Wavelet Based Image Denoising Using Ant Colony Optimization Technique for Identifying Ice Classes in SAR Imagery

Abstract

Interpretation of satellite radar images is an important ongoing research field in monitoring river ice for both scientific and operational communities. This research focus on the development of optimal recognition strategies for the purpose of identifying different ice classes in SAR imagery. However acquisitions of SAR images produce certain problems. SAR images contain speckle noise which is based on multiplicative noise or rayleigh noise. Speckle noise is the result of two phenomenons, first phenomenon is the coherent summation of the backscattered signals and other is the random interference of electromagnetic signals. This therefore degrades the appearance and quality of the captured images. Ultimately it reduces the performance of important techniques of image processing such as detection, segmentation, enhancement and classification etc. This research contributes the major objectives towards speckle filtering using wavelet techniques optimized by Ant Colony Optimization (ACO). First is to remove noise in uniform regions. Second is to preserve and enhance edges and image features and third is to provide a good visual appearance. The work is carried out in three stages. First stage is to transform the noisy image to a new space (frequency domain). Second stage is the manipulation of coefficients. Third is to transform the resultant coefficients back to the original space (spatial domain). Results show that statistical wavelet shrinkage filters are good in speckle reduction but they also lose important feature details. Here the challenge is to find an appropriate threshold value, which is achieved using intra-scale dependency of the wavelet coefficients to estimate the signal variance only using the homogeneous local neighboring coefficients. Moreover, to determine the homogeneous local neighboring coefficients, the ACO technique is used to classify the wavelet coefficients. Experimentation is conducted on SAR images which are further used for development of optimal recognition system for ice classes.