Medição precisa de áreas de feridas crônicas usando Structure from Motion

Abstract

Chronic wounds are ulcers that have a difficult or almost interrupted healing process, leading to an increased risk of health complications, such as amputations and changes. The need for quantitative areas is of great importance in clinical trials, pathological analysis of wounds and daily patient care. Manual and 2D manuals cannot solve the problems caused by the curvatures of the human body and different camera angles. This work proposes the calculation of the area of wounds through the use of a non-invasive methodology, performing a 3D reconstruction of the surface of the human body. This combined work of images and Structure from Motion (SfM) using different feature descriptors (SIFT, SURF, BRIEF and ORB). The results show that the precise measurements of 3D surface areas can be adapted from images acquired with a smartphone using the proposed methodology. For the different mean error descriptors, they were 1.7 % for SIFT, 3.6 % for SURF, 6.4 % for ORB and 20.8 % for BRIEF, using a setting of 10 images, while the average error for while using a single 2D image was 32.7%.