Paper Info
Title
A hybrid block-based motion estimation algorithm using JAYA for video coding techniques.
Abstract
In any video coding solution, motion estimation (ME) and motion compensation (MC) techniques are widely used and they play an inevitable role in reducing the temporal redundancies between successive frames. Block-based motion estimation (BME) is one of the widely utilized techniques in the recently developed video compression standards by JCT-VC including HEVC due to its efficacy and ease of implementation. In BME, the frames in a video sequence are partitioned into a number of non-overlapping blocks. Then, for each of the blocks, a best-matched block is obtained within a definite search region in the reference frame to minimize certain cost function or fitness value such as the sum of absolute difference (SAD), mean of absolute difference (MAD), or mean square error (MSE). However, the key challenge lies in the evaluation of these cost functions, since they are computationally expensive and involves the most time-taking operations in the BME process. Hence, BME-based approaches can be viewed as an optimization problem and meta-heuristic algorithms can be effectively exploited for the problem under consideration. In this paper, a hybrid BME technique using a recently developed optimization algorithm, namely, JAYA algorithm, is proposed. Besides this proposal, several other modules, namely, the fitness approximation technique, search history preservation, and early & adaptive termination criterion are utilized. The main objective behind the use of the aforementioned modules is to avoid the unnecessary evaluation of the fitness function. Exhaustive simulations are carried out to demonstrate the efficacy of the proposed method over that of the benchmark schemes with respect to different performance measures, namely, the peak signal-to-noise ratio (PSNR), PSNR degradation ratio (DPSNR), search efficiency, structural similarity index measure (SSIM), and computation time. Comparative analysis and quantitative evaluation clearly show that the present technique produces more reliable motion vectors with competitive time complexity as compared to that of the benchmark schemes.
Year
DOI
Venue
2019
10.1016/j.dsp.2019.01.016
Digital Signal Processing
Keywords
Field
DocType
Motion estimation (ME),Sum of absolute difference (SAD),Block matching (BM),JAYA algorithm,Peak signal-to-noise-ratio (PSNR)
Reference frame,Pattern recognition,Motion compensation,Fitness function,Fitness approximation,Artificial intelligence,Motion estimation,Time complexity,Data compression,Optimization problem,Mathematics
Journal
Volume
ISSN
Citations 
88
1051-2004
0
PageRank 
References 
Authors
0.34
0
4
Name
Order
Citations
PageRank
Bodhisattva Dash192.15
Suvendu Rup2115.55
Figlu Mohanty372.44
M. N. Swamy410418.85