Computing differential properties of 3-D shapes from stereoscopic images without 3-D models

摘要: We are considering the problem of recovering three-dimensional geometry a scene from binocular stereo disparity. Once dense disparity map has been computed pair images, one often needs to calculate some local differential properties corresponding 3-D surface such as orientation or curvatures. The usual approach is build reconstruction surface(s) which all shape will then be derived without ever going back original images. In this paper, we depart paradigm and propose use images directly compute properties. thus new method extending classical correlation estimate accurately both its derivatives image data. relate those >