Hybrid Regression Analysis based Technique for Removal of Impulse Noise from Gray-Scale Images

Abstract

Noise is major limiting factor in digital images, due to noise quality of image is degrade. For improving quality of image filtering techniques are used. There are various filtering technique are reported in last few years. In this presented work the impulse noise and their effect in image is measured. In addition of that the recent contributions on the image filters for impulse noise detection and correction is also investigated. The Filtering algorithms considered are: Fuzzy Logic Based Adaptive Noise Filter [1], Cloud Model Filter [2]. Several runs on many images were made using these algorithms. Whereas the noise detection process of the CM filter was good, and the correction process of Fuzzy Logic Based Adaptive Noise Filter was better. By combining the concepts of both these algorithms and also using some additional concept like regression analysis, L ZERO smoothing, an improved impulse noise filtering algorithm is developed. Proposed algorithm enhances the image quality. This algorithm is iterative, so as much iteration as the user may desire can be made considering image quality. The performance of the proposed algorithm is measure in term of visual quality and PSNR. General Terms Error Filtering, Image Processing, Image Enhancement, Performance Improvement.