A dermoscopic image registration method based on adaptive SURF feature extraction and weighted kernel norm minimization fusion is proposed in this chapter, which is tried to overcome the multi-angle dermoscoptic images registration. The experiments are designed to show that the improved method have good rotation invariance, scale invariance, and affine invariance etc. and at the same time, this method can uniformly and stablly extract points such as moles,scars,and hair, and further can be effectively applied to different interferences.