IMAGE DENOISING USING HARD THRESHOLD TECHNIQUES ON WAVELET TRANSFORM AND SHEARLET TRANSFORM

Abstract

- In this data age century with increment in the modern technology there is a development in the theory of multidimensional data to provide the higher directional sensitivity in imaging. A numeric image is a portrayal of a real image which is taken as a set of numbers that can be gathered and picked up by a digital computer. In order to decode the image into numbers it is divided into small segments called pixels (picture elements). Whenever there is a transmission of images or due to some environment factor there is an addition of noise to the images takes place that ultimately results in the reduction of originality of the image. It is very important to remove the noise from the images so that it is safeguard. Shearlets are a multiscale foundation which authorize efficient encoding of anisotropic feature in multivariate problem classes. In this paper, we have set forth the noise removal transform by hard thresholding for denoising. We can denoise the noisy image by wiping out the fine details, to enhance the quality of the images. This paper presents the denoising of a natural image based on wavelet and shearlet trans- form with hard thresholding techniques which is used to eliminate noise from the image. The images are corrupted with Gaussian, salt pepper and speckle noise. The multiscale and multi-directional outlook of shearlet transform are methodical in take care of edges of an image in denoising procedures. Shearlet comes out as a methodical transform for edges analysis and detection. Quantitative performance parameters such as PSNR, MSE are used to evaluated the denoised image effect. And hence came to the conclusion that the Shearlet Transform with hard thresholding in pyshear lab (python) is an methodical technique for enhancing the overall quality of the image.