期刊名称：International Journal of Engineering and Computer Science
印刷版ISSN：2319-7242
出版年度：2013
卷号：2
期号：6
页码：1834-1844
出版社：IJECS
摘要：The visual effects of blocking artifacts can be reduced by using deblocking filter. Appearance of the image is not clear while an artifact occurs. While parallelization, deblocking filter has complicated data dependencies to provide insufficient parallelization. The performance can be easily affected by the synchronization, when the number of cores increases. The load imbalance problem will be occurs. The proposed algorithm has three separate filtering strong filter, normal filter & no filtering. The vertical (flat) directions used as strong filter because of human visual system is sensitive. The horizontal directions used as smoothing filter. The deblocking filter process is divided into two parts BSC & EDF. BSC is Boundary Strength Computation; EDF is Edge Discrimination & Filtering. BS value determines the strength of the filtering, when BS value is zero it is no filtering, BS value of 4 is performed strong filtering, other values uses normal filtering. BSC uses METPMHT (Markov Empirical Transition Probability Matrix and Huffman Tree). EDF uses IPCAP (Independent Pixel Connected Area Parallelization). These two methods are used to increase the parallelization and reduce the synchronization
关键词：Deblocking filter; blocking Artifacts; METPMHT (Markov Empirical Transition Probability Matrix and Huffman Tree). IPCAP (Independent Pixel Connected Area Parallelization); Discrete cosine Tranform
