Block-removal filtering method and device

Abstract

The invention discloses a method for video or image deblocking filter and the device thereof. According to the block types of the two sides of a boundary and a motion vector, whether to perform deblocking filter is judged, and a needed filter strength when deblocking filter is needed is selected, and then deblocking filter is carried out in accordance with the filter strength and image content. Compared with current deblocking filter method, the deblockign filter method can improve the objective quality and the subjective quality of an image and the adaptability of filter without increase of calculation complexity and storage complexity.

Description

technical field [0001] The present invention relates to a digital data processing technology, in particular, to a method and device applied to video or image deblocking filtering. Background technique [0002] Traditional video or image coding standards such as JPEG, H.261, MPEG-1, MPEG-2, H.264, AVS, etc. are all based on block-based discrete cosine transform (BDCT). Quantization and encoding do not consider the correlation of pixels in adjacent blocks, so the image will have false edges on the sub-block boundaries, that is, block effects. When the degree of quantization of transform coefficients is too large, serious block effects will appear in the image, resulting in a decline in the subjective quality of the image. By performing deblocking filtering on the decoded image, the block effect can be effectively removed and the subjective quality can be improved. The latest international video coding standard H.264 and the Chinese national standard AVS both use deblocking f...

AI Technical Summary

Problems solved by technology

In the deblocking filtering process, it is necessary to perform multiple conditional judgments at the block level and pixel level, so it is difficult to perform parallel operations on hardware platforms such as DSP

Secondly, due to the many pixels involved in the deblocking filtering process, basically all pixels in an image need to be filtered, so operations such as memory read and write control are also very time-consuming and complicated.

Images