Cancer is considered to be a fatal disease. In present era, skin cancer known as melanoma appears on skin. Shape, border, intensity, and size of cancer cells play an important role in detecting cancer. Segmentation separates objects and analyzes each region separately to infer what it is. Suspicious lesion extracted from segmentation is further used for feature extraction to identify melanoma at earlier stages. However, selecting an apt segmentation technique for various dataset images is a major challenge in the medical field. In this paper, we have derived a new method using K-means clustering and Canny edge detection which detects the edges and borders of lesion accurately. This research derives an improved method of segmentation with perfect edge detection especially for low contrast, small lesion, and irregular border images. This method is evaluated on parameters such as MSE and PSNR which achieved better results. On an average MSE values are reduced by 99%, and PSNR values are increased by 70%.