Feature Extraction Techniques for Efficient Analysis and Classification of Biomedical Images

This paper develops and evaluates different techniques for detection, classification and analysis of pattern in medical images. The research work studies the structure of fractal and multi-fractal images; and extracts the statistical self-similarity features characterized by the Holder exponent for pattern classification. The effectiveness of local and global features has recently attracted growing attention in the field of texture image classification and retrieval. Global features from multi-fractal descriptors are extracted and combined with local features from fractal descriptors to generate new descriptors for efficient discrimination of images. The experimental approaches are validated for different scales of images during the classification process in order to determine the appropriate image size that could yield the maximum classification accuracy. The experimental results show that the descriptors extracted from the combined features considerably improve the performance of the classifiers. The results achieved in this paper have greatly demonstrated the effectiveness of local and global features in the analysis and classification of biomedical images.