This study proposes two efficient blockmatching algorithms for fast motion estimation using a histogram ordering model in order to reduce the computational cost of video coding. Since the representative pixels that c...
详细信息
This study proposes two efficient blockmatching algorithms for fast motion estimation using a histogram ordering model in order to reduce the computational cost of video coding. Since the representative pixels that consist of edge or texture mainly contribute to the blockmatching criterion, it is important to analyse the details of the current block. To analyse the characteristics of the block with low complexity, the pixel histogram is used for the observation of the block. Based on this pixel histogram for the current block, an optimal blockmatching order is determined for the lossless fast matching algorithm. In addition, adaptive partial blockmatching algorithm for lossy fast motion estimation using histogram-based blockmatching ordering is also presented to further reduce the complexity of blockmatching. By capturing the representative pixels, blockmatching distortion is maximised soon enough during the blockmatching. For this reason, partial blockmatching is available instead of full blockmatching, which is also known as a full search algorithm. Our experimental results show that the proposed algorithm not only reduces the computational complexity of blockmatching criterion, but also tends to maintain the image quality when compared to the conventional fast matching algorithms.
暂无评论