Algorithm and Architecture for Adaptive Motion Estimation in Video Processing

摘要

This paper introduces a block-based motion estimation algorithm based on projection with Adaptive Window Size Selection (AWSS) along with its architecture. Motion field of pixel is either horizontal or vertical, this paper assume horizontal and all the processing is done based on it. This projection method is combined with AWSS in which appropriate search window for each block is determined on the basis of motion vectors and prediction errors obtained for the previous block, which makes this novel method several times faster than exhaustive search with negligible performance degradation. Encoding QCIF-size video by the proposed method results in reduction of computational complexity of motion estimation by roughly 45% and overall encoding by 23%, while maintaining image/video quality. A hardware architecture for the adaptive motion estimation algorithm was also done in this paper, which has several features such as low latency less area and more speed which is applicable for 30 frames per sec with frame size of QCIF (176 × 144) which has a small chip area of (3.57 × 3.57 mm~2) and operating frequency of 83 MHz 1.8 v, 195 mW power dissipation designed for video coding, with UMC 0.18 μm 1P6M cell library.