Contrast Enhancement of Dermoscopic Images using Spatial and Frequency Domain Techniques

摘要

Melanoma is one of the deadliest forms of skin cancer with a high mortality rate. If detected in the early stages, melanoma can be easily treated. Dermoscopy is an imaging technique to diagnose melanoma in the early stages. One of the most important steps in the computer analysis of dermoscopic images is segmentation of the lesion from the healthy skin. Segmentation is a challenging task due to the lack of edges and low contrast in dermoscopic images. In this work, an attempt has been made to improve the contrast of dermoscopic images. The contrast enhanced dermoscopic images could be used to improve segmentation in dermoscopic images. Preprocessing is carried out to remove hair from the images. Contrast enhancement is performed in the frequency domain using a weighted multistage wavelet transform. Subsequently, the contrast is improved by stretching the histogram in the spatial domain. The results of the proposed technique are evaluated visually and quantitatively. An average increase of 21% is observed between the contrast enhanced images and the raw images. Image sharpness measure for melanoma images is found to be higher compared to non-melanoma images. The results obtained indicate that this technique could be used to increase the contrast of dermoscopic images.