A Comparative Evaluation of Denoising of Remotely Sensed Images Using Wavelet, Curvelet and Contourlet Transforms
- PDF / 2,471,089 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 52 Downloads / 227 Views
RESEARCH ARTICLE
A Comparative Evaluation of Denoising of Remotely Sensed Images Using Wavelet, Curvelet and Contourlet Transforms Rizwan Ahmed Ansari & Kirshna Mohan Budhhiraju
Received: 29 January 2014 / Accepted: 14 September 2014 # Indian Society of Remote Sensing 2016
Abstract This paper presents an overview of remotely sensed image denoising based on multiresolution analysis. In this paper, the wavelet, curvelet and contourlet transforms are used for denoising of remotely sensed images with additive Gaussian noise. The curvelets and contourlets are two kinds of new multi-scale transforms which can capture the intrinsic geometrical structure of data. At first, we outline the implementation of these multiscale representation systems. The paper aims at the analysis of denoising of image using wavelets, curvelets and contourlets on high resolution multispectral images acquired by the QuickBird and medium resolution Landsat Thematic Mapper satellite systems. We apply these methods to the problem of restoring an image from noisy image and compare the effects of denoising. Two comparative measures are used for evaluation of the performance of the three methods for denoising. One of them is the peak signal to noise ratio and the second is the ability of the denoising scheme to preserve the sharpness of the boundaries. By both of these comparative measures, the curvelet has proved to be better than the other two.
Keywords Ridgelets . Curvelets . Contourlets . Remote sensing images . Directional filter bank . Additive noise
R. A. Ansari (*) : K. M. Budhhiraju Satellite Image Processing Lab, Centre of Studies in Resources Engineering (CSRE), Indian Institute of Technology, Bombay, Powai, Mumbai 400076, India e-mail: [email protected] R. A. Ansari e-mail: [email protected] K. M. Budhhiraju e-mail: [email protected]
Introduction In the last few years, advancements in observing the Earth from space have led to a new class of images with very high spatial resolution. These high resolution images offer very good quality of detailed information about the properties of the objects. For various purposes, remote sensing images are used to extract some parameters, detect the presence or extent of various phenomena, or for interpretation. These applications require high signal-to-noise ratio (SNR) to get correct results to get better performance. The data that are contaminated with noise can cause a failure to extract valuable information and degrade the interpretability. A large number of noise filtering algorithms are present in the literature. In the recent years, there has been a lot of interest in wavelet based methods for noise removal from signals and images. In literature, a wide range of wavelet-based tools and ideas have been proposed and studied (Donoho 1995; Coifman and Donoho 1995; Donoho and Johnstone 1995; Bui and Chen 1998; Bose and Chappalli 2004; Kang and Zhang 2008; Zhao et al. 2010). Wavelet transform showed great effect when dealing with one and two-dimensional signal with point singularity fea
Data Loading...
