Features derived from Grey Level Co-occurrence Matrix (GLCM) and Grey Level Run-Length (GLRL) matrix are widely used for image characterization based on texture analysis. In this paper, we propose the application of suitably selected texture discriminating features for classification of oral cancer lesions in digital camera images into six groups. Backpropagation based Artificial Neural Network (BPANN) is used to compare and validate the performance of different feature sets. The classification accuracy is observed to improve with combination of GLCM, GLRL and intensity based first order features. Further improvement in accuracy is obtained by application of feature selection using boxplot analysis. A set of 61 features is formulated and applied on 192 sections of images taken from 16 patients. Such a classification of malignancies is helpful in prognosis and treatment of oral cancer which is the most common form of cancer in India.