A Metrical Approach to Classification of the Object Modelling with Point Cloud Data

27853

Object classification using 3D point cloud data is an emerging issue in the field of research. Objectdetection and recognition are some of the main problems in the computer graphics works. In this study, we propose a novel method with a metrical approach for effectively detecting and classification of 3D point cloud data from real objects. The efficiency of the method is validated according to the real 3D point cloud data. The point cloud data and the combination of metric information has been implemented in a software framework to improve the results of the classification phase. Robust segmentation and metrical feature extractions were evaluated with different objects which markedly separated from each other for classification of 3D point cloud data. Success is evaluated by comparing the raw data of different objects with classified metric data which executed classification process. In the evaluation, the metric classification algorithm has been shown to be efficient process for the automatic classification of different objects which have a simple planimetric shape.

128

Point cloud , Classification , Kinect , Object recognition , 3D segmentation , Modelling

Afyon Kocatepe University Journal Of Science an‎d Engineering

136

Afyon Kocatepe University Journal Of Science an‎d Engineering