Abstract : Despite the advance of functional exploration based on sophisticated medical imaging techniques, the digitization of anatomical surfaces still rely on manual imprecise and expensive clinical practice. From color images acquired with a hand held digital camera, an innovative tool for assessing chronic wounds has been developed. It combines both types of assessment, namely color analysis and dimensional measurement of injured tissue, in a user-friendly system, to provide the largest spreading in care staffs. Based on a ground truth established by clinicians, a sample database of wound tissue images has been constructed. These samples come from unsupervised color image segmentation after color correction to ensure stability under lighting conditions, viewpoint and camera type changes. They are characterized by color and texture descriptors, selected and re-sampled with data analysis techniques, before the learning stage of four categories of tissues of a Support Vector Machine classifier with perceptron kernel. The results of single view classification are merged and directly mapped on the mesh surface of the 3D wound model captured using uncalibrated vision techniques applied on a stereoscopic image pair. The result is a significative improvement in the robustness of the classification, equally stable over several reconstructions. The exact tissue areas are simply obtained by retro projection of the tissue regions on the 3D model. This geometric model is also strengthened since the automatic delineation of the wound uses skin detection to remove extra triangles from the mesh.