New color texture features have been designed based on second-order statistics features that are calculated from a new color co-occurrence matrix (CCM). These CCM features have three main novel design aspects. First, they incorporate perceptual color differences in their computation. Second, the second-order probability distributions are based on the CIECAM16 color appearance model and its derived Uniform Color Space (CIECAM16-UCS). Third, to avoid high-dimensional and sparse cooccurrence matrices, low-dimensional CCMs are calculated using perceptual clustering for color quantization. The ability of these new CCM metrics to analyze colored textures has been validated in two experiments using biomedical color texture images: Basal Cell Carcinoma (BCC) dermoscopic pattern detection and hemangioma depth estimation from color photographs. Both experiments demonstrated that the proposed CCM features outperformed other texture analysis methods.