Since the past few years, security manufacturers have begun to emphasize video structure. Through the evolution of algorithms and technological innovation, video structure has begun to be applied on a large scale. So, what is video structure?
Video structure is a technology for extracting video content information. It adopts time-space segmentation according to semantic relationships of video content,Feature Extraction, object recognition and other processing methods are organized into technology that can be understood by computers and people. From the perspective of data processing process, video structured description technology can convert surveillance video into information that can be understood by humans and machines, and further transform it into intelligence used in public security operations, realizing the transformation of video data into information and intelligence.
Why do we need video structure
Video structured technology firmly grasps the main line of video content information processing and network sharing application. The industry strives to fully realize the intelligence of surveillance video information after several years of technical research and system construction.Video surveillanceThe intelligence of the Internet strengthens the universality of police video applications. That is, it realizes video information processing and analysis mainly based on automatic processing by machines, and transforms it into information available for public security work through technical means; realizes information sharing and active interoperability between monitoring networks, terminals, and police types, and realizes network functions such as active monitoring and automatic network analysis; comprehensively expands the application mode of video in police work, greatly improves the ease of use of technology, and realizes flexible, simple and diverse video on-demand service applications centered on business police officers anytime, anywhere.
In addition to the public security industry, the application scenarios of video structured technology are also inSmart transportationExpand. At present, the application needs and frequency of electric police checkpoints in image investigation have long surpassed that of traffic police, because cases basically have to be connected with vehicles, which can find many clues. The snap angle of the vehicle of the checkpoint is relatively fixed, and corresponding vehicle feature recognition technology can be developed. The electric warning lockpoint is a good matching point for business needs and technical implementation. This is the application reserve of video structure. Some domestic manufacturers have developed cameras that can break through the limitations of plane image characteristics and obtain more accurate three-dimensional information, such as the number of human bodies, height, object length, etc.
Video structured technology is widely used
In terms of the content of the video structured description, the video information that public safety focuses on is mainly: personnel, vehicles, and behavior. In the video, the person is presented as a descriptive individual, including the precise positioning of the person's face, the extraction of facial features, facial features comparison, the gender, age range, approximate height, hair accessories, clothing, object carrying, and walking shape of the person; the description information for the vehicle includes: license plate, car color, car model, brand, sub-brand, car sticker, car accessories information and other vehicle description information; the description information for behavior includes: cross-border, area, wandering, legacy, gathering and other behavior description information.
After the video is structured, the following purposes can be achieved:
First of all, the video search speed has been greatly improved. After the video is structured, search for the suspect pedestrian target on a screenshot from a million-level target library (corresponding to high-definition videos of hundreds to one thousand hours), and complete it in a few seconds; search for the library of tens of million-level target library, and complete it in a few minutes (if cloudization is achieved, the speed will be faster). Performing search queries on a structured basis can solve the problem of fast target search.
Secondly, the storage capacity is greatly reduced. After structured video, the structured search information and target data of the person stored in the video are less than 2% of the video data volume; for vehicles, less than 1%; for behaviors, it is more reduced. The storage capacity is greatly reduced, which can solve the problem of long-term video storage.
Finally, video structure can activate video data, which can be used asData MiningBasics: After the video is structured, it is stored in the corresponding structure.Data Warehouse, in-depth data mining can be carried out on various types of data warehouses, give full play to the role of big data, enhance the application value of video data, and improve the analysis and prediction functions of video data.
Basic requirements for video structured analysis platform
The system has a cloud computing architecture, and the deployment should adopt centralized processing and hierarchical review methods, multi-level cascade, automatic allocation of dynamic structured processing services, balancing loads of multiple nodes of the system, and can be handled during major and major events.parallelNetwork service computing capabilities and cluster capabilities;
The client should support C/S architecture, and a single node device has at least 20 users concurrent access;
Backend management should support B/S architecture and have functions such as permissions, users, structured task settings, and comprehensive network management.
Have the ability to obtain video streams and video files from the hard disk recorder;
Have local video files uploaded by users;
It has unified transcoding of uploaded files to satisfy online on-demand broadcast;
Have structured uploaded filesVideo synthesisPlay and be able to play backtrack the original video based on the event tag of the structured video.
It can perform structured analysis of uploaded video files or video streams, meet the extraction of feature information of activity events in the video, and then structured storage is performed after the time stamp is associated with the video file;
It has the ability to search the content of the processed file, search and filter according to the running direction, activity range, color value, size and other characteristics of the activity events in the video; it also has the ability to search and filter all video time and location conditions.
It has real-time operation detection of each processing unit in the platform, provides a comprehensive network management module, and can visually display the resource utilization of various service modules and equipment on the platform through charts;
It has permission control function, and controls the user's operable range by assigning corresponding roles to different users.
In short, with big data andDeep LearningWith the development of video structure, video structure will definitely provide strong support for the intelligence of video data, thus making active prevention and control systems possible.