Classification of dermoscopic images is an essential research topic as it may be advantageous in monitoring skin related problem, and detect the many type of skin cancer disease as soon as they appear. Therefore the need for fast, automatic, less expensive and accurate method to classify skin disease is of great realistic significance. This work aims to provide accurate, robust, reliable and automated dermoscopic image analysis technique, to allow for early detection of malignant melanoma disease. The optimal classifiers in light of using FFT transform with multilayer perceptron (MLP)Neural Network. An alternate Cross-Validation dataset is used for authentic appraisal of the proposed gathering computation with respect to basic execution measures, for instance, MSE and request accuracy. The Average Classification Accuracy of MLP Neural Network containing one covered layers with 9 PE's dealt with in an ordinary topology is seen to be unrivaled (97.22%) for Training and cross-validation. Finally, perfect count has been delivered dependent on the best classifier execution.

Neural solution, MatLab, Matlab program, Microsoft excel, ELM images.