Method for Removing Salt and Pepper Noise in Ultra-high Density Images

Abstract

The invention discloses a method for reducing salt and pepper noise of an image with ultra-high density. It mainly solves the problem that the existing methods cannot filter out the ultra-high density salt and pepper noise. The implementation steps are: (1) use the extreme value method to detect noise points; (2) first replace the value of the noise points with the mean value of the non-noise points of the four-point template in the 3×3 window, recursively process 3 times, and then use the 3×3 The mean value of the non-noise point of the cross template of the window replaces the value of the noise point, and the recursive processing is performed 3 times; in each recursive processing, it is detected whether the noise point is processed, and if the processing is not completed, the window is enlarged, and the four-point template and the cross template are used respectively Perform the same recursive mean filtering until the end of the 11×11 window; (3) If the noise density is above 0.85, use a 3×3 cross window or global window to continue the mean filtering process, and finally achieve ultra-high density noise reduction. The invention has the advantages of low computational complexity, large denoising range and effective suppression of ultra-high-density salt-and-pepper noise.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method for reducing ultra-high-density salt-and-pepper noise in images that can be used in the fields of industrial images, agricultural images, medical images, and astronomical images. Background technique [0002] More than 50% of the pixels in the image with high salt and pepper noise are polluted by salt and pepper noise, and the noise level exceeds the range that conventional filtering algorithms in machine vision can handle. The traditional point of view is that: as a nonlinear filter, the median filter has the ability to filter salt and pepper noise and retain image details, and is one of the most suitable algorithms for image filtering with high salt and pepper noise in conventional filtering algorithms. On the basis of value filtering, some median filtering algorithms for high salt and pepper noise images are proposed. Esakkirajan S et al. propose...

AI Technical Summary

Problems solved by technology

These improved methods improve the effect of removing salt and pepper noise, but the experimental results show that: for more than 90% of the pixels are polluted by ultra-high density salt and pepper noise, the denoising effect of these methods is not ideal

Other common denoising methods include methods based on Markov field models and partial differential equations, wavelet domain methods, principal component methods, and geometric scale analysis methods, etc., but some of these methods fail when dealing with ultra-high-density noise. , some calculation complexity is extremely high, it is difficult to apply

Images