ABSTRACT
Cancer is one of the major health concerns with skin cancer playing a significant contribution. Skin cancer can be broadly classified into three: Basel cell carcinoma, squamous cell carcinoma and melanoma. Among this melanoma is the most fatal which spreads through metastasis. It can be completely cured if detected at early stages. This work presents a non-invasive method for detecting skin cancer at early stages where images are captured using a smart phone at real time. The proposed system mainly consists of four stages: Pre-Processing, Segmentation, Feature extraction and Classification. Pre-Processing and segmentation are performed using LoG SOBEL algorithm and K-Means respectively. Features extracted in this method are statistical features, color features and texture features. The extracted features are given to a SVM classifier which classifies the image into benign or cancerous. This method allows users to identify skin cancer at early stages.
