Combination of Weber, oriented gradient, fourier and color features for 3D textured matching
Hero Yudo Martono · 2016
3D object retrieval is one popular research domain in last years. One of challenges is how to determine similarity between 3D objects. Some methods are already proposed and some of them are based on image processing. Inspired by promising recent results from image processing and the reality that no single descriptor is powerfull and combination of different descriptors usually generate improved performance over any single methods then in this paper we implement combination of Weber, Oriented Gradient, and fourier transform to extract global 2D image descriptor. To get appropriate result, 3D pose normalization is processed for getting normal pose. 2D Image is obtained by projecting 3D object from normal view point. Since 3D object came with color and texture, color histrogram is choosen as color descriptor. Finally, all descriptors are merged by linear combination. Experiment result shows that this combination provides competitive performance and able to be 3D object descriptors.