Compression enhancement using the hybrid motion estimation in sub-image array transformed from elemental image array in three-dimensional integral image

Abstract

In this paper, we suggest the enhanced compression scheme using the hybrid motion-estimation in sub-image array (SIA) transformed from elemental image array (EIA) in three-dimensional integral imaging. In the proposed method, firstly, a macroblock in the reference sub-image (SI) is matched in the local SI applying to MSE (Motion Square Estimation) by three step search (TSS) to search the coordinates of current macro block. The second step is to search the most matched area in SIs and the coordinate of macro block for full searching, and finally, the start point for searching in the local SI is altered by the former coordinate. Accordingly, the computed motion estimation from the block-matching with TSS and the full searching is presented as MV and the object in the reference SI is shifted to the object position of each current SIs to compensate their MV based on the motion estimation. The computation time for block-matching by the TSS to search approximate current macro block range in the first step and the full searching to acquire the exact current macro block range in the second step is decreased. In addition, the video compression such as MPEG-4 is applied to encode the data of the consecutive frames. Some experiments are carried out and compression efficiency of the proposed scheme has been improved 676.30%, 357.19% and 3.37% on the condition of 30[dB] approximately, compared with the JPEG compression, EIA compression and Full searching method. The computation time is also improved 196.97% compared with the full search scheme for motion estimation and compensation.