Research On Image Enhancement Methods Based On Shearlet And NSST Transform

In the form, transmission and reception of images, the quality of images declined because of various reasons. To facilitate the follow-up analysis of image processing, the image enhancement is necessary in the image processing field. In this paper, select the remote sensing image and medical image as the test sample. The purpose of remote sensing image enhancement is let the users gen the information they need from the enhanced image easily and improve the visual effect of remote sensing image as a whole. And the medical image enhancement purpose is to enhance the accuracy of medical diagnosis, so that the enhanced image can be meet the conditions in some specific application requirements. Regardless of remote sensing images and medical images, according to different application requirements, the needed enhancement algorithm is also different. But now there is no universal algorithm, so there is no the unified image evaluation standard. In this paper, we have evaluate the effect of the enhanced image by using the subjective and objective index evaluation.Image enhancement processing has been mainly researched from the three aspects such as improve contrast, image denoising, edge details strengthen. First, introduced the features and theory of spatial domain image enhancement, the frequency domain of image enhancement such as wavelet transform, contourlet transform,NSCT transform, shearlet transform and NSST transform. In general, the de-noising, contrast enhancement, edge and detail enhancement are the main contents of image enhancement. The gray level enhancement and histogram modification technology belongs to the contrast enhancement, In this paper, enhance the overall contrast of image by through the adaptive histogram equalization. The gradient operator, the laplace operator and the sharpning mask technique is belongs to the edge and detail enhancement method and the proposed method in the paper using the linear unsharpning mask and laplace operator to enhance the edge and detail information. But the adaptive histogram equalization leads to noise amplification and the sharpning mask method is very sensitive to noise, so that these excessive noise influence the enhancement effect. Therefore, the image after HE is decomposed into the frequency domain and the threshold de-noising method is used to romove noise in the high-frequency, which contains the noise and details information. This paper adopts the adaptive de-noising method based on the BayesShrink and NSST domain.At the end of the paper, test the performance of the two kinds proposed method and their compared methods. The experimental results show that the two enhancement algorithm is achieved satisfactory results not only the visual effect but also the objective evaluation standard.