Reduction of Unclassified Region of Support Vector Machine Using RBF Network for Colour Image Classification

Unclassified region decreases the efficiency and performance of multi-class support vector machine. The proper selection of feature sub set reduced the unclassified region of multi-class support vector machine. Now a day’s multi-class classification are widely used in image classification. The feature selection or mapping of data one space to another space creates diversity of outlier and noise and generate unclassified region for image classification. For the reduction of unclassified region we used radial basis function for sampling of feature and reduce the noise and outlier for feature space of data and increase the performance and efficiency of image classification. Our proposed method optimised the feature selection process and finally sends data to multiclass classifier for classification of data. Here we used support vector machine for multi-class classification. As a classifier SVM suffering two problems (1) how to choose optimal feature sub set input and (2) how to set best kernel parameters. These problems influence the performance and accuracy of support vector machine. Now the pre-sampling of feature reduced the feature selection process of support vector machine for image classification.