Author : Deepthi P. Jordhana and Soundararajan. K

The objective of the paper is to classify the objects occurring in SAR Images as Natural or Manmade using Kernel based Support vector machines method. The non linear image data is mapped into piecewise linear data using a kernel function. Support vector machine uses training images to train a classifier and this classifier is tested on the test images. Training Images are classified into positive for natural images and negative for manmade objects. Test Images are the SAR regions under consideration. SVM method with Radial Basis kernel (RBF) function is used to categorize the regions of a SAR Image. The parameters that measure the performance of using this method are False Alarm(FA) and Target Miss(TM).They indicate the percentage of incorrectly detected objects. Simulation results based on Matlab proves the efficiency for manmade classification for different sets of SAR Images.

Keywords: Support vector machines; RBF Kernel; Image Classifier

Article published in International Journal of Current Engineering and Technology, Vol.4, No.6 (Dec-2014)