Detection of Melanoma Skin Cancer Using Image Processing Techniques

The most dangerous type of human skin cancer is malignant melanoma, and its incidence has increased rapidly. Early detection of malignant melanoma in dermoscopic images is very important and critical as it can be helpful to cure it in the early stage. Computer Aided Diagnosis programs can be very helpful in helping dermatologists to identify cancers early. System uses a new approach to diagnose melanoma skin cancer in this proposed system. To detect hair and several noise from pictures, pre-processing stage is done by adding a bank of directional filters and thus Image processing method is applied to fill the unknown regions. Program has checked the precision, accessibility, specificity of the publicly available PH2 dataset. It is found that good results are produced using the principle of repetition on applications to eliminate the same function or picture during the training and testing process, thereby confirming the validity of the method suggested.