Performance Analysis of Machine Learning Based Classifiers for the Diagnosis of Melanoma Cancer and Comparison
Melanoma cancer is the most injurious form of cancer which affects the human. Skin cancer has quickly increased in western part of the country among the world. In this paper, diagnosing melanoma in premature stages a detection system has been designed which contains the following digital image processing techniques. First, dermoscopy image of skin is taken, and it is subjected to the pre-processing step for noise removal and post-processing step for image enhancement. Then the processed image undergoes image segmentation using Otsu method and Morphological processing. Second, features are extracted using feature extraction technique-ABCD parameter, GLCM, and FOS. Various feature combinations are given as the input to the KNN, SVM, ANN and Bag of Visual Words classifiers. KNN classifier is used to classify the data set into two classes, SVM classifier is used to classify the data set into three classes, ANN classifier is used to classify the data set based on the number of layers and Bag of Visual Words are used to classify the data set into two classes. Performance is analyzed based on the accuracy of the learning classifier output.