Characterization of Human Fingernails Using Iterative Thresholding Segmentation

In this paper, we present new fingernail biometric as a possibility in pattern recognition and establish its prospects experimentally. In this perspective composition, we propose a method adapted in three stages. In stage one, the finger biometric which exhibits Gaussians of objects is identified as background and foreground elements is modeled for segmentation of relevant regions. Preliminary methods include intensity adjustment to reduce noise, contrast enhancement for edge detection, and morphological operation to improve finger region from noisy background. In stage two, iterative histogram-based thresholding of multispectral image (R, G, and B components) to binarize fingernail region from finger object is adapted. In stage three, geometric feature calculation makes it possible to identify fingernail into different shapes as oval, round, and rectangular. Nail dimension and shape features are used for the recognition. With this designed system, we are able to achieve 80 % recognition rate and initial results are encouraging.