A Method to Determine the Characteristic of Object Based on 2D/3D Correspondance

摘要

This paper presents a method to retrieve semantic using combination of images and the corresponding 3D model reconstructed using photogrammetry. Our proposed method for determining the characteristics consists of two major steps. First, recognize characteristics based on the curvature of the point cloud and image processing of input photographies. Second, determine the characteristics of both spaces (2D and 3D) simultaneously. Our proposed method has the advantage of a combination of additional missing information on images and 3D model, to enhance detection and extraction of features. Experimental examples are presented to present the occurrely and the flexitricity of the method that can be applied to a large variety of objects.