A comprehensive intelligent data monitoring system for large-scale exhibitions

Abstract

The present invention provides a comprehensive intelligent monitoring system for data of a large-scale exhibition, belonging to the field of intelligent monitoring technology, and comprising: a temperature and humidity monitoring module, which is used to monitor the temperature and humidity data of the exhibition venue in real time, automatically adjust the temperature and humidity, and maintain the best environmental conditions; an air quality monitoring module, which is used to analyze the air quality in the exhibition venue, timely discover abnormal conditions, link air purification equipment, and ensure the air quality of the exhibition site; a light monitoring module, which is used to monitor the operating status of the lighting system in the exhibition venue and timely discover faults; an energy consumption monitoring module, which is used to obtain the energy consumption data of various equipment in the exhibition venue in real time; a video monitoring module, which comprises a monitoring submodule, a split storage submodule, and a playback submodule; wherein the monitoring submodule is used to comprehensively collect monitoring videos in the exhibition venue; and the playback submodule is used to merge the split compressed video files into monitoring images during playback and play them sequentially.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring, and in particular relates to an all-round intelligent monitoring system for data of a large-scale exhibition. Background Art

[0002] With the rapid development of information technology, large-scale exhibitions are playing an increasingly important role in meeting people's various needs. Large-scale exhibitions usually refer to large-scale and influential conferences, exhibitions or related activities. Such activities often involve many participants, including industry professionals, corporate representatives, experts and scholars, etc. They are held regularly or irregularly in a certain geographical space to promote information exchange, business cooperation, technology display and cultural heritage. Large-scale exhibitions usually involve vast venues, a large number of exhibitors and visitors, and put forward high requirements for venue management, equipment operation and maintenance, and environmental monitoring. Most of the existing exhibition management systems use traditional methods such as manual inspections and simple monitoring, which make it difficult to discover and deal with various problems in a timely manner, and cannot ensure the efficient, safe and comfortable conduct of exhibition activities.

[0003] At present, more and more exhibition venues are beginning to introduce intelligent monitoring technology, using various sensors and video equipment to collect and analyze information such as temperature and humidity, air quality, lighting, and energy consumption inside the venue in real time. Some large exhibition venues have deployed relatively complete intelligent monitoring systems that can detect problems in a timely manner, automatically adjust environmental parameters, and maximize the exhibition experience. However, these systems usually have the problem that most video recordings use traditional encoding compression and storage methods, which cannot effectively reduce the storage overhead of monitoring data, resulting in the system being unable to store a large amount of monitoring videos for a long time. This not only increases storage costs, but also greatly reduces the efficiency of data retrieval and analysis. Summary of the invention

[0004] In view of this, the present invention provides a comprehensive intelligent data monitoring system for large-scale exhibitions, which can solve the technical problems in the prior art of video monitoring, that is, the system cannot effectively reduce the storage overhead of monitoring data, resulting in the system being unable to store a large amount of monitoring videos for a long time, which increases storage costs and greatly reduces the efficiency of data retrieval and analysis.

[0005] The present invention is achieved in that:

[0006] The first aspect of the present invention provides a comprehensive intelligent monitoring system for large-scale exhibition data, which includes the following functional modules:

[0007] Temperature and humidity monitoring module, used to monitor the temperature and humidity data of the exhibition venue in real time, automatically adjust the temperature and humidity, and maintain the best environmental conditions;

[0008] The air quality monitoring module is used to analyze the air quality in the exhibition venue, detect abnormal conditions in time, and link air purification equipment to ensure the air quality at the exhibition site;

[0009] Light monitoring module, used to monitor the operating status of the lighting system in the exhibition venue and detect faults in time;

[0010] Energy consumption monitoring module, used for real-time energy consumption data of various equipment in the exhibition venue;

[0011] The video surveillance module includes a monitoring submodule, a split storage submodule and a playback submodule; wherein the monitoring submodule is used to collect surveillance videos in the exhibition venue in all directions; the split storage submodule adopts the frame-fill content splitting method to store the surveillance video as split compressed video files, and the playback submodule is used to merge the split compressed video files into surveillance images during playback and play them sequentially.

[0012] On the basis of the above technical solution, the comprehensive intelligent monitoring system for large-scale exhibition data of the present invention can also be improved as follows:

[0013] The monitoring submodule is used to perform the following steps:

[0014] S11. Collect panoramic video data of the exhibition venue, including surveillance video information from different angles and locations;

[0015] S12, performing time synchronization and spatial splicing on the collected video data to form complete panoramic video monitoring information;

[0016] S13, encoding and compressing the stitched panoramic video to obtain video file blocks for storage, and sending each video file block to the split storage submodule.

[0017] The step S11 comprises:

[0018] Step 1) Multiple high-definition cameras with panoramic shooting capabilities are arranged at key locations of the exhibition venue to cover the main areas of the venue;

[0019] Step 2) For each video stream collected by the camera, the NTP network time synchronization protocol is used for time synchronization processing to ensure that the timestamps between the video clips of each channel are consistent;

[0020] Step 3) Use efficient video coding standards such as H.265 to encode and compress the synchronized video data, and set different compression parameters according to the importance of different areas.

[0021] The step S12 comprises:

[0022] Step 1) Using the camera's position information and shooting angle, perform spatial registration and geometric correction on each channel of video data to remove distortion and perspective differences;

[0023] Step 2) Using feature point-based image stitching algorithms, such as SIFT and SURF, the corrected video data is spatially stitched to generate a complete panoramic video image;

[0024] Step 3) Perform H.265 encoding compression on the stitched panoramic video, and adjust the bit rate and quality parameters according to the complexity of the picture.

[0025] The step S13 comprises:

[0026] Step 1) The panoramic video data is encoded and compressed using the H.265 encoding standard, and different compression parameters can be set according to the importance of different areas;

[0027] Step 2) splitting the encoded and compressed panoramic video data into video file blocks of fixed length, such as 10 to 30 seconds per file block;

[0028] Step 3) Send each generated video file block to the split storage submodule for further processing.

[0029] The split storage submodule is used to perform the following steps:

[0030] S21, obtaining the video file block and extracting a semantic key frame, wherein the semantic key frame is a frame containing the most comprehensive semantic segmentation objects among a plurality of adjacent frames;

[0031] S22, performing semantic segmentation on each frame in the video file block to obtain a plurality of segmented image blocks;

[0032] S23, extracting a frame from each segmented block, and splitting the frame into a frame and a filling pattern;

[0033] S24, establishing a storage block for storing the frame of each segmented image block and each filling content, as well as the timestamp and position of the segmented image block in the original video;

[0034] S25, deleting repeated filling content;

[0035] S26: Store the storage block in a storage device.

[0036] The semantic key frame is the frame with the largest number of semantic objects detected among multiple adjacent frames. The number mark of the segmented block includes its coordinate information and timestamp information in the original video frame. The frame part is extracted using an edge detection algorithm, and the filling pattern part is the internal area obtained by subtracting the frame part from the segmented block. The information recorded in the storage block includes: the frame coordinates, filling pattern data, timestamp and space coordinates of each segmented block in the original video. The storage device is a high-speed hard disk or a distributed storage system, and performs backup and off-site disaster recovery processing. The storage block can be hierarchically managed by time period or area.

[0037] The playback submodule is used to perform the following steps:

[0038] S31, reading the stored compressed video file from the storage device;

[0039] S32, decoding the read compressed video file to restore the original segmented blocks;

[0040] S33, resynthesizing each segmented image block according to the stored frame and filling content information;

[0041] S34, splicing the resynthesized segmented image blocks in chronological order to restore a complete surveillance video image;

[0042] S35, playing back and displaying the restored surveillance video images.

[0043] The step S31 comprises:

[0044] Step 1) According to the user's query conditions, such as time period, region, etc., the corresponding storage block data is located from the storage device;

[0045] Step 2) Read the storage block data into the memory, preparing for subsequent decoding and reconstruction operations.

[0046] The step S32 comprises:

[0047] Step 1) decompressing the storage block data read in step S31 to restore the frame coordinates and filling content data of each segmented block;

[0048] Step 2) According to the stored spatiotemporal information, each segmented tile is restored to the corresponding time and space position to form a complete set of segmented tiles.

[0049] The step S33 comprises:

[0050] Step 1) for each segmented image block, regenerate pixel data of the segmented image block according to the stored frame coordinates and filling content data;

[0051] Step 2) splicing the resynthesized segmented blocks to corresponding positions according to the stored spatiotemporal information to form a complete video frame;

[0052] Step 3) Arrange the continuous video frames in time sequence to complete the reconstruction of the entire video picture.

[0053] The step S34 comprises:

[0054] Step 1) traverse all video frames reconstructed in step S33 and arrange them in the order of stored timestamps;

[0055] Step 2) splicing the arranged video frames in sequence to form a continuous video picture sequence;

[0056] Step 3) Smooth the spliced ​​video image sequence to eliminate the visual discontinuity caused by the splicing of segmented blocks.

[0057] The step S35 comprises:

[0058] Step 1) decoding the video picture sequence spliced ​​in step S34 into a playable video format through a video decoder;

[0059] Step 2) The decoded video data is played back in real time through a video player, providing basic control functions such as pause and fast forward;

[0060] Step 3) According to user needs, advanced playback functions such as perspective switching and area zooming can be added to enhance the visualization effect of surveillance video.

[0061] Further, the step S22 comprises:

[0062] Step 1) Using the semantic segmentation algorithm in step S21, semantic segmentation is performed on each frame of the video file block;

[0063] Step 2) According to the segmentation results, each frame image is divided into multiple semantic segmentation blocks, each block corresponds to a semantic object in the scene;

[0064] Step 3) Number each segmented block and record its position coordinates and timestamp information in the original video frame.

[0065] Further, the step S23 comprises:

[0066] Step 1) performing edge detection on each segmented image block obtained in step S22 to extract the contour line of the image block, i.e., the frame part;

[0067] Step 2) The inner area obtained by subtracting the frame part from the segmented block is the filling pattern part;

[0068] Step 3) Record the frame coordinate information of each segmented block and the pixel data of the filling pattern.

[0069] Further, the step S24 comprises:

[0070] Step 1) creating a storage block for storing the following information: frame coordinates of each segmented tile, fill pattern data of each segmented tile, timestamp of each segmented tile in the original video, and spatial coordinates of each segmented tile in the original video;

[0071] Step 2) storing the above information of each segmented block in the storage block according to a certain data structure;

[0072] Step 3) Use LZW or Huffman coding algorithm to compress the data in the storage block.

[0073] Further, the step S25 comprises:

[0074] Step 1) traverse the filling contents of all segmented blocks in the storage block to check whether there is repeated pixel data;

[0075] Step 2) For repeated filling content, only one copy of the actual data is retained and replaced by a pointer reference in other segmented blocks;

[0076] Step 3) Update the filling content information of the split tiles in the storage block, and replace the repeated actual data with pointer references.

[0077] Further, the step S26 comprises:

[0078] Step 1) storing the optimized storage blocks in the form of files or databases in a high-speed hard disk or distributed storage system;

[0079] Step 2) Back up the storage blocks and perform off-site disaster recovery to ensure data security and reliability.

[0080] The method also includes step 3) whereby the storage blocks can be managed hierarchically as required, such as by time period or region.

[0081] Furthermore, the semantic segmentation algorithm adopts a deep learning-based method.

[0082] Compared with the prior art, the beneficial effects of the all-round intelligent monitoring system for large-scale exhibition data provided by the present invention are:

[0083] 1. More efficient surveillance data collection and storage. This system uses advanced video coding compression technology and layered storage methods to significantly reduce the storage overhead of surveillance data. At the same time, through methods such as semantic segmentation and repeated deletion of padding content, the storage structure is further optimized, and a large amount of surveillance video data can be stored for a long time at a very low cost. This provides a good foundation for subsequent data retrieval and analysis.

[0084] 2. Deeper monitoring information mining. This system not only collects conventional monitoring data such as temperature and humidity, air quality, and energy consumption, but also uses advanced video analysis technology to perform deep semantic segmentation and understanding of monitoring videos, and can extract more abundant monitoring information such as the activity trajectory of participants and exhibition hot spots. This information plays an important role in decision-making support for exhibition operation management.

[0085] 3. Stronger system integration and scalability. This system adopts a modular design and can be seamlessly connected with other subsystems of the exhibition venue, such as security, fire protection, and intelligent lighting, to achieve comprehensive intelligent management. At the same time, each functional module of the system has good scalability and can flexibly add new monitoring and analysis capabilities according to needs to meet the continuous upgrading needs of exhibition venue management in the future.

[0086] In summary, the solution of the present invention solves the technical problem that the prior art in video surveillance cannot effectively reduce the storage overhead of monitoring data, resulting in the system being unable to store a large amount of monitoring videos for a long time, increasing storage costs and significantly reducing the efficiency of data retrieval and analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0088] Figure 1 A schematic diagram of the structure of a comprehensive intelligent monitoring system for large-scale exhibition data provided by the present invention;

[0089] Figure 2 A schematic diagram of the execution steps of the monitoring submodule;

[0090] Figure 3 A schematic diagram of the steps to be performed for splitting the storage submodule;

[0091] Figure 4 Schematic diagram of the steps performed for the playback submodule. DETAILED DESCRIPTION

[0092] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0093] like Figure 1 FIG. 1 is a schematic diagram of a large-scale exhibition data all-round intelligent monitoring system provided by the present invention. The system includes the following functional modules:

[0094] Temperature and humidity monitoring module, used to monitor the temperature and humidity data of the exhibition venue in real time, automatically adjust the temperature and humidity, and maintain the best environmental conditions;

[0095] The air quality monitoring module is used to analyze the air quality in the exhibition venue, detect abnormal conditions in time, and link air purification equipment to ensure the air quality at the exhibition site;

[0096] Light monitoring module, used to monitor the operating status of the lighting system in the exhibition venue and detect faults in time;

[0097] Energy consumption monitoring module, used for real-time energy consumption data of various equipment in the exhibition venue;

[0098] The video surveillance module includes a surveillance submodule, a split storage submodule and a playback submodule; wherein the surveillance submodule is used to collect surveillance videos in the exhibition venue in all directions; the split storage submodule adopts the frame-fill content splitting method to store the surveillance video as split compressed video files, and the playback submodule is used to merge the split compressed video files into surveillance images and play them sequentially during playback.

[0099] like Figure 2 As shown, the monitoring submodule is used to perform the following steps:

[0100] S11. Collect panoramic video data of the exhibition venue, including surveillance video information from different angles and locations;

[0101] S12, performing time synchronization and spatial splicing on the collected video data to form complete panoramic video monitoring information;

[0102] S13, encoding and compressing the spliced ​​panoramic video to obtain a video file block for storage, and sending each video file block to the split storage submodule;

[0103] like Figure 3 As shown, the split storage submodule is used to perform the following steps:

[0104] S21, obtaining the video file block and extracting a semantic key frame, wherein the semantic key frame is a frame containing the most comprehensive semantic segmentation objects among a plurality of adjacent frames;

[0105] S22, performing semantic segmentation on each frame in the video file block to obtain a plurality of segmented image blocks;

[0106] S23, extracting a frame from each segmented block, and splitting the frame into a frame and a filling pattern;

[0107] S24, establishing a storage block for storing the frame of each segmented image block and each filling content, as well as the timestamp and position of the segmented image block in the original video;

[0108] S25, deleting repeated filling content;

[0109] S26: Store the storage block in a storage device.

[0110] like Figure 4 As shown, the playback submodule is used to perform the following steps:

[0111] S31, reading the stored compressed video file from the storage device;

[0112] S32, decoding the read compressed video file to restore the original segmented blocks;

[0113] S33, resynthesizing each segmented image block according to the stored frame and filling content information;

[0114] S34, splicing the resynthesized segmented image blocks in chronological order to restore a complete surveillance video image;

[0115] S35, playing back and displaying the restored surveillance video images.

[0116] The specific implementation methods of the modules mentioned above and the related steps of some modules are described in detail below:

[0117] The specific implementation of the temperature and humidity monitoring module is as follows:

[0118] 1. Sensor layout: This module requires multiple temperature and humidity sensors to be placed in different areas of the exhibition venue to obtain comprehensive environmental data. The sensor type can be a high-precision digital temperature and humidity sensor, or a more economical analog sensor. Regardless of the type of sensor used, regular calibration is required to ensure data accuracy.

[0119] 2. Data collection and transmission: The temperature and humidity data from each sensor are transmitted to the central monitoring system via wired or wireless methods. The wired method is mainly Ethernet, and the wireless method can choose low-power wide area network technologies such as ZigBee, WiFi or LoRaWAN.

[0120] 3. Data processing and analysis: The monitoring system collects real-time temperature and humidity data from all sensors and analyzes and compares them according to the preset target value range. If the temperature and humidity in a certain area exceed the target range, the system will trigger an alarm and prompt the corresponding air conditioning equipment to be adjusted.

[0121] 4. Environmental control: The air conditioning control system is integrated with the temperature and humidity monitoring module. According to the monitoring data, the cooling and heating equipment in each area can be automatically adjusted to make it work in the best state. At the same time, the air conditioning equipment can also be manually adjusted remotely according to specific needs.

[0122] The specific implementation of the air quality monitoring module is as follows:

[0123] 1. Sensor selection: This module needs to use sensors that can detect major air pollutants, such as PM2.5 / PM10 particulate matter, TVOC, CO2, formaldehyde, etc. You can use a single pollutant sensor or an integrated multi-gas sensor for flexible selection.

[0124] 2. Sensor layout: Similar to temperature and humidity sensors, air quality sensors need to be reasonably arranged according to the area of ​​the exhibition venue, venue zoning, etc., in order to comprehensively monitor the air conditions in each area.

[0125] 3. Data processing and analysis: Compare the collected air quality data with the established health standard values, determine the air quality level (excellent, medium, polluted, etc.), and set alarm thresholds for suspicious abnormal data.

[0126] 4. Purification: Once the air quality in a certain area is found to be declining, the system will notify the on-site staff and automatically start the air purification equipment in the corresponding area to quickly improve the on-site air quality.

[0127] 5. User interaction: Considering that the air quality requirements in different occasions are different, the system should also have the function of manual adjustment and control, so that users can adjust the air purification effect according to their needs.

[0128] The specific implementation of the light monitoring module is as follows:

[0129] 1. Illumination monitoring: This module includes a series of light intensity sensors for real-time monitoring of the illumination level of the exhibition venue. The lighting of the exhibition venue usually consists of two parts: main lights and auxiliary lights, which need to be monitored and adjusted separately.

[0130] 2. Sensor layout: According to the site plan, the illumination sensors should be reasonably arranged in various areas of the site, especially in areas where the required illumination level is high and changes frequently.

[0131] 3. Preset lighting mode: The system has preset several standard lighting modes, such as exhibition mode, conference mode, performance mode, etc. Each mode has a specific target illumination level. Users can switch to the required lighting mode at any time.

[0132] 4. Fault diagnosis and maintenance: If the actual illumination of a certain area deviates too much from the target value, the system will trigger an alarm and mark the fault area. Maintenance personnel can quickly determine the fault location based on the alarm area and repair or debug related lighting equipment in a timely manner.

[0133] 5. Energy-saving control: The light monitoring module is also connected to the energy management system to dynamically adjust the power of the lamps according to the illumination requirements of different areas to achieve energy-saving effects. When no one is active in the area, it can also actively switch to low-power mode.

[0134] The specific implementation of the energy consumption monitoring module is as follows:

[0135] 1. Energy consumption metering terminal: This module needs to measure the power consumption of various electrical equipment in the exhibition venue in real time, which requires the installation of smart meters or communication-type energy metering terminals on the equipment side. The selection of smart meters based on current transformers can accurately measure the real-time power consumption of each device.

[0136] 2. Data collection and storage: As the collection center of energy consumption data, the energy management system collects electricity consumption data transmitted by each electricity metering terminal in real time through industrial Ethernet or Internet of Things communication technology, and stores it in the database.

[0137] 3. Energy consumption analysis and statistics: The energy management system analyzes and compiles statistics on the collected raw data, generates various energy consumption reports by equipment type, time period and other dimensions, and directly reflects the energy usage. At the same time, historical data can be applied to energy consumption forecast analysis.

[0138] 4. Remote control of equipment: For controllable load equipment, such as air conditioners, lighting fixtures, etc., the energy management system can remotely issue control instructions to achieve remote start and stop and parameter adjustment of the equipment, thereby optimizing the operating status and energy consumption level of the equipment.

[0139] Step S11: Collect panoramic video data of the exhibition venue, including surveillance video information from different angles and locations:

[0140] The purpose of this step is to collect panoramic video data of the exhibition venue and provide raw materials for subsequent video processing and storage. The specific implementation method is:

[0141] 1) Multiple high-definition cameras should be placed at key locations in the exhibition venue to cover the main areas of the venue. The cameras should have panoramic shooting capabilities and can capture 180-degree or even 360-degree panoramic videos.

[0142] 2) For each video stream captured by the camera, time synchronization processing is required to ensure that the timestamps between the video clips are consistent. This can be achieved through the NTP network time synchronization protocol.

[0143] 3) Encode and compress the synchronized video data to reduce storage space usage. Efficient video encoding standards such as H.265 can be used, and different compression parameters can be set according to the importance of different areas to retain higher image quality for key areas.

[0144] Step S12: Time synchronization and spatial splicing of the collected video data to form complete panoramic video monitoring information:

[0145] The purpose of this step is to spatially stitch the multi-channel synchronized video data to form a complete panoramic video surveillance screen. The specific implementation method is:

[0146] 1) Using the camera’s location information and shooting angle, perform spatial registration and geometric correction on each video data to remove distortion and perspective differences.

[0147] 2) According to the camera position and viewing angle, use the image fusion algorithm to stitch the corrected video data to generate a complete panoramic video screen. Here, feature point-based image stitching algorithms such as SIFT and SURF can be used.

[0148] 3) Encode and compress the stitched panoramic video to reduce storage space usage. H.265 encoding can be used, and the bit rate and quality parameters can be adjusted according to the complexity of the picture.

[0149] Step S13: Encode and compress the spliced ​​panoramic video to obtain storable video file blocks, and send each video file block to the split storage submodule:

[0150] The purpose of this step is to encode and compress the spliced ​​panoramic video to generate storable video file blocks and send them to the subsequent split storage submodule for processing. The specific implementation method is:

[0151] 1) Use efficient video coding standards such as H.265 to encode and compress panoramic video data. Different compression parameters can be set according to the importance of different areas to retain higher image quality for key areas.

[0152] 2) Split the encoded and compressed panoramic video data into video file blocks of fixed length, for example, each file block is 10 seconds. This will facilitate subsequent storage and retrieval.

[0153] 3) Each generated video file block is sent to the split storage submodule for further processing.

[0154] Step S21: Obtain the video file block and extract the semantic key frame, which is the frame containing the most comprehensive semantic segmentation objects among multiple adjacent frames:

[0155] The purpose of this step is to extract semantic key frames from the video file blocks to provide a basis for subsequent segmentation and storage. The specific implementation method is:

[0156] 1) Get the video file block transmitted from the video monitoring submodule.

[0157] 2) Perform semantic segmentation on each frame of the video file block to identify the semantic objects in the scene, such as people, equipment, signs, etc. This can be done using a semantic segmentation algorithm based on deep learning, such as Mask R-CNN, DeepLab, etc.

[0158] 3) Count the number of semantic objects detected in each frame, and find the frame with the richest semantic objects among the adjacent frames, which is the semantic key frame. This can preserve all the semantic information in the scene as much as possible.

[0159] Step S22: semantically segment each frame in the video file block to obtain multiple segmented blocks:

[0160] The purpose of this step is to perform semantic segmentation on each frame of the video file block to generate multiple segmented blocks. The specific implementation method is:

[0161] 1) Using the semantic segmentation algorithm mentioned in step S21, semantic segmentation is performed on each frame image in the video file block.

[0162] 2) According to the segmentation results, each frame of the image is divided into multiple semantic segmentation blocks. Each block corresponds to a semantic object in the scene, such as a person, equipment, sign, etc.

[0163] 3) Mark each segmented block with a number and record its position coordinates and timestamp information in the original video frame.

[0164] Step S23: Extract the frame of each segmented block and split it into a frame and a filling pattern:

[0165] The purpose of this step is to further split each semantic segmentation tile into a frame and a filling pattern to prepare for subsequent efficient storage. The specific implementation method is:

[0166] 1) Perform edge detection on each segmented block obtained in step S22 to extract the outline of the block, that is, the frame part.

[0167] 2) The inner area obtained by subtracting the frame part from the segmented block is the filling pattern part.

[0168] 3) Record the frame coordinate information of each segmented block and the pixel data of the filling pattern.

[0169] Step S24: Establish a storage block for storing the frame of each segmented tile and each filling content, as well as the timestamp and position of the segmented tile in the original video:

[0170] The purpose of this step is to establish a unified storage structure for storing the frame, filling content and spatiotemporal information of each segmented tile in the original video. The specific implementation method is:

[0171] 1) Create a storage block to store the following information:

[0172] -Frame coordinates of each split tile

[0173] -Fill pattern data for each split tile

[0174] - Timestamp of each segmented tile in the original video

[0175] -The spatial coordinates of each segmented tile in the original video

[0176] 2) For each segmented block, its frame coordinates, fill pattern data, timestamp and spatial coordinate information are organized and stored in the storage block according to a certain data structure.

[0177] 3) Use efficient data compression algorithms, such as LZW and Huffman coding, to compress the data in the storage block to further save storage space.

[0178] Step S25-delete duplicate filling content:

[0179] The purpose of this step is to further optimize the storage space and delete the repeated padding content in the storage block. The specific implementation method is:

[0180] 1) Traverse the filling contents of all segmented tiles in the storage block to check whether there is duplicate pixel data.

[0181] 2) For repeated filling content, only one copy of the actual data is retained and replaced by a pointer reference in other segmented tiles.

[0182] 3) Update the filling content information of the split tiles in the storage block, and replace the repeated actual data with pointer references.

[0183] Step S26: storing the storage block in a storage device:

[0184] The purpose of this step is to store the optimized storage block persistently in the storage device to provide data support for subsequent video playback. The specific implementation method is:

[0185] 1) The storage blocks generated in steps S24 and S25 are stored in a high-speed hard disk or a distributed storage system in the form of files or databases.

[0186] 2) Back up storage blocks and perform disaster recovery in different locations to ensure data security and reliability.

[0187] 3) Storage blocks can be managed in layers as needed, such as by time period or area.

[0188] Step S31: Read the stored compressed video file from the storage device:

[0189] The purpose of this step is to read the previously stored compressed video file from the storage device to prepare for subsequent video reconstruction. The specific implementation method is:

[0190] 1) According to the user's query conditions, such as time period, area, etc., locate the corresponding storage block data from the storage device.

[0191] 2) Read the storage block data into the memory, preparing for subsequent decoding and reconstruction operations.

[0192] Step S32: Decode the compressed video file read to restore the original segmented blocks:

[0193] The purpose of this step is to decode the compressed video file read in step S31 and restore the original semantic segmentation block. The specific implementation method is:

[0194] 1) Decompress the storage block data read in step S31 to restore the frame coordinates and filling content data of each segmented block.

[0195] 2) According to the stored spatiotemporal information, each segmented block is restored to the corresponding time and space position to form a complete set of segmented blocks.

[0196] Step S33: Resynthesize each segmented image block according to the stored frame and fill content information:

[0197] The purpose of this step is to use the segmented block information restored in step S32 to re-synthesize a complete video screen. The specific implementation method is:

[0198] 1) For each segmented tile, the pixel data of the segmented tile is regenerated according to the stored frame coordinates and filling content data.

[0199] 2) The resynthesized segmented blocks are spliced ​​to the corresponding positions according to the stored time and space information to form a complete frame of video.

[0200] 3) Arrange the continuous video frames in time sequence to complete the reconstruction of the entire video image.

[0201] Step S34: stitch the resynthesized segmented image blocks in chronological order to restore the complete surveillance video image:

[0202] The purpose of this step is to splice the segmented blocks resynthesized in step S33 in chronological order to restore the complete surveillance video screen. The specific implementation method is:

[0203] 1) Traverse all video frames reconstructed in step S33 and arrange them in the order of stored timestamps.

[0204] 2) Splice the arranged video frames in sequence to form a continuous video image sequence.

[0205] 3) Smooth the spliced ​​video image sequence to eliminate the visual discontinuity caused by the splicing of segmented blocks.

[0206] Step S35: Play back and display the restored surveillance video:

[0207] The purpose of this step is to play back and display the complete surveillance video screen restored in step S34 to provide users with visual surveillance information. The specific implementation method is:

[0208] 1) Decode the video frame sequence spliced ​​in step S34 into a playable video format through a video decoder.

[0209] 2) The decoded video data is played back in real time through a video player. The player can provide basic control functions such as pause and fast forward.

[0210] 3) According to user needs, advanced playback functions such as perspective switching and area zooming can be added to enhance the visualization effect of surveillance videos.

[0211] In order to better understand the solution of the present invention, the steps involved in the specific calculation are described in more detail below in combination with the formula:

[0212] Step S21: Obtain video file blocks and extract semantic key frames

[0213] The purpose of this step is to extract the key frames containing the richest semantic information in the scene from the video file block, providing a basis for subsequent segmentation and storage. The specific implementation process is as follows:

[0214] Assume that the video file block contains N video frames, denoted as {I 1 ,I 2 ,...,I N}. For each video frame I i , use the semantic segmentation algorithm to segment it and get M i semantic segmentation region, denoted as Each area R ij Contains information about the location, size, category, etc. of the semantic object.

[0215] Next, count the number of semantic objects detected in each video frame, denoted as n i =M i Find the frame with the largest number of semantic objects among multiple adjacent video frames, which is the semantic key frame and is denoted by I k .

[0216] Formally, it can be expressed as:

[0217]

[0218] In this way, it can be ensured that the semantic keyframe can contain all the semantic information in the scene as much as possible.

[0219] Step S22: Semantically segment each frame in the video file block

[0220] The purpose of this step is to perform semantic segmentation on each frame of the video file block and generate multiple segmentation blocks. The specific implementation process is as follows:

[0221] For each video frame I i , use semantic segmentation algorithms (such as Mask R-CNN, DeepLab, etc.) to segment and obtain M i Semantic segmentation region Each segmented region R ij Contains the location coordinates (x ij ,y ij )、Size(w ij ,h ij ) and the category label c ij .

[0222] Formally, it can be expressed as:

[0223] R ij ={(x ij ,y ij ),(w ij ,h ij ),c ij}

[0224] Among them, (xij ,y ij ) is the segmentation area R ij In the original video frame I i The position coordinates in (w ij ,h ij ) is the width and height of the segmented area; c ij is the category label of the semantic object, such as personnel, equipment, signs, etc.

[0225] For each segmented region R ij Mark with numbers and record them in the original video frame I i The spatiotemporal information in the video provides a basis for subsequent storage and playback.

[0226] Step S23: Extract and split each segmented block into its own frame

[0227] The purpose of this step is to further split each semantic segmentation tile into a frame and a filling pattern to prepare for subsequent efficient storage. The specific implementation process is as follows:

[0228] For each segmented block R obtained in step S22 ij First, use the edge detection algorithm (such as Canny operator) to extract the outline of the block, that is, the frame part. The edge point set of the frame part is ε ij .

[0229] Then, the segmented block R ij Subtract the frame part ε ij The obtained internal area is the filling pattern part. The pixel value set of the filling pattern is

[0230] Formally, it can be expressed as:

[0231] ε ij =EdgeDetect(R ij )

[0232]

[0233] Among them, EdgeDetect(·) represents the edge detection algorithm, and \ represents the set subtraction operation.

[0234] Finally, record each segmented block R ij The frame coordinate information ε ij and pixel data for the fill pattern Prepare for subsequent storage optimization.

[0235] Step S24: Establish a unified storage structure

[0236] The purpose of this step is to establish a unified storage structure to store the frame, filling content, and spatiotemporal information of each segmented tile in the original video. The specific implementation process is as follows:

[0237] Create a storage block Used to organize the following information:

[0238] 1. The frame coordinates ε of each segmented tile ij

[0239] 2. Fill pattern data for each segmented tile

[0240] 3. The timestamp t of each segmented tile in the original video i

[0241] 4. The spatial coordinates (x ij ,y ij )

[0242] For each segmented block R obtained in step S23 ij , its frame coordinate ε ij , Fill Pattern Timestamp i and spatial coordinates (x ij ,y ij ) information, organized and stored in storage blocks middle.

[0243] Formally, storage blocks It can be expressed as:

[0244]

[0245] To further optimize storage space, you can The data in the data is compressed and encoded. Common compression algorithms include LZW, Huffman coding, etc. You can choose the appropriate compression method according to the actual situation.

[0246] Step S25: Delete duplicate filling content

[0247] The purpose of this step is to further optimize the storage space and delete the repeated padding content in the storage block. The specific implementation process is as follows:

[0248] 1. Traverse the storage block Fill content of all split tiles in Check whether there is duplicate pixel data. You can use data structures such as hash tables or KD trees to quickly find duplicate filling content.

[0249] 2. For repeated filling content, only one copy of the actual data is retained and replaced by a pointer reference in other segmented tiles.

[0250] 3. Update storage blocks The filling content information of the split blocks is replaced by pointer reference instead of the repeated actual data.

[0251] Doing so can significantly reduce storage space usage and improve storage efficiency.

[0252] Step S26-Persistent Storage

[0253] The purpose of this step is to optimize the storage block Persistent storage in the storage device provides data support for subsequent video playback. The specific implementation process is as follows:

[0254] 1. The optimized storage block Stored in high-speed hard disks or distributed storage systems in the form of files or databases.

[0255] 2. Storage blocks Perform backup and off-site disaster recovery to ensure data security and reliability. RAID technology or distributed storage systems can be used to achieve data redundancy.

[0256] 3. As needed, the storage block can be Perform hierarchical management, such as storing by time period or region. For example, time can be divided by year-month-day, and spatial regions can be managed in a grid-like manner.

[0257] In summary, by establishing a unified storage structure Optimizing and persistently storing it can greatly improve the storage efficiency and reliability of monitoring data, and provide a good data foundation for subsequent video playback.

[0258] Step S31: Read the compressed video file from the storage device

[0259] The purpose of this step is to read the previously stored compressed video files from the storage device to prepare for subsequent video reconstruction. The specific implementation process is as follows:

[0260] 1. According to the user's query conditions, such as time period [t s ,t e ]、area(x s ,y s ,x e ,y e ) etc., locate the corresponding storage block from the storage device Data. Hierarchical indexing can be used to quickly locate target data.

[0261] 2. The storage block to be located The data is read into memory, ready for subsequent decoding and reconstruction operations.

[0262] Step S32: Decode the compressed video file

[0263] The purpose of this step is to decode the compressed video file read in step S31 and restore the original segmented blocks. The specific implementation process is as follows:

[0264] 1. The storage block read in step S31 The data is decompressed to restore the frame coordinates ε of each segmented block ij and fill content data

[0265] 2. According to the stored timestamp t i and spatial coordinates (x ij ,y ij ) information, restore each segmented block to the corresponding time and space position, and form a complete set of segmented blocks.

[0266] Step S33: Resynthesize the segmented blocks

[0267] The purpose of this step is to use the segmented block information restored in step S32 to resynthesize a complete video image. The specific implementation process is as follows:

[0268] 1. For each segmented block R ij , according to the stored frame coordinates ε ij and fill content data Regenerate the pixel data of the segmented tile.

[0269] 2. Resynthesize each segmented block R ij , according to the stored timestamp t i and spatial coordinates (x ij ,y ij ) information, spliced ​​to the corresponding position to form a complete video frame I′ i .

[0270] 3. For consecutive video frames {I′ 1 ,I′ 2 ,...,I′ N}Arrange the images in chronological order to complete the reconstruction of the entire video screen.

[0271] Step S34: Splice the segmented blocks in time order

[0272] The purpose of this step is to splice the segmented blocks resynthesized in step S33 in chronological order to restore the complete surveillance video screen. The specific implementation process is as follows:

[0273] 1. Traverse all video frames reconstructed in step S33 {I 1 ′ ,I 2 ′ ,...,I ′ N}, according to the stored timestamp {t 1 ,t 2 ,...,t N}Arrange them in order.

[0274] 2. Splice the arranged video frames in sequence to form a continuous video picture sequence

[0275] 3. For the spliced ​​video sequence Smoothing is performed to eliminate visual discontinuity caused by the splicing of segmented blocks. This can be done using interpolation smoothing algorithms, such as linear interpolation, cubic spline interpolation, etc.

[0276] Step S35: Monitoring video playback

[0277] The purpose of this step is to play back the complete surveillance video restored in step S34 to provide users with visual surveillance information. The specific implementation process is as follows:

[0278] 1. The video image sequence spliced ​​in step S34 Decoded into a playable video format by a video decoder (such as H.264, VP9, ​​etc.).

[0279] 2. Play back the decoded video data in real time through a video player. The player can provide basic control functions such as pause, fast forward, and slow play.

[0280] 3. According to user needs, advanced playback functions such as perspective switching and area zooming can be added to enhance the visualization effect of surveillance videos. For example, users can be allowed to customize the zoom ratio of the area of ​​interest, or automatically switch to the perspective of the relevant camera when a specific event occurs.

[0281] The explanation of relevant variables is shown in Table 1:

[0282] Table 1 Variable explanation table:

[0283]

[0284] Specifically, the principle of the present invention is:

[0285] 1. Semantic segmentation technology based on deep learning. This system uses advanced deep learning models, such as Mask R-CNN and DeepLab, to perform fine-grained semantic segmentation on each frame in the surveillance video and accurately extract each target object (such as personnel, equipment, signs, etc.) in the scene. This lays the foundation for subsequent monitoring information mining and data storage optimization.

[0286] 2. Hierarchical storage structure based on frame-fill content. In order to efficiently store the huge amount of surveillance video data, this system adopts an innovative hierarchical storage scheme. First, each semantic segmentation area is divided into two parts: the outline (frame) and the internal filling pattern. This structured representation greatly reduces the storage space occupied. Secondly, by identifying and deleting repeated filling content, the storage efficiency is further optimized. Finally, this structured segmentation tile is organized and stored according to the spatiotemporal information, which is convenient for subsequent rapid retrieval and playback.

[0287] 3. Hierarchical indexing based on time and space. In order to support efficient monitoring data query and playback, this system adopts a hierarchical indexing mechanism in both time and space dimensions. In the time dimension, the monitoring data is hierarchically managed in the form of year-month-day; in the space dimension, the venue area is divided in a grid-like manner. This multi-dimensional indexing structure can quickly locate the monitoring data that users are interested in, greatly improving the response speed of the system.

[0288] 4. Modular system design. This system adopts a modular design concept. Each functional module (such as temperature and humidity monitoring, air quality monitoring, video monitoring, etc.) is relatively independent and has good scalability. This not only facilitates the flexible configuration of the system and future functional expansion, but also facilitates seamless integration with other subsystems of the exhibition venue to achieve comprehensive intelligent management.

[0289] Through the above-mentioned technological innovations, the large-scale all-round intelligent monitoring system for exhibition data of the present invention can efficiently collect, store and analyze various monitoring data, and provide more intelligent and refined support for exhibition venue management.

[0290] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0291] The following is a specific embodiment of the present invention: Take a large international exhibition center as an example. The exhibition center covers an area of ​​100,000 square meters and has multiple large exhibition halls and conference rooms. In order to achieve all-round intelligent monitoring of the exhibition center, the large-scale exhibition data all-round intelligent monitoring system proposed by the present invention is deployed. The specific implementation process is as follows:

[0292] 1. Monitoring data collection

[0293] First, more than 200 high-definition surveillance cameras were deployed at key locations in the exhibition center, covering the main areas of the venue. These cameras have the ability to capture panoramic video data of 180 degrees or even 360 degrees. In order to ensure the time synchronization between the various video data, the NTP network time synchronization protocol was used to uniformly calibrate the timestamps of all cameras.

[0294] In terms of video encoding and compression, the H.265 encoding standard is adopted, and differentiated compression parameters are set for different areas. For areas with dense personnel activities and key monitoring areas, higher image quality is retained; for some secondary areas, the bit rate is appropriately reduced to save storage space. After encoding and compression, the storage space of each monitoring video is reduced by about 70%.

[0295] 2. Monitoring data storage

[0296] In order to effectively manage massive surveillance video data, an optimization solution based on semantic segmentation and structured storage is designed. The specific steps are as follows:

[0297] 1) Semantic segmentation: The Mask R-CNN model based on deep learning is used to perform fine-grained semantic segmentation on each frame of surveillance video and extract each target object (such as people, equipment, signs, etc.) in the scene. After segmentation, each frame of video is divided into multiple semantic regions.

[0298] 2) Frame-fill content splitting: For each semantic segmentation region, it is further split into two parts: the outline (frame) and the internal filling pattern. This structured representation method can not only significantly reduce the storage space occupied, but also lay the foundation for subsequent data compression and fast retrieval.

[0299] 3) Deletion of duplicate content: When storing the filling patterns of the semantically segmented areas, we carefully compared the areas and found that there was a lot of duplicate pixel data. Therefore, we used a hash table to only retain the unique filling content, and stored the other areas using pointer references. This further optimized the storage efficiency.

[0300] 4) Spatiotemporal information recording: In addition to the frame coordinates and filling content, the timestamp and spatial position coordinates of each semantic segmentation area in the original video are also recorded. These spatiotemporal information provide support for subsequent video playback and data analysis.

[0301] 5) Hierarchical storage: Finally, the structured semantic segmentation data is stored hierarchically according to the time dimension (year-month-day) and the spatial dimension (grid). This hierarchical indexing mechanism can greatly improve the efficiency of data query and positioning.

[0302] Through the above optimization measures, the storage cost of surveillance video data has been successfully reduced by more than 80%. Taking a 6-hour surveillance video of a day as an example, the traditional method requires about 2TB of storage space, while the solution of the present invention only requires about 400GB.

[0303] 3. Monitoring data analysis

[0304] In addition to efficient data collection and storage, this system also has powerful monitoring data analysis capabilities. Using the aforementioned semantic segmentation technology, various types of targets detected in the monitoring video are deeply analyzed to extract rich monitoring information.

[0305] 1) Personnel activity trajectory analysis: By tracking the personnel targets detected in the surveillance video, the activity trajectory of the participants at the exhibition site can be reconstructed. Combined with time information, the flow of people in each exhibition area and conference room can be analyzed to provide valuable decision support for venue managers.

[0306] Taking a large machinery exhibition as an example, the distribution of the length of time participants stayed in each exhibition area from 10:00 am to 12:00 pm on the first day was counted. Figure 1 As shown in the figure, it can be seen that the participants like to stay in exhibition area C and exhibition area F the most. In contrast, the flow of people in exhibition area A and exhibition area B is relatively small. This provides a basis for exhibitors to adjust the layout of exhibits and the arrangement of on-site activities.

[0307] 2) Hotspot area identification: In addition to the activity trajectories of people, the hotspot areas of the exhibition site can also be identified based on the semantic segmentation results. These hotspot areas are usually areas with high population density and strong interactive activities, which are of great significance for optimizing the on-site layout and increasing the interactive experience.

[0308] Taking the same machinery exhibition as an example, we calculated the heat map of personnel distribution throughout the day. Figure 2 As shown in the heat map, it can be clearly seen that exhibition areas C and F have always been the most popular areas, while exhibition areas A and B are relatively deserted. This provides a basis for the organizer to adjust the layout of the booths and increase hot spot interactions.

[0309] 3) Abnormal event monitoring: In addition to conventional crowd flow analysis, the system can also automatically monitor abnormal events at the exhibition site. For example, when a large number of people suddenly gather in the monitoring screen and there are obvious signs of riots, the system will promptly issue an alarm to remind the on-site staff to deal with it in time. At the same time, according to the analysis results, the specific area and time point of the incident can be located, providing important basis for subsequent investigation and evidence collection.

[0310] In general, with the help of advanced semantic segmentation and video analysis technology, it is not only possible to efficiently collect and store monitoring data, but also to extract rich monitoring information from it, providing strong support for the intelligent management of the exhibition center. This can not only improve the overall operational efficiency of exhibition activities, but also create a safer and more comfortable environment for participants.

[0311] 4. Surveillance video playback

[0312] In addition to real-time monitoring, the system also supports playback and retrieval of historical monitoring data. Based on the aforementioned hierarchical storage structure, it is possible to quickly locate the monitoring video clips that users are interested in.

[0313] Take a participant as an example. The staff needs to check the activity track of the participant at the exhibition site. Just input the characteristics of the participant (such as photos, clothing characteristics, etc.), and the system can quickly locate the relevant surveillance video clips based on the semantic segmentation results, and restore them to complete video images for playback. The entire query process only takes about 10 seconds, greatly improving work efficiency.

[0314] During the video playback process, users are also provided with a wealth of interactive functions. For example, the viewing angle of the surveillance camera can be switched at will, and the area of ​​interest can be zoomed in for detailed observation; the playback speed can also be adjusted to carefully analyze the behavior trajectory of the participants. The provision of these functions further enhances the user experience of the system.

Claims

1. A comprehensive intelligent monitoring system for large-scale exhibition data, characterized in that: Includes the following functional modules: Temperature and humidity monitoring module, used to monitor the temperature and humidity data of the exhibition venue in real time, automatically adjust the temperature and humidity, and maintain the best environmental conditions; The air quality monitoring module is used to analyze the air quality in the exhibition venue, detect abnormal conditions in time, and link air purification equipment to ensure the air quality at the exhibition site; Light monitoring module, used to monitor the operating status of the lighting system in the exhibition venue and detect faults in time; Energy consumption monitoring module, used for real-time energy consumption data of various equipment in the exhibition venue; The video monitoring module includes a monitoring submodule, a split storage submodule and a playback submodule; wherein the monitoring submodule is used to collect monitoring videos in the exhibition venue in all directions; the split storage submodule uses a frame-filling content splitting method to store the monitoring video as split compressed video files, and the playback submodule is used to merge the split compressed video files into monitoring images and play them sequentially during playback; the split storage submodule is used to perform the following steps: S21, obtaining a video file block and extracting a semantic key frame, wherein the semantic key frame is a frame containing the most comprehensive semantic segmentation objects among a plurality of adjacent frames; S22, performing semantic segmentation on each frame in the video file block to obtain a plurality of segmented image blocks; S23, extracting a frame from each segmented block, and splitting the frame into a frame and a filling pattern; S24, establishing a storage block for storing the frame of each segmented image block and each filling content, as well as the timestamp and position of the segmented image block in the original video; S25, deleting repeated filling content; S26: Store the storage block in a storage device.

2. According to claim 1, a large-scale exhibition data all-round intelligent monitoring system is characterized in that: The monitoring submodule is used to perform the following steps: S11. Collect panoramic video data of the exhibition venue, including surveillance video information from different angles and locations; S12, performing time synchronization and spatial splicing on the collected video data to form complete panoramic video monitoring information; S13, encoding and compressing the stitched panoramic video to obtain video file blocks for storage, and sending each video file block to the split storage submodule.

3. The all-round intelligent monitoring system for large-scale exhibition data according to claim 1 is characterized in that: The playback submodule is used to perform the following steps: S31, reading the stored compressed video file from the storage device; S32, decoding the read compressed video file to restore the original segmented blocks; S33, resynthesizing each segmented image block according to the stored frame and filling content information; S34, splicing the resynthesized segmented image blocks in chronological order to restore a complete surveillance video image; S35, playing back and displaying the restored surveillance video images.

4. The all-round intelligent monitoring system for large-scale exhibition data according to claim 3 is characterized in that: The step S22 comprises: Step 1) Using the semantic segmentation algorithm in step S21, semantic segmentation is performed on each frame of the video file block; Step 2) According to the segmentation results, each frame image is divided into multiple semantic segmentation blocks, each block corresponds to a semantic object in the scene; Step 3) Number each segmented block and record its position coordinates and timestamp information in the original video frame.

5. The all-round intelligent monitoring system for large-scale exhibition data according to claim 3 is characterized in that: The step S23 comprises: Step 1) performing edge detection on each segmented image block obtained in step S22 to extract the contour line of the image block, i.e., the frame part; Step 2) The inner area obtained by subtracting the frame part from the segmented block is the filling pattern part; Step 3) Record the frame coordinate information of each segmented block and the pixel data of the filling pattern.

6. The all-round intelligent monitoring system for large-scale exhibition data according to claim 3 is characterized in that: The step S24 comprises: Step 1) creating a storage block for storing the following information: frame coordinates of each segmented tile, fill pattern data of each segmented tile, timestamp of each segmented tile in the original video, and spatial coordinates of each segmented tile in the original video; Step 2) storing the above information of each segmented block in the storage block according to a certain data structure; Step 3) Use LZW or Huffman coding algorithm to compress the data in the storage block.

7. The all-round intelligent monitoring system for large-scale exhibition data according to claim 3 is characterized in that: The step S25 comprises: Step 1) traverse the filling contents of all segmented blocks in the storage block to check whether there is repeated pixel data; Step 2) For repeated filling content, only one copy of the actual data is retained and replaced by a pointer reference in other segmented blocks; Step 3) Update the filling content information of the split tiles in the storage block, and replace the repeated actual data with pointer references.

8. The all-round intelligent monitoring system for large-scale exhibition data according to claim 3 is characterized in that: The step S26 comprises: Step 1) storing the optimized storage blocks in the form of files or databases in a high-speed hard disk or distributed storage system; Step 2) Back up the storage blocks and perform off-site disaster recovery to ensure data security and reliability.

9. The all-round intelligent monitoring system for large-scale exhibition data according to claim 4 is characterized in that: The semantic segmentation algorithm adopts a deep learning-based method.

Images