Image stitching, color correction and superresolution of a video sequence for dermatological analysis

Abstract

Skin cancer is the most common cancer on humans, some kinds of skin cancer can be deadly if not addressed correctly and in time. The diagnosis is usually stated based on the appearance of the lesions in the skin. This paper describes how to apply stitching techniques to images extracted from a video acquired with a smartphone. The method has been implemented in Matlab, and it has as objective to obtain a panoramic image of the patient skin. The goal of this project is to overcome to the resolution limitation of a typical smartphone, by applying super-resolution and colour correction techniques. The quality of the result is measured qualitatively using different methods at the merging of the images. The use of VLFeat library is required to develop and follow the steps explained in this paper, the compilation of this library and how to address any error in the installation process. The panoramic image does not provide any diagnosis of the illness but is meant as a pre-processing step for an automatic tool aimed at helping the dermatologist to diagnose skin cancer.