The Research And Design Of Digital Image Stabilization Technology Base On DM6446

Along with the video processing technology is used more and more widely, electronic image stabilization (EIS) technology as a video processing technology, it has wide application prospects. In our daily life, unstable video will cause the observer's visual fatigue, and in the military would cause monitoring system of the leak and false alarm etc, so the quality of the video quality is not only related to people's life, but also related to national security.This paper is mainly divided into two parts. At the beginning of the first part, we introduced the electronic image stabilization technology classification, and analyzed the history of the electronic image stabilization technology and the DSP technology. Second, through the data access, we studied the development of electronic image stabilization at home and abroad, and introduced its principles and mainly algorithms; meanwhile we made clear the process of electronic image stabilization. The electronic image stabilization is mainly processing digital images, so the electronic image stabilization is also called the digital image stabilization (DIS). The digital image stabilization is mainly composed of four parts:image preprocessing, motion estimation, vectors filter and motion compensation. Third, we introduced the mathematical model of image movement, throughout the movement mathematics model, according to the different video features, we can choose different sports model, so as to reduce the complexity of the algorithm. Fourth, we mainly researched two classic algorithms of digital image stabilization:gray projection algorithm (PA) and representative point matching algorithm (RPM). Besides of researching the digital image stabilization algorithms, we also introduced the general method of image preprocessing and the image edge processing methods. Finally, we proposed a new digital image stabilization algorithm, which combined the gray projection algorithm and representative point matching algorithm. Compared with gray projection algorithm, the new algorithm improved the image stabilized precision; meanwhile improved the image processing speed more than representative point matching algorithm; it can completely meet the real-time processing requirements.Another part of the paper introduced the digital image stabilization system hardware realize part. In this part, we first simply introduced the DSP TMS320DM6446 (called DM6446), which was produced by TI Company. The DM6446 is DSP+ARM dual-core structure. We respectively introduced the DSP subsystem, ARM subsystem and video process subsystem. Second, we introduced the digital image stabilization system hardware realized process in DM6446 platform. ARM controlled the whole system, DSP was responsible for the digital image stabilization algorithm processing, and both of the two cores could exchange information through the interrupt mode and shared memory. Finally, we completed program and program optimization, and achieved the images collection from camera, digital image stabilization processing, at last outputted the stabilized video in the LCD.