Classification of nailfold capillary images using wavelet and Discrete Cosine Transform

This paper presents technique for classification of nail fold capillary images based on Discrete Wavelet Transform (DWT) and Discrete Cosine Transform (DCT). An approach for image enhancement and feature extraction using wavelet transform using its property of multilevel decomposition in pattern recognition application has been proposed. The main idea is to achieve better accuracy in classification by extracting more relevant features after dimensional reduction. Data compression and energy compaction is the main feature of DCT and therefore in the proposed method, DCT was applied on LL sub band image obtained from decomposition of the nail fold capillary image using DWT. A subset of the most significant coefficients was retained as the feature vector and using K Nearest Neighbor classifier, nail fold image was classified as healthy, normal and others. In this paper, classification rate of 73.3% using 500 coefficients of DCT and 70% using 5000 coefficients of DWT as feature vector has been achieved.