Feature Extraction and Classification for Textured Images
Kai Fang · Microelectronics & Computer · 2005
A method of effective extraction features for textured images is presented in this paper. Based on frequency domain distribution and scale feature of textural information, we can do texture classification effectively. We use support vector machine with good classification performance as classifier. The experiment results show that the feature vectors extracted are steady and higher classification accuracy can be attained even if the number of the classes is larger.