Color Space Analysis for Iris Recognition

Color Space Analysis for Iris Recognition by Matthew K. Monaco Master of Science in Electrical Engineering West Virginia University Arun A. Ross, Ph.D., Chair This thesis investigates issues related to the processing of multispectral and color infrared images of the iris. When utilizing the color bands of the electromagnetic spectrum, the eye color and the components of texture (luminosity and chromaticity) must be considered. This work examines the effect of eye color on texture-based iris recognition in both the near-IR and visible bands. A novel score level fusion algorithm for multispectral iris recognition is presented in this regard. The fusion algorithm based on evidence that matching performance of a texture-based encoding scheme is impacted by the quality of texture within the original image ranks the spectral bands of the image based on texture quality and designs a fusion rule based on these rankings. Color space analysis, to determine an optimal representation scheme, is also examined in this thesis. Color images are transformed from the sRGB color space to the CIE Lab, YCbCr, CMYK and HSV color spaces prior to encoding and matching. Also, enhancement methods to increase the contrast of the texture within the iris, without altering the chromaticity of the image, are discussed. Finally, cross-band matching is performed to illustrate the correlation between eye color and specific bands of the color image. iii I dedicate this work to my friends and family.