Research On The Key Technologies Of DSP-Based Intelligent Video Surveillance

With the developments of electronics and information technologies, more and morecountries and governments begin to pay attention to the researches and applications ofcomputer vision technology. In recent years, as an emerging area of computer vision，intelligent video surveillance technologies have been widely used in security equipment,intelligent traffic, unattended surveillance and so on. They have greatly promted thedevelopments of intelligent video surveillance systems.Although a lot of efforts have been done by researchers and organizations，theycan’t meet the needs of people for intelligent video surveillance systems. Nowdays，more and more public infrastructure have been built.Some video surveillance systemshave to be equipt to detect missing object in such places. The existing methods formissing object detection have two shortcomings：They are not well adapted to theenvironment. They are easily influenced by the change of environments. The operationis very complex,and it is difficult to achieve real-time surveillance. Aiming at tacklingthe above problems, the main contributions include：(1) A new algorithm is designed to recognition the missing object with simpleoperations，less hardware resources，and the strong adaptability to the changeg ofenvironments. Then, the algorithm is simulated in computer.(2) In order to realize the algorithm on the DSP，a master control program based onmulti-threading mechanism under Linux system was designed to process the video datacollection, transmission and display.Then the communications program based onDSPLink driver was designed.(3) According to the characteristics of the C64+DSP processor，the algorithm wastransplanted to DSP and optimized effectively.Therefore the resources costs of thesystem were greatly reduced and the system resources were allocated reasonably. Theoverall performance of the system were enhanced effectively.