CRF Adaptive Method Based on Block Effect Detection

Abstract

The invention discloses a CRF adaptive method based on block effect detection, which belongs to the technical field of image coding. The method includes: dividing an original frame and a reconstructed frame into blocks, performing edge detection on a macro block, and judging whether it is a block effect edge ; Calculate the blockiness strength of the current frame according to the blockiness edge; finally, perform adaptive adjustment of CRF according to the blockiness strength. The present invention uses the block effect that has a great impact on subjective results as the quality evaluation index, and the encoder can adaptively adjust the CRF parameter value according to this quality index, so that when the quality is too high, the CRF can be increased to reduce the code rate and quality. When it is too low, CRF can be reduced to improve the bit rate and quality, and obtain better overall quality and lower average bit rate.

Description

technical field [0001] The invention relates to the technical field of image coding, in particular to a CRF adaptive method based on block effect detection. Background technique [0002] In the encoding process, if the quantization step is too long, there will be a clear border line between the macroblock and the macroblock after encoding, and the pixel values ​​​​in the macroblock will tend to be uniform, which results in a mosaic like a checkerboard. Grid, that is, block effect. Although the deblocking filter is introduced in the x264 encoder to reduce the block effect, when the block effect is serious, even the use of deblock filter cannot completely eliminate the block effect, and it will cause blurred details. Another way to reduce the block effect is to reduce the quantization step size, but the bit rate also increases accordingly, and the optimal quantization step size values ​​for different scenes of different videos are different. The CRF (Constant Rate Factor) us...

AI Technical Summary

Problems solved by technology

[0003] In addition, the current x264 encoder estimates the strength of the block effect before deblocking filtering. The method used can only use the information of the reconstructed frame (that is, the encoding result) and cannot use the original frame. This is because the block effect detection used by x264 The method needs to be used at the encoding end and the decoding end at the same time, and the methods used at both ends need to be consistent, and the original frame information cannot be obtained at the decoding end

The accuracy of blockiness intensity estimation using only reconstructed frames will be affected

Images