article · 2010
Novel step search of block-size selection for variable block-size motion estimation
Hsueh-Yen Yang, Yu‐Cheng Fan, Hen‐Wai Tsao
Cited by 2Abstract and details only
Abstract
Variable block-size motion estimation (VBSME) of H.264/AVC utilizes seven block-sizes to describe the movement of video details. The VBSME can extremely improve the efficiency of compressed image but the problems of the block-size selection and aspect effects impact the resultant quality. Therefore, the number of motion vector (MV) determines the efficiency of video coding. Additionally, because the block matching algorithm (BMA) of small block would be interfered by surrounding aperture, the searching results often mistake the originals of vision reception and cause the unreliability of reconstructed frame. In the paper, we propose dual issues to solve current problems of VBSME. First, weighted calculation of VBSME is presented to avoid the affects of aperture effects. However, a step search of block-size selection is proposed to detect the object motion and efficiently apply the MVs of small-size block.
Cite this paper
Yang, H.-Y., Fan, Y., & Tsao, H. (2010). Novel step search of block-size selection for variable block-size motion estimation. https://doi.org/10.1109/icalip.2010.5685187
Yang, Hsueh-Yen, et al. “Novel step search of block-size selection for variable block-size motion estimation.” 2010, pp. 1705–09, https://doi.org/10.1109/icalip.2010.5685187.
Yang, Hsueh-Yen, Yu‐Cheng Fan, and Hen‐Wai Tsao. 2010. “Novel step search of block-size selection for variable block-size motion estimation.” https://doi.org/10.1109/icalip.2010.5685187.
Yang, H.-Y., Fan, Y. and Tsao, H. (2010) ‘Novel step search of block-size selection for variable block-size motion estimation’. Available at: https://doi.org/10.1109/icalip.2010.5685187.
@article{yang2010novel,
title = {{Novel step search of block-size selection for variable block-size motion estimation}},
author = {Yang, Hsueh-Yen and Fan, Yu‐Cheng and Tsao, Hen‐Wai},
pages = {1705--1709},
year = {2010},
doi = {10.1109/icalip.2010.5685187}
}Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.