This work describes identification of human skin diseases/cancer, namely, Actinic Keratosis, Eczema, Melanoma, Psoriasis, and Vascular Lesion using image processing technique. Colour models, namely, hue–saturation–intensity (HSI) and red–green–blue (RGB) are used for extracting eighteen colour features. The texture features are calculated using co-occurrence matrix, run-length matrix and wavelet decomposition techniques. A total of 34 features pertaining to texture are extracted. The identification of skin diseases is performed using classifier based on Linear Discriminant Analysis (LDA) and its performance is also compared to that of other statistical classifiers. The performance based on the above colour and texture feature extraction techniques towards identification of skin cancers are also compared. It is found that colour features yields better classification accuracy as compared to texture features based on run-length and co-occurrence matrices using LDA classifier. However, it is observed that wavelet features based on wavelet decomposed technique achieves overall best accuracy of 79% unlike other feature evaluation methods presented in this study. This study shows that identification of skin cancer based on RGB colour model achieves better average accuracy of 74% as compare to HSI model. This work also compares the performance of combined features (colour plus texture) and it is found that combination of wavelet and RGB colour features achieves classification accuracy up to 92% as compared to other colour and texture combination. Results also show that the performance of LDA classifier is comparatively better as compare to other classifier considered in this work.