Signal Processing:Image Communication 28 (2013)1435-1447 Contents lists available at ScienceDirect IMAGE Signal Processing:Image Communication ELSEVIER journal homepage:www.elsevier.com/locate/image Fuzzy quantization based bit transform for low bit-resolution CrossMark motion estimation Chuan-Ming Song 3.b.*,Yanwen Guob,Xiang-Hai Wang 3.b,Dan Liua College of Computer and Information Technology.Liaoning Normal University,Dalian 116029.China National Key Laboratory for Novel Software Technology.Nanjing University.Nanjing 210093.China ARTICLE INFO ABSTRACT Article history: This study proposes a novel fuzzy quantization based bit transform for low bit-resolution Received 6 February 2013 motion estimation.We formalize the procedure of bit resolution reduction by two Received in revised form successive steps,namely interval partitioning and interval mapping.The former is a 6 August 2013 Accepted 24 September 2013 many-to-one mapping which determines motion estimation performance,while the latter Available online 12 October 2013 is a one-to-one mapping.To gain a reasonable interval partitioning.we propose a non- uniform quantization method to compute coarse thresholds.They are then refined by Keywords: using a membership function to solve the mismatch of pixel values near threshold caused Video coding by camera noise,coding distortion,etc.Afterwards,we discuss that the sum of absolute Motion estimation difference(SAD)is one of the fast matching metrics suitable for low bit-resolution motion Block matching Fuzzy quantization estimation in the sense of mean squared errors.A fuzzy quantization based low bit- Low bit-resolution resolution motion estimation algorithm is consequently proposed.Our algorithm not only Low bit-depth can be directly employed in video codecs,but also be applied to other fast or complexity scalable motion estimation algorithms.Extensive experimental results show that the proposed algorithm can always achieve good motion estimation performances for video sequences with various characteristics.Compared with one-bit transform,multi- thresholding two-bit transform,and adaptive quantization based two-bit transform,our bit transform separately gains 0.98 dB,0.42 dB,and 0.24 dB improvement in terms of average peak signal-to-noise ratio,with less computational cost as well. 2013 Elsevier B.V.All rights reserved. 1.Introduction heterogeneous networks in real time.These clients may have varying capabilities in display resolution,computing power. Various video services,such as surveillance,video tele- network bandwidth,etc.An efficient video codec is essential phony/conferencing,mobile streaming.wireless LAN video in for the video services in such a challenging scenario.Sullivan home network,and even beyond high definition video,are et al.[3]believed that most of the efficiency improvement of becoming available in more and more application scenarios state-of-art video codecs,e.g,H.264/AVC,results from better with the rapid development of Internet,wireless communica- temporal prediction and compensation.Motion estimation/ tion,and pervasive computing technologies [1.2].Mass videos compensation therefore plays a key role in coding system. are required to be reliably delivered to diverse clients over Unfortunately,motion estimation is usually remarked as the most computationally intensive component,consuming up to 50%[4],even to 60-80%[5.of the computation evolved in the .Corresponding author at:College of Computer and Information entire codec.Such a heavy computation load will inhibit Technology.Liaoning Normal University.Dalian 116029,China. practical video communication on portable and wireless Te1:+8615140397690. E-mail addresses:chmsong@163.com, devices with limited battery power.Hence an efficient motion chmsong@Innu.edu.cn(C.-M.Song).ywguo@nju.edu.cn (Y.Guo). estimation algorithm with low computational complexity is of xhwang@Innu.edu.cn (X.-H.Wang),liudan_di@163.com (D.Liu). great importance for real-time video services cross platforms. 0923-5965/S-see front matter 2013 Elsevier B.V.All rights reserved. http://dx.doi.org/10.1016/j.image.2013.09.007Fuzzy quantization based bit transform for low bit-resolution motion estimation Chuan-Ming Song a,b,n , Yanwen Guo b , Xiang-Hai Wang a,b , Dan Liu a a College of Computer and Information Technology, Liaoning Normal University, Dalian 116029, China b National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210093, China article info Article history: Received 6 February 2013 Received in revised form 6 August 2013 Accepted 24 September 2013 Available online 12 October 2013 Keywords: Video coding Motion estimation Block matching Fuzzy quantization Low bit-resolution Low bit-depth abstract This study proposes a novel fuzzy quantization based bit transform for low bit-resolution motion estimation. We formalize the procedure of bit resolution reduction by two successive steps, namely interval partitioning and interval mapping. The former is a many-to-one mapping which determines motion estimation performance, while the latter is a one-to-one mapping. To gain a reasonable interval partitioning, we propose a non￾uniform quantization method to compute coarse thresholds. They are then refined by using a membership function to solve the mismatch of pixel values near threshold caused by camera noise, coding distortion, etc. Afterwards, we discuss that the sum of absolute difference (SAD) is one of the fast matching metrics suitable for low bit-resolution motion estimation in the sense of mean squared errors. A fuzzy quantization based low bit￾resolution motion estimation algorithm is consequently proposed. Our algorithm not only can be directly employed in video codecs, but also be applied to other fast or complexity scalable motion estimation algorithms. Extensive experimental results show that the proposed algorithm can always achieve good motion estimation performances for video sequences with various characteristics. Compared with one-bit transform, multi￾thresholding two-bit transform, and adaptive quantization based two-bit transform, our bit transform separately gains 0.98 dB, 0.42 dB, and 0.24 dB improvement in terms of average peak signal-to-noise ratio, with less computational cost as well. & 2013 Elsevier B.V. All rights reserved. 1. Introduction Various video services, such as surveillance, video tele￾phony/conferencing, mobile streaming, wireless LAN video in home network, and even beyond high definition video, are becoming available in more and more application scenarios with the rapid development of Internet, wireless communica￾tion, and pervasive computing technologies [1,2]. Mass videos are required to be reliably delivered to diverse clients over heterogeneous networks in real time. These clients may have varying capabilities in display resolution, computing power, network bandwidth, etc. An efficient video codec is essential for the video services in such a challenging scenario. Sullivan et al. [3] believed that most of the efficiency improvement of state-of-art video codecs, e.g., H.264/AVC, results from better temporal prediction and compensation. Motion estimation/ compensation therefore plays a key role in coding system. Unfortunately, motion estimation is usually remarked as the most computationally intensive component, consuming up to 50% [4], even to 60–80% [5], of the computation evolved in the entire codec. Such a heavy computation load will inhibit practical video communication on portable and wireless devices with limited battery power. Hence an efficient motion estimation algorithm with low computational complexity is of great importance for real-time video services cross platforms. Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/image Signal Processing: Image Communication 0923-5965/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.image.2013.09.007 n Corresponding author at: College of Computer and Information Technology, Liaoning Normal University, Dalian 116029, China. Tel.: þ86 15140397690. E-mail addresses: chmsong@163.com, chmsong@lnnu.edu.cn (C.-M. Song), ywguo@nju.edu.cn (Y. Guo), xhwang@lnnu.edu.cn (X.-H. Wang), liudan_dl@163.com (D. Liu). Signal Processing: Image Communication 28 (2013) 1435–1447