scheme, and P is the lighting power of the lighting control scheme.

[0065] In the lighting control function, lig is the fitness judgment score of the lighting control scheme, which is used to compare the advantages and disadvantages of different lighting control schemes and select the optimal lighting control scheme. The lighting control function is divided into three items, namely the evaluation scores of three indicators. The first item is the matching degree between the brightness of the user area where all users are located in the lighting control scheme and the user behavior. The larger the score of this item, the better the matching degree; the second item is the intersection degree between the lighting light direction of the user area where all users are located in the lighting control scheme and the user's line of sight direction. The smaller the score of this item, the better the effect, so the reciprocal is taken to reduce the glare interference of the lighting control scheme and avoid glare; the third item is the reciprocal of the lighting power of the lighting control scheme. The smaller the score of this item, the lower the energy consumption. At the same time, the weight of the influence is set for each item to perform weighted summation, comprehensively evaluate the quality of a lighting control scheme, and compare and select the optimal lighting control scheme.

[0066] Subsequently, for all the light sources in the lighting source array, an on state and an off state are set for each light source, and all the light sources are randomly turned on to form a random on combination. Then, based on the random on combination, a random brightness value is independently generated for each turned-on light source. At the same time, for the light sources with adjustable direction, a random direction parameter is independently generated based on each group of random brightness, so that the above light source on state, brightness value, and dimming direction parameter combination forms a random lighting control scheme. Repeat the entire process multiple times, store the generated random lighting control schemes, form a lighting control optimization library, and provide sample support for control scheme optimization. Afterwards, for multiple user feature recognition results, read the lighting control schemes in the lighting control optimization library in turn, and substitute them into the constructed lighting control function to calculate the control fitness under the current multiple user feature recognition results as the lighting fitness of each lighting control scheme. Then, compare the lighting fitness corresponding to the lighting control scheme in the lighting control optimization library under the current multiple user feature recognition results, select the lighting control scheme with the largest lighting fitness as the optimal lighting control scheme, and complete the automatic optimization determination of the lighting control scheme based on user features.

[0067] According to the optimal lighting control scheme, the lighting source array is controlled through wireless communication.

[0068] In the embodiment of the present application, after obtaining the optimal lighting control scheme, the obtained optimal lighting control scheme is sent to the control terminal of the lighting source array through wireless communication methods, such as WiFi, Bluetooth, ZigBee, etc., to realize intelligent control of the lighting source. Among them, the optimal lighting control scheme includes the corresponding opening status, brightness value and irradiation direction parameters of multiple lighting sources in the target area. Subsequently, the control terminal of the lighting source array controls and adjusts the switch status, brightness and lighting angle of each lighting source in the lighting source array according to the optimal lighting control scheme. In this way, the combined optimization and personalized control of the lighting source array for multiple user positions and user characteristics in the target scene are realized, and automatic adaptive dimming is realized to improve the lighting effect.

[0069] In summary, the intelligent lighting communication control method provided by the embodiment of the present application has the following technical effects:

[0070] Collect images of the target area covered by the lighting source array to obtain regional images, providing source data support for subsequent feature recognition and analysis. Transmit the regional images through wireless communication, perform portrait recognition and division, obtain multiple user areas, realize the transmission of the collected images, locate and identify different user areas, and prepare for the refined identification of different users. Collect images of multiple user areas to obtain multiple marked area images, perform user type recognition on the multiple marked area images, obtain multiple user type information, and provide a basis for matching and selecting feature recognition channels. Based on multiple user type information, map and select multiple matching feature recognition channels, perform user feature recognition on multiple marked area images, obtain multiple user feature recognition results, and accurately extract the behavior and sight information of different users. According to the multiple user feature recognition results, optimize the lighting control scheme of the lighting source array to obtain the optimal lighting control scheme, and determine the best lighting control strategy based on the specific characteristics of each user. According to the optimal lighting control scheme, control the lighting source array through wireless communication to achieve intelligent and rapid response to complex lighting scenes and improve the lighting effect.

[0071] Embodiment 2

[0072] Based on the same inventive concept as the intelligent lighting communication control method in the aforementioned embodiment, Figure 3 As shown, an embodiment of the present application provides an intelligent lighting communication control system, the system comprising:

[0073] The regional image acquisition module 11 is used to acquire images within the target area covered by the illumination light source array to obtain a regional image;

[0074] The area image division module 12 is used to transmit the area image through wireless communication, and perform portrait recognition and division to obtain multiple user areas;

[0075] The user type recognition module 13 is used to collect images of multiple user areas, obtain multiple marked area images, perform user type recognition on the multiple marked area images, and obtain multiple user type information;

[0076] A user feature recognition module 14 is used to map and select a plurality of matching feature recognition channels based on the plurality of user type information, perform user feature recognition on the plurality of marked area images, and obtain a plurality of user feature recognition results, wherein the user feature recognition results include user behavior and user sight direction;

[0077] A control scheme optimization module 15, configured to optimize the lighting control scheme of the lighting source array according to the plurality of user feature recognition results to obtain an optimal lighting control scheme;

[0078] The lighting source control module 16 is used to control the lighting source array through wireless communication according to the optimal lighting control solution.

[0079] Furthermore, the region image division module 12 includes the following execution steps:

[0080] Acquire a sample area image set according to the monitoring data records in the target area;

[0081] Marking and dividing the portrait area in the sample area image set to obtain a sample user area division result set;

[0082] Using the sample area image set and the sample user area division result set, based on semantic segmentation, construct a user identifier;

[0083] The user identifier is used to perform portrait recognition and region division on the region image to obtain the multiple user regions.

[0084] Furthermore, the user type identification module 13 includes the following execution steps:

[0085] Obtain a set of sample marked area images, and mark the user type in each sample marked area image to obtain a set of sample user types;

[0086] Using the sample marked area image set and the sample user type set, based on a convolutional neural network, a user type identifier is constructed;

[0087] The user type identifier is used to perform user identification on the multiple marked area images to obtain multiple user type information.

[0088] Furthermore, the user feature identification module 14 includes the following execution steps:

[0089] Based on multiple sample user type information, multiple sample marked area image sets are obtained;

[0090] Marking user behaviors and user sight lines in the plurality of sample marked area images to obtain a plurality of sample user behavior sets and a plurality of sample user sight line direction sets;

[0091] Using the multiple sample marked area image sets, the multiple sample user behavior sets, and the multiple sample user sight direction sets, construct multiple feature recognition channels corresponding to the multiple sample user type information;

[0092] According to the multiple user type information, multiple matching feature recognition channels are mapped and selected, and input recognition is performed on the multiple marked area images to obtain multiple user behaviors and multiple user sight lines as multiple user feature recognition results.

[0093] Furthermore, the user feature identification module 14 further includes the following execution steps:

[0094] Using the multiple sample marked area image sets and the multiple sample user behavior sets, constructing multiple user behavior recognition branches;

[0095] Using the multiple sample marked area image sets and the multiple sample user sight direction sets, constructing multiple user sight direction recognition branches;

[0096] By combining the multiple user behavior recognition branches and the multiple user sight direction recognition branches, multiple feature recognition channels are obtained.

[0097] Furthermore, the control scheme optimization module 15 includes the following execution steps:

[0098] Constructing a lighting control function according to the user behaviors and the user sight directions in the plurality of user feature recognition results;

[0099] Randomly combining the turned-on lighting sources, lighting directions and lighting brightness in the lighting source array to generate a lighting control optimization library;

[0100] According to the lighting control function, lighting control optimization is performed in the lighting control optimization library, the lighting adaptability of the lighting control scheme is calculated using the lighting control function, the lighting control scheme with the maximum lighting adaptability is output, and the optimal lighting control scheme is obtained.

[0101] Furthermore, the control scheme optimization module 15 also includes the following execution steps:

[0102] Construct the lighting control function as follows:

[0103]

[0104] Among them, lig is the control fitness, w1 is the first weight, w2 is the second weight, w3 is the third weight, F i is the matching degree between the brightness of the user area where the i-th user is located and the user behavior in the lighting control scheme, K i is the intersection degree between the lighting light direction of the user area where the i-th user is located and the user's sight direction in the lighting control scheme, and P is the lighting power of the lighting control scheme.

[0105] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0106] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to multiple elements that can be selected individually or in full. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. An intelligent lighting communication control method, characterized in that: The method comprises: Collect images within the target area covered by the illumination light source array to obtain an area image; The regional image is transmitted through wireless communication, and portrait recognition and segmentation are performed to obtain multiple user regions; Collecting images of multiple user areas to obtain multiple marked area images, and performing user type recognition on the multiple marked area images to obtain multiple user type information; Based on the multiple user type information, mapping and selecting multiple matching feature recognition channels, performing user feature recognition on the multiple marked area images, and obtaining multiple user feature recognition results, wherein the user feature recognition results include user behavior and user sight direction; Optimizing the lighting control scheme of the lighting source array according to the plurality of user feature recognition results to obtain an optimal lighting control scheme; According to the optimal lighting control scheme, the lighting source array is controlled through wireless communication.

2. The method according to claim 1, characterized in that Performing portrait recognition and segmentation on the regional image to obtain multiple user regions, including: Acquire a sample area image set according to the monitoring data records in the target area; Marking and dividing the portrait area in the sample area image set to obtain a sample user area division result set; Using the sample area image set and the sample user area division result set, based on semantic segmentation, construct a user identifier; The user identifier is used to perform portrait recognition and region division on the region image to obtain the multiple user regions.

3. The method according to claim 1, characterized in that Performing user type recognition on the multiple marked area images to obtain multiple user type information includes: Obtain a set of sample marked area images, and mark the user type in each sample marked area image to obtain a set of sample user types; Using the sample marked area image set and the sample user type set, based on a convolutional neural network, a user type identifier is constructed; The user type identifier is used to perform user identification on the multiple marked area images to obtain multiple user type information.

4. The method according to claim 1, characterized in that Based on the multiple user type information, mapping and selecting multiple matching feature recognition channels, performing user feature recognition on the multiple marked area images, and obtaining multiple user feature recognition results, including: Based on multiple sample user type information, multiple sample marked area image sets are obtained; Marking user behaviors and user sight lines in the plurality of sample marked area images to obtain a plurality of sample user behavior sets and a plurality of sample user sight line direction sets; Using the multiple sample marked area image sets, the multiple sample user behavior sets, and the multiple sample user sight direction sets, construct multiple feature recognition channels corresponding to the multiple sample user type information; According to the multiple user type information, multiple matching feature recognition channels are mapped and selected, and input recognition is performed on the multiple marked area images to obtain multiple user behaviors and multiple user sight lines as multiple user feature recognition results.

5. The method according to claim 4, characterized in that Using the multiple sample marked area image sets, the multiple sample user behavior sets, and the multiple sample user sight direction sets, multiple feature recognition channels corresponding to the multiple sample user type information are constructed, including: Using the multiple sample marked area image sets and the multiple sample user behavior sets, constructing multiple user behavior recognition branches; Using the multiple sample marked area image sets and the multiple sample user sight direction sets, constructing multiple user sight direction recognition branches; By combining the multiple user behavior recognition branches and the multiple user sight direction recognition branches, multiple feature recognition channels are obtained.

6. The method according to claim 1, characterized in that Optimizing the lighting control scheme of the lighting source array according to the plurality of user feature recognition results to obtain an optimal lighting control scheme includes: Constructing a lighting control function according to the user behaviors and the user sight directions in the plurality of user feature recognition results; Randomly combining the turned-on lighting sources, lighting directions and lighting brightness in the lighting source array to generate a lighting control optimization library; According to the lighting control function, lighting control optimization is performed in the lighting control optimization library, the lighting adaptability of the lighting control scheme is calculated using the lighting control function, the lighting control scheme with the maximum lighting adaptability is output, and the optimal lighting control scheme is obtained.

7. The method according to claim 6, characterized in that According to the user behaviors and user sight directions in the multiple user feature recognition results, a lighting control function is constructed as follows: Among them, lig is the control fitness, w1 is the first weight, w2 is the second weight, w3 is the third weight, F i is the matching degree between the brightness of the user area where the i-th user is located and the user behavior in the lighting control scheme, K i is the intersection degree between the lighting light direction of the user area where the i-th user is located and the user's sight direction in the lighting control scheme, and P is the lighting power of the lighting control scheme.

8. An intelligent lighting communication control system, characterized in that: A system for implementing an intelligent lighting communication control method according to any one of claims 1 to 7, comprising: A regional image acquisition module, the regional image acquisition module is used to acquire an image within a target area covered by the illumination light source array to obtain a regional image; A region image division module, the region image division module is used to transmit the region image through wireless communication, and perform portrait recognition and division to obtain multiple user regions; A user type recognition module, the user type recognition module is used to collect images of multiple user areas, obtain multiple marked area images, perform user type recognition on the multiple marked area images, and obtain multiple user type information; A user feature recognition module, the user feature recognition module is used to map and select a plurality of matching feature recognition channels based on the plurality of user type information, perform user feature recognition on the plurality of marked area images, and obtain a plurality of user feature recognition results, wherein the user feature recognition results include user behavior and user sight direction; A control scheme optimization module, the control scheme optimization module is used to optimize the lighting control scheme of the lighting source array according to the multiple user feature recognition results to obtain an optimal lighting control scheme; An illumination light source control module, wherein the illumination light source control module is used to control the illumination light source array through wireless communication according to the optimal illumination control scheme.

Images