Motion Vector Re-estimation for Transcoding using Kalman Filter

Autor: Jen-Chun Lin, 林仁俊
Rok vydání: 2007
Druh dokumentu: 學位論文 ; thesis
Popis: 95
Recently, digital multimedia products have been fast developed, which changes our life. Visual devices are applied to mobile phone, PDA, or other products for different requirements. These devices can receive real-time videos from wireless network or internet. In order to watch video programs, the video frame sizes must be adaptively changed for each device. Although many fast down scaling algorithms have been proposed, such as SA, AWA MDN, their performances can not achieve high visual quality. In this thesis, we proposed a method to solve this problem. We consider the motion correlation of neighbor blocks in temporal and spatial domain, and embed Kalman filter to traditional fast down scaling algorithms. We adopt linear interpolation to reconstruct float pixel blocks for motion compensation, which improves visual quality of videos. Since the computational complexity of Kalman filter is low, the proposed method can be used in real-time systems.
Databáze: Networked Digital Library of Theses & Dissertations