Tahmina Akhtar†, Rahima Akter†, Chhalma Sultana Chhaya †, Ashfaqur Rahman ‡ † Military Institute of Science and Technology/Dept of CSE, Dhaka, Bangladesh, ‡ Central Queensland University/Centre for Intelligent and Networked Systems, QLD, Australia firstname.lastname@example.org, email@example.com, firstname.lastname@example.org, email@example.com
Video coding is a process for representing video sequences in a compact manner. A significant step in video coding is searching for similar segments in previous frames and use only the difference information for reconstruction thus reducing space requirement. Different search techniques including Full search and 2-D logarithmic search etc. are used in the current literature. Full search restricts its application because of its computational load. 2D logarithmic search is computationally less expensive although there are some spaces for improvement. In this paper we propose a new search technique by modifying the 2-D logarithmic search that requires less search points with insignificant loss in visual quality. Experimental results demonstrate the effectiveness of the proposed technique. Keywords: video coding, 2-D logarithmic search.
Video is a sequence of still images representing scenes in motion. A video is created by capturing a numbers of still images in a short time interval. When these still images are displayed very quickly, it represents the motion of the object in the images. Video represent the huge amount of data. In order to transfer video data from one place to another efficiently it is required to compress the size of video data. One way to compress the size of video data is video coding [  ] [  ]. The principal goal in the design of a video-coding system is to reduce the transmission rate subject to some picture quality constraint. In transmission side, the first frame (normally called the reference frame) is transmitted as it is and the remaining frames are sent as a function of the reference frame. The frame to be sent is divided into a number of blocks and the best match for the block is looked for in the search window of the reference frame. This processing is called the search technique in video coding literature. There exist a number of video coding techniques including MPEG-1/2/4 [  ] [  ], H.26X [  ] etc. uses search techniques like Full search [  ], 2-D logarithmic search [  ], Coarse-Fine-Three-Step search [  ], Conjugate Direction search [  ], and Pyramid search [  ]. Each of these search techniques has merits and demerits in their favor. Full search finds the best match for a block as it searches all the candidate positions in the search window. Full search however is computationally expensive and renders difficulty for real time implementation. Some variants exist that applies some heuristics to reduce the candidate search points and reduce the computational complexity although compromising the image quality a bit. 2-D logarithmic search is one such search technique that reduces the search points to a subset of the search window (to be detailed in literature review) and finds the near-optimal best match with reduced computational complexity. Although computationally inexpensive it contains some redundancy in the search space. We aim to reduce this redundancy and aim to find a modified 2-D logarithmic search technique with even reduced computational load. Experimental results demonstrate that the proposed technique reduces the number of search points and thus reduces search time with insignificant sacrifice of image quality. The paper is organized as follows. In Section II we elaborate some related works. In Section III we present our proposed search approach. Some experimental results to demonstrate the effective of the proposed approach is presented in Section IV. Finally Section V concludes the...