In this paper, effective approach for early detection of skin lesion using segmentation and classification is proposed. Skin images contains distortion in terms of unwanted hairs and noise initially, hence filtering is required in order to remove noise from the image. Segmentation is the next step used to extract lesion area. Segmentation yield parameters such as Geometric Mean, Standered Deviation, Kurtosis, Moments etc. Results of segmentation are compared against various well known measures and result obtained is commendable. Support vector machine and fuzzy neural technique is used for classification. Results show significant improvement over SVM+KNN. Accuracy is improved by 45%, F-Score is improved by 40%, Precision is improved by 45% and recall is also improved by 45%.