Research On Preprocessing Technique Of Real-time Video Encoding

Video coding is the key technology of the multimedia application. Video preprocessing is one of the important parts of the video coding, which process the raw video sample for the sake of encoder.Aiming at the two questions, video deinterlacing and video denoising, which are the key parts of video preprocession, this paper carries out the in-depth study . with new ideas to the two questions, the high-powered algorithms are put forward.Chapter 2 is aimed at the study of video deinterlacing. First, it explains the interlacing phenomenon and the principle of deinterlacing. Second, it describes some normal traditional algorithm of video deinterlacing. Last, it brings forward the high-powered realtime algorithm in detail, which is suitable to the DSP application, and this algorithm's improvement to traditional algorithm is presented.Chapter 3 is aimed at the study of video denoising. First, it introduces the background and the normal categorization of video denoising technology in concise. Second, it describes some normal traditional algorithm of video denoising. Last, it brings forward the high-powered algorithm in detail, which is recommended in this paper, and highlights this algorithm's advantage relative to traditional algorithm.The video deinterlacing algorithm, recommended in this paper, is based on the motion detection and nonlinear interpolation. Most complicated image and complicated motion, which affect the performance of video deinterlacing, have been taken into consideration, so the recommended algorithm is suitable to most scene's application.The video deinterlacing algorithm, recommended in this paper, is a robust algorithm based on motion compensation and self-adapting spatiotemporal filter. In the procedure of video denoising, the recommended algorithm makes the best of related data within frame and between frames, so it keeps considerable image detail while getting rid of random noise. The recommended algorithm gets high performance.The algorithms proposed have been implemented on EQUATOR platform, and have been applied in the real-time video coding.