Performance Analysis of Object Shape Classification and Matching from Tactile Images Using Wavelet Energy Features

Tactile images while grasping objects are acquired and wavelet based features are extracted for matching and classification. The performance of matching and classification is evaluated in terms of matching rate and classification accuracy along with the computation times. This comparison will help in determining the applicability of classification or matching in future works including real time applications. Highest classification accuracy is found to be 86%, in 0.0619 sec, while the best matching r ate obtained is 96% in 0.0041 sec. Thus Image matching is suitable for real time applications taking less computation time while providing significant performance improvement at the same time.