Video coding with dynamic background

Manoranjan Paul,Weisi Lin,Bu-Sung Lee,Chiew Tong Lau

doi:10.1186/1687-6180-2013-11

Abstract

Motion estimation (ME) and motion compensation (MC) using variable block size, sub-pixel search, and multiple reference frames (MRFs) are the major reasons for improved coding performance of the H.264 video coding standard over other contemporary coding standards. The concept of MRFs is suitable for repetitive motion, uncovered background, non-integer pixel displacement, lighting change, etc. The requirement of index codes of the reference frames, computational time in ME & MC, and memory buffer for coded frames limits the number of reference frames used in practical applications. In typical video sequences, the previous frame is used as a reference frame with 68–92% of cases. In this article, we propose a new video coding method using a reference frame [i.e., the most common frame in scene (McFIS)] generated by dynamic background modeling. McFIS is more effective in terms of rate-distortion and computational time performance compared to the MRFs techniques. It has also inherent capability of scene change detection (SCD) for adaptive group of picture (GOP) size determination. As a result, we integrate SCD (for GOP determination) with reference frame generation. The experimental results show that the proposed coding scheme outperforms the H.264 video coding with five reference frames and the two relevant state-of-the-art algorithms by 0.5–2.0 dB with less computational time.

Highlights

The H.264/AVC video coding standard improves ratedistortion performance significantly compared to its predecessors and competitors by introducing a number of innovative ideas in Intra- and Inter-frame coding [1,2,3]
We found that the absolute different matrix between ground-truth background (GTB) and the most common frame in scene (McFIS) generated by traditional DBM (tDBM) has maximum 135 and average 5.3 values, whereas the counterpart generated by the McFIS by proposed DBM (pDBM) has 125 and 4.8
For scene change detection (SCD) using McFIS, we find the sum of absolute difference (SAD) between the McFIS and the current frame

Summary

Introduction

The H.264/AVC video coding standard improves ratedistortion performance significantly compared to its predecessors and competitors by introducing a number of innovative ideas in Intra- and Inter-frame coding [1,2,3]. The above-mentioned AGOP and/or SCD techniques require comparison between the current frame and a number of previous frames for better rate-distortion performance. We generate a most common frame in scene (McFIS), for SCD and AGOP, and as an effective reference frame for better rate-distortion performance in coding. Using McFIS as a reference frame we have the following advantages compared to the existing methods based on MRFs and SCD for AGOP: Only one McFIS is used instead of a number of reference frames so the overheads of index codes are reduced. Comprehensive analysis and simulation results [on computational time, SCD, amount of referencing based on dynamic background (i.e., McFIS), and rate-distortion performance] are provided to understand the effectiveness of McFIS in video coding. The overall experimental set up and results for the proposed scheme are presented in Section “Overall experimental results”, while Section “Conclusions” concludes the article

GMM-based DBM

Methods

Findings

Conclusions