Sparse-region residual error compensating and revising method for improving marginal definition during image sampling

Abstract

The invention discloses a sparse-region residual error compensating and revising method for improving marginal definition during image sampling. The method comprises the following steps of: carrying out upper sampling treatment on an image, and carrying out overlapping and blocking on the upper sampled image; searching non-local similar blocks on each image block; establishing a data pool by taking index values of the image blocks and the non-local similar blocks as data streams; carrying out sparse-region residual error compensating and revising iteration treatment on the data streams in the data pool in a parallel manner; and when the iteration is finished, updating the data pool and integrating the image blocks into a high-resolution image. The sparse-region residual error compensating and revising method disclosed by the invention utilizes the non-local similarity among the image blocks and the sparsity of signals in the blocks to better overcome the defects of a marginal saw tooth effect, noise detail loss and the like during the upper image sampling process, so that the marginal definition of the image can be greatly improved.

Description

Technical field [0001] The invention belongs to the technical field of image processing and display, in particular to a sparse domain residual compensation correction method for improving edge definition in image upsampling. Background technique [0002] In the field of image processing and image display, it is often necessary to display images on display devices with different resolutions, or to zoom in to display local details of the image, so image upsampling is a commonly used technique. For another example, in the field of image compression and coding, in order to reduce the transmission bit rate of the image, an effective method is to first downsample the image to reduce the image resolution, and then use the traditional video coding technology to compress, and the decoded image is processed on the decoding end. The up-sampling operation restores the original resolution image. This method can greatly reduce the transmission code rate, but the down-sampling filter and the u...

AI Technical Summary

Problems solved by technology

These methods are faster, but have the following disadvantages: 1) When the image is enlarged in a large scale, it is easy to cause the image to be jagged near the edge; 2) When the image has noise, it is easy to amplify the noise

These methods are upsampling methods for special applications. For large-scale upsampling, it is easy to cause serious image edge aliasing, and it is easy to amplify noise.

Images