Performance Analysis of Feature Combination in Object Classification

Abstract

The objective of this paper is to evaluate the strategy of feature combination i.e. to show the effectiveness of feature combination over single features (in terms of classification accuracy) and thus to study the behavior of recognition metrics by changing certain parameters. We can conclude from the results that feature combination is quite efficient in practical approach. The addition of new features to the existing ones gradually enhances the classification accuracy. Thus such systems can be a great help from the realistic point of view and can be utilized in day to day life. The recognition metrics depend differently on different number of aspects, thus if we can manage such parameters wisely, we can achieve better results by extending the application of feature combination.