From a practical point of view, the current mature video analysis and processing technologies, including border guard, motion detection, text (license plate) recognition, applications and more focused on the security level of traffic monitoring and management, home security, as well as banks, prisons and other more high places. With the recent things, smart city and other construction projects in the country flourished, monitoring systems are entering the era of massive amounts of data, to improve efficiency and business value of video resources, we must use more advanced intelligent video analysis technology for in-depth the video data mining, detection and management. Whether traditional security units, or commercial application areas, all of the intelligent video analysis technology generated a lot of demand. At present, the development of intelligent video analysis program has made great progress, but because of the complexity of its own research,making diverse research methods and tools, algorithms, high complexity and limited scope, there is no universality, accuracy, speed meets common method of demand. Networking and distributed processing requirements of video surveillance systems, as well as restrictions on large-scale project installation cost, size and power consumption, making the computing power and bandwidth are rising in embedded processors become the mainstream choice of video surveillance systems . In this trend, at this stage of the development of intelligent video analysis technology focus is to improve the algorithm level, reducing the false positive rate, with more cost-effective processor platform, curing algorithms to hardware accelerator, leaving the processor MIPS to be more flexible software for operation, thereby enhancing the reliability and usefulness of intelligent video analysis and processing. All along, intelligent video analysis and processing are generally based on the DSP chip architectures under the front or the back-end platform for surveillance cameras, and now is moving in the direction of the ARM hardware platform evolution. Advantages of ARM embedded system is low power consumption, the price of equipment programs are cheaper, with the ARM embedded device applications market share increased year by year, the whole industry chain has basically taken shape, a growing number of security vendors began using ARM processor development of new video surveillance camera products. Of course, to carry out a variety of complex video analysis algorithms on a single ARM platform, its difficult to imagine. Intelligent video analysis to use a lot of CPU resources, but because of their need to deal with a lot of work already, which makes the same ARM platform is often difficult to provide sufficient space for the operation of intelligent video algorithms, in order to achieve fluency in the ARM platform, accurate video it is difficult analysis. The way to solve this problem is still the further optimization of video analysis algorithms. Currently intelligent video analysis technology development focus lies on how you can reduce the interference of the video screen to improve the accuracy of identification systems analysis. In addition to the special analysis and processing for the video data for intelligent analysis and processing sound more and more people's attention. In the past, surveillance cameras is received with the pickup by detecting whether the audio signal reaches the threshold value to start the alarm mode. Today, users want to be able to carry out a more refined sound judgment, extracted from the audio signal in the hazardous characteristic, and integrate with video surveillance, early warning and improve the overall capacity of the system. If intelligent video analysis is intelligent monitoring 'eye', then intelligent audio analysis is intelligent monitoring 'ears'. Intelligent video surveillance total solution sound intelligent voice processing solutions will be integrated into HD network camera solutions, in addition to the traditional definition monitor, the system can identify and locate remote sound monitoring and feature recognition, and intelligent sound source. Using this technology, intelligent monitoring system to capture ambient sounds, when unexpected events occur, the program analyzes the sound source can be identified and implemented to determine the orientation, the main platform for the integration of the camera, the camera may need to be automatically converted to the sound source direction and angle monitoring. In the positioning of the smart sound source is simulated using two human binaural microphone function arrival time difference between the effects of different microphone based on the sound, the sound source localization is achieved. In the course of an analysis of the voice recognition, the biggest difficulty comes from how well the background noise suppression, improve the accuracy of the voice feature extraction. The noise may sound indoor environment including noise fan, air conditioning issue, as well as outdoor construction sites mixer, on the road automobile produced by the primary task of the system is possible to reduce the impact of noise arising from the above, which accurately extract the broken glass, children crying, and so the risk of a sound signal, and immediately turned the camera monitoring system.
Related Articles -
intelligent video analysis, surveillance camera, HD network camera,
|