In this research breast thermal infrared images have been color segmented for the first time. The color segmentation has been done with three clustering techniques: k - means, mean shift and fuzzy c - means. Furthermore , t wo breasts are separated automatically using a proposed algorithm . Since symmetry usually indicates healthy subjects, asymmetrical temperature distribution between left breast and right breast could be a strong indicator of abnormality. In addition, m utual infor mation between two breasts can be used as a measure of symmet ry betwee n two breasts. An algorithm to compute the mutual information between right breast .....| Key Words Thermography, F uzzy c - means , Mutual infor mation , Registration, Fractal Dimension, L yapunov exponents
