DCT block effect elimination method based on Gaussian filtering

Abstract

The invention discloses a DCT block effect elimination method based on Gaussian filtering. The DCT block effect elimination method comprises the following steps: S1, constructing 8 * 8 pixel point blocks, adjacent blocks and an overlapping area in an image; S2, adaptively distinguishing the types of the image blocks based on the matrix characteristics of the image blocks, wherein the image blocks comprise flat area image blocks and texture area image blocks; and S3, carrying out Gaussian filtering on the flat area image block by using a large Gaussian filtering radius and a large Gaussian filtering coefficient, and carrying out Gaussian filtering processing on the texture area image block by using a small Gaussian filtering radius and a small Gaussian filtering coefficient. A flat region and a texture region of the image are adaptively distinguished according to the matrix characteristics of each 8 * 8 image block, so that the reliability and stability of the adaptive method are higher; gaussian filtering processing is carried out on different region types by adopting Gaussian filtering coefficients and Gaussian filtering radiuses according with image block characteristics based on Gaussian filtering, so that the high-frequency information of block edges is well reserved while the block effect is more effectively removed.

Description

technical field [0001] The invention belongs to the technical field of image denoising, and more specifically, the invention relates to a Gaussian filter-based DCT block effect elimination method. Background technique [0002] The international standard JPEG for image compression is based on the DCT transformation coding of image blocks. This transformation first takes out the spatial redundancy between adjacent pixels in the block, and then performs quantization, which can effectively realize data compression. In the compression coding based on DCT transformation, an image is divided into several 8*8 pixel blocks, and then DCT transformation is performed on each block to obtain 64 DCT coefficients, which greatly reduces the amount of calculation. However, this block-wise DCT processing method ignores the correlation between blocks. When quantizing each block of DCT coefficients, the DCT coefficients are divided by the quantization coefficients and then rounded up, and some...

AI Technical Summary

Problems solved by technology

When the compression rate is high, the disadvantages of this block DCT transformation will be particularly obvious. This transformation method will cause a large amount of high-frequency information at the edge of the block to be lost, resulting in discontinuity at the boundary of the block in the reconstructed image. Jump, this is the so-called DCT transform block effect

[0003] The existing DCT transform block effect elimination method is based on empirical thresholds (fixed thresholds) to determine flat areas and textured areas. The empirical thresholds are only good for distinguishing specific images, but they are not universal. There are flat areas and Inaccurate recognition of texture areas affects the quality of filtered images

Images