The Video Image Processing Key Algorithms And Intelligent Video Surveillance System

Video & image post-processing is the key technique in video & image applications. This technique can be used to improve the quality, extract the information and make data transformation of the video & image for particular purpose. This thesis studied the status and developing trend of video & image post-processing and focused on several key techniques in this area, such as image scaling, gamma correction and so on. The conventional algorithms were studied in this thesis and the innovations and their corresponding hardware implementations were also proposed. This thesis proposed a scaling architecture without frame buffer, where the bi-cubic algorithm, free zooming function and error compensation were implemented by only 5 pieces of line buffer; a new gamma correction method based on edge directed error diffusion was also proposed to improve the gamma correction precision of conventional methods. The basic principles and algorithms of a new application of the video & image technique in the safe guarding area, named intelligent video surveillance, were studied in this thesis. A feasible system architecture and its detailed implementation were also involved. We implemented the proposed architecture on FPGA and the system was able to work in real time to detect some special behaviors in the surveillance regions.