A Fast Mode Decision Algorithm Based on
Temporal-Spatial Correlation for P-Frame
Video Coding*
Zhigang Yang, Lingyong Xu, and Tao Chen
College of Information and Communication Engineering,
Harbin Engineering University
Harbin, 150001, China
E-mail: {zgyang, xulingyong, chentao}@hrbeu.edu.cn
Abstract—In this paper, a fast mode decision algorithm based
on temporal-spatial correlation for P-frame coding are proposed
to accelerate the video encoding procedure. The algorithm
includes three strategies, mode prejudgment based on temporal-
spatial neighborhood macroblocks (MBs), mode combination
based on motion vector (MV) coherence, and temporal-spatial
correlation analysis for intra mode. Firstly, the left and up MBs
in current frame and the current-position and right MBs in
previous frame are used to predict the mode of current MB.
Then, adjacent blocks in sub-type with similar MVs can be
combined into a block in macro-type. Furthermore, intra mode
in P-frame coding is limited by the temporal-spatial correlation
analysis. Finally, the performance of the proposed scheme is
verified through comparative experimental results using JM
reference software.
Keywords—fast mode decision, temporal-spatial correlation,
video coding
I. INTRODUCTION
The transform and motion compensation hybrid coding
framework is commonly adopted in the state-of-the-art video
coding standards like H.264/AVC [1]. The variable block-size
motion compensation is a key technique in this coding scheme.
Full inter modes including SKIP, 16×16, 16×8, 8×16, 8×8, 8×4,
4×8, and 4×4 blocks, then followed by intra mode are usually
used in P-frame coding to obtain a high compression efficiency
[2], as shown in Fig. 1. These structures cause the
computational complexity of full-mode motion estimation/
compensation is much higher than other video coding modules.
Many kinds of fast inter-mode decision schemes have been
reported [3]-[6]. For instance, a MB tracking-based fast mode
decision algorithm is proposed in [3]. The information of
adjacent MBs is used for predicting the possible encoding
modes to reduce the computation [4].
In this paper, fast mode decision strategies for P-frame
coding are proposed to reduce the computational complexity of
video coding. The rest parts of this paper are organized as
follows. In Section II, different strategies including mode
prejudgment based on neighborhood MBs, mode combination,
and temporal-spatial correlation analysis are introduced, and
then the overall fast mode decision scheme is presented. In
Section III, experimental results are shown to demonstrate the
effectiveness of the proposed method. Finally, the paper is
concluded in Section IV.
Current
MB
8x8
4x88x4 4x4
8x8SKIP 16x16 16x8 8x16
16x16 4x4
intra
inter
Fig. 1. Full mode decision of current MB
II. FAST MODE DECISION
A. Mode Prejudgment Based on Temporal-Spatial
Neighborhood MBs
If the possible mode of current MB can be early predicted,
the motion estimation procedure of the rest modes can be
reduced. In general, SKIP and 16×16 modes account for the
largest proportion of the final mode statistic. Therefore, a mode
prejudgment strategy for SKIP and 16×16 modes is proposed.
This work was supported in part by the National Natural Science
Foundation of China (61201238 and 61571146), and the Fundamental
Research Funds for the Central Universities (HEUCF140812).