Image filtering algorithm based on genetic algorithm and Shearlet wavelet

Abstract

The invention relates to an image filtering algorithm based on a genetic algorithm and a Shearlet wavelet. The algorithm is carried out according to the following steps: 1), designing a change threshold of each scale and each direction for the distribution characteristics of different scales and different directions of a Shearlet transformation domain; 2), measuring image quality by use of a peak signal-to-noise ratio (PSNR) and a mean square error (MSE), and taking it as a target function of a multi-target genetic algorithm, wherein the target function is shown in the descriptions, and W1 and W2 are weights; and 3), adaptively determining an optimal threshold of each scale and each direction by use of the multi-target genetic algorithm to realize self-adaptive denoising based on image content. According to the invention, compared to an original Shearlet wavelet, the problem of searching for wavelet coefficients under the conditions of different scales and different directions is taken into consideration, a set soft threshold parameter is employed, and images are smoother and softer and are suitable for comprehension of human eyes. Since an optimization algorithm is employed, the reduction denoising effect of the images is greatly improved.

Description

technical field [0001] The invention belongs to the field of image filtering, specifically an image filtering algorithm for Rician noise [0002] Background technique [0003] During the process of image acquisition, transmission and processing, it will be disturbed by noise. If it is not filtered, it will seriously affect the post-processing work. People's requirement for filtering is to filter out noise pixels and retain image details, but since details and noise are difficult to distinguish, this becomes a more difficult problem. [0004] Magnetic resonance imaging has brought considerable convenience to medical diagnosis, and can easily detect many lesions in patients. But unfortunately, most MRI images will be polluted by noise, and the object information contained in them will become blurred and insufficient, which adds difficulties to the further correct analysis of the images. In the field of magnetic resonance imaging, Rician noise is more common. This is a re...

AI Technical Summary

Problems solved by technology

However, through careful analysis and reference to other papers, we believe that this method still has two aspects that can be further improved. One is that the energy distribution of Shearlet coefficients at different scales and directions can be considered. We can The optimization algorithm is used to adjust the coefficients according to different situations, so as to achieve the purpose of improvement; second, the existing hard threshold filtering method is too simple, and does not fully consider the transition problem of details and noise. This paper intends to use soft threshold to try

Images