Recognition of SAR Images Using Pattern Classification Methods

The purpose of this paper is to present a system for the classification of synthetic aperture radar SAR images. SAR is considered as an effective scanning system for the earth surface. It is operated by using coherent electromagnetic wave (EMW) which lies in the range of microwave (MW). The importance of this system has been increased through the last few decades, because it has been found as to be very efficient scanning tool, does not affected by weather condition, and can be used at all day s time. The main shortcoming associates with this imaging system is the speckle noise (coherent fading), which is causes by the coherent nature of the EMW. In fact, classifying SAR images is not an easy task if not impossible, because the complicated corruption of the image signal. As will be illustrated, to reduce the speckle effects, we have pre-processed the SAR images by implementing different smoothing techniques. In this paper, two techniques are considered in analysis operation (Minimum Distance (MDS) and Maximum Likelihood (MLH) classification methods). The quality test showed that the introduced methods in this work have high accuracy results in classifying image regions.