Image processing method and device based on interpolation shear wave

Abstract

The invention relates to an image processing method and device based on interpolation shear waves, wherein the method includes using parameterized wavelet basis functions to construct interpolation shear waves; according to the constructed interpolation shear waves, by introducing a staggered shear transformation matrix , construct an interpolation Shearlet function; perform cross-cut transformation on the image to be processed according to the cross-cut transformation matrix; perform multi-scale interpolation wavelet transform on the image to be processed after the cross-cut transformation according to the interpolation Shearlet function to achieve Decomposition and reconstruction of the image to be processed; denoising the decomposed and reconstructed image to be processed through the threshold method. The present invention can realize multi-scale and multi-directional decomposition of images, has faster transformation speed, and can extend the range of image processing from Hilbert space to Banach space, can better retain the texture of the image, and can identify the image more accurately. Texture and noise, thereby avoiding the appearance of artificial artifacts caused by the diffusion of texture.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image processing method and device based on interpolation shear wave. Background technique [0002] Texture is an important feature for identifying objects in images. Therefore, the current image processing methods, especially the noise reduction process, hope not to destroy the texture structure of the image. The common method of image noise reduction is to regard the noise as a high-frequency signal, and perform noise reduction through various filters, such as mean filter, median filter, Wiener filter and wavelet transform. [0003] In fact, the texture, boundary contour, etc. in the image contain rich high-frequency information. The simple filtering method will inevitably destroy the texture and contour of the image while reducing the noise, so that the image after noise reduction is blurred and affects the subsequent processing of the image. . To solve ...

AI Technical Summary

Problems solved by technology

[0005] In view of the defect that the existing noise reduction method cannot correctly identify texture and noise, resulting in an unsatisfactory noise reduction effect, the present invention proposes the following technical solutions:

Images