Cross Search Frequency Domain Motion Estimation Algorithm for the High Efficiency Video Coding Standard

Development in video coding technology is at the heart of recent rapid advances in streamed and live digital video technologies. This paper proposes the implementation of the Large Cross Search Algorithm in the H.265/High Efficiency Video Coding motion estimation process, a technique common to most hybrid block-based video encoders. Our speed optimization through frequency domain phase-correlation enables the user to compress a video in a short time while maintaining high image quality with bit rates similar to the standard implementation. The implementation was tested with preliminary results on common video sequences.

I. Introduction

In recent years, video coding standards groups have built on prior methods accumulated through MPEG, H.263, and H.264, attempting to further eliminate some redundancy in data from encoded video sequences [2]. These standards had some limitations such as having small encoding blocks. In January 2013, the latest video coding standard for video compressing techniques was created, the H.265 codec [3], [4]. H.265 is known as HEVC which stands for High Efficiency Video Coding.