| 摘要 |
Video deblocking can be implemented in video content processing and delivery environments when displaying decompressed/decoded video information. Discontinuities are identified and smoothed to reduce blockiness, for example at macroblock boundaries in conjunction with Moving Picture Expert Group (MPEG)-based schemes. In exemplary implementation(s), a client device detects discontinuities at or near block boundaries, determines which discontinuities are artificial, and smooths at least those discontinuities that are determined to be artificial. These actions may be accomplished using (i) one or more Laplacian of Gaussian (LoG) operations as applied to both an image and a quantization matrix and (ii) windowed comparisons between values and a threshold resulting from the LoG operations. Alternatively, these actions may be accomplished using a spatio-temporally varying filter. Additionally, in an MPEG context, intra image units may be deblocked using the former approach while non-intra image units may be deblocked using the latter approach.
|
