Melanoma is the most serious form of skin cancer that affects millions of people globally. Through image analytics, early identification of skin cancer is enabled, resulting in more effective treatment and a lower mortality rate. The ph2 and human against machine datasets were used to collect images. After preprocessing the image with a weighted median filter, segmentation is investigated using a number of common techniques, with the best result generated by combining watershed transform and maximum similarity region merging. U-net architecture is explored for segmentation. Segmentation efficiency is calculated by dice loss and Jaccard coefficient. Segmentation architecture outperform the conventional method. Additionally, a novel wavelet transform-based approach is used to extract features, followed by local ternary pattern analysis. The intersection of the histograms, the Bhattacharya distance, the Chi-square distance, and the Pearson correlation coefficients are all computed. This inquiry makes use of only the Histogram intersection and Chi-square distance characteristics. Additional categorization is examined through the use of a range of machine learning algorithms, including the k-nearest neighbour approach, Bayesian classification, decision trees, and Support Vector Machines (SVM). When a Radial Basis Function (RBF) kernel based SVM is applied, the classification accuracy is maximised. This work is entirely devoted to binary categorization. As evidenced by the data, they outperform other state-of-the-art approaches reported in the literature. SVM classifies data with an accuracy of 98.6 percent. Weighted median filter, Watershed transform, Merging regions with the highest degree of similarity, Wavelet transform, Local Ternary Pattern, Histogram intersection Pearson correlation coefficient, chi-square distance Distance between Bhattacharya and support vector machine.