Five New Trends in Intelligent Video Surveillance Analytics

With the rapid improvement of monitoring clarity and the rapid rise of storage devices, intelligent analysis has become the most effective way to deal with current back-end equipment conflicts. 1. Smart Streaming Technology The intelligent code stream is that after the system recognizes the image, it adjusts the stream of the normally recorded video according to the motion speed of the moving subject in the screen, and stops the final video storage. For slow moving subjects, the code stream can be recorded as low as 8 frames per second (fps) or less. For subjects moving at normal speed, the code stream is set to a normal 25 fps, and the video code stream of moving objects with sensitive images can be set to Set at 30fps or more. Regarding high-speed moving subjects, with the cooperation of high-speed cameras, the code stream can be as high as 1000fps or more. The intelligent stream technology can reduce the occupation of video storage resources by non-sensitive images, thus providing sufficient recording capacity for sensitive images. 2. Intelligent monitoring technology of dynamic region self-adaptation Taking the video surveillance screen of a garage as an example, the sensitive information on the screen is moving vehicles and people. The insensitive picture is the background (air and roof). But in the actual video recording, the background (air and roof) occupies more than 50% of the storage space. Through the image recognition technology, it is possible to determine the fixed background and the moving object image, so it is possible to have the technical possibility of moving only in the recorded picture. Dynamic area intelligent monitoring is to record only moving objects other than the background, which greatly reduces the demand for video storage space. The same storage space, the length of video time that can be saved can be doubled. The camera system, through the image recognition technology, can intelligently learn and identify which images are the background. Even the camera controlled by the PTZ can automatically determine the new background in the recorded video after automatic learning within a set period of time. Intelligent monitoring of images of moving objects. 3. Low-code stream recording technology in non-sensitive areas The 25% to 30% area above the video, usually the top image of the sky or building, is simply an area where sensitive features are unlikely to be present. For example, the partial image of the red mosaic in some typical surveillance images does not contain the video content that people care about at all, so the negligence of the insensitive area of ​​the video image or the low code stream can be recorded separately. High-bit-stream video to do a synchronization. It is worth clarifying that the non-sensitive areas are different in different application scenarios. Some may be below the video image, and in some monitoring scenarios, the non-sensitive areas are irregular. The situation then stops the manual setting at the application level. 4. Face/vehicle recognition (or other sensitive moving objects) drives HD video recording technology In special scenes, such as the entrance of a building, the center of an elevator, etc., the face is a sensitive image. Vehicles and their number plates are sensitive images in garages and at the entrances and exits of the community. If all high-definition video is used, it can meet the monitoring needs, but video storage, especially long-term storage, will require a large amount of storage space; if the image recognition technology is used to determine the time when the set sensitive image is displayed, the camera is driven to start. For high-definition recording, for ordinary non-sensitive images, start standard-definition or even low-bit rate video streams to record. In this way, high-definition and standard-definition monitoring records are separated, which ensures the quality of recorded sensitive images, and at the same time reduces the video storage capacity to a large extent. 5. Distributed storage technology of sequence frame video files The frames generated by the video within one second are identified as sequence frames, and the storage and playback sequences are compiled at the same time, and the frames of different sequences are divided into several files for storage; a single frame sequence file can be played alone, and the effect is equivalent to the video recorded by the low-code stream Effect. When all frame sequences can be synthesized into a complete video and played together, it is a high-definition (or standard-definition) video effect. When the storage space needs to be reclaimed, the area occupied by a partial sequence frame video file can be covered according to the storage strategy plan. The other part is kept for more efficient use of storage space. For example, only one month of video data can be saved according to the original storage. After the video frame files are distributed and stored, the file data of ted sequence frame videos can be saved for several months. Partial masking of sequence frame video files that have been saved in the medium and long term is stopped, and videos that have completed fading are gradually discarded. The long-term saved video data does not disappear completely, but gradually disappears and discards. This maximizes the time that surveillance video is saved. Video streaming of intelligent video surveillance based on the above five technologies The video stream of intelligent monitoring is recorded and stored after a series of image recognition at different logical levels, and after the intelligent recognition is stopped. Among them, the application of dynamic area adaptive intelligent monitoring technology and non-sensitive low-code stream recording technology can be used throughout the entire storage process according to practical needs; or it can be tively implemented according to manual settings at the application level. Image recognition from preliminary recognition to precise recognition, according to the needs of different levels. For the sampling frequency of video image recognition, the specific frequency can be set according to the requirements, and it is not necessary to recognize the image of each frame.