Efficient segmentation of skin lesion is necessary for computational methods of identification and classification. Clinical segmentation by dermatologists, however, has a margin of error and substantial inter- and intra-individual variability, making it insufficiently accurate for segmentation investigations. By solving these constraints, this study makes it possible to extract nonlinear features of region of interest border lines and conduct a comprehensive examination of the geometry of the lesion. There are a variety of methods used for identifying skin lesions in images, such as morphological operations, clustering techniques, edge-based segmentation, region-growing approaches, threshold-based methods, and active contour models. Preprocessing and segmentation are the two primary methods in the proposed strategy. The segmentation phase uses Otsu thresholding and K-mean clustering combined with color attributes learned from the training images to improve the boundaries of the segments. In the preprocessing step, noise like illumination, hair, and rulers are eliminated using filtering techniques. To evaluate the proposed method, ISIC 2018 challenge dataset is used.