Abstract:In the last decades, 3D models have known a significant expansion, as we can see them in many fields such as computer vision, biology, augmented reality, engineering, and even medicine. They become more and more popular and accessible than ever. Due to these facts, there is an increasing need for a retrieval system, which can return the most similar 3D objects existing in the database to the query. In the present paper, a new retrieval approach is proposed, based on a multi-criteria method to generate a compact descriptor, which represents a signature for each 3D model. The key idea behind this approach is to try to extract the best out of each criterion (i.e., measure) by extracting a combined score using the Data envelopment analysis method (DEA). The results of the proposed method are very satisfactory and outperform some commonly used retrieval methods which highlight the potential and the performance of our approach.

3D mesh segmentation has become an essential step in many applications in 3D shape analysis. In this paper, a new segmentation method is proposed based on a learning approach using the artificial neural networks classifier and the spectral clustering for segmentation. Firstly, a training step is done using the artificial neural network trained on existing segmentation, taken from the ground truth segmentation (done by humane operators) available in the benchmark proposed by Chen et al. to extract the candidate boundaries of a given 3D-model based on a set of geometric criteria. Then, we use this resulted knowledge to construct a new connectivity of the mesh and use the spectral clustering method to segment the 3D mesh into significant parts. Our approach was evaluated using different evaluation metrics. The experiments confirm that the proposed method yields significantly good results and outperforms some of the competitive segmentation methods in the literature.

Decomposing a 3D mesh into significant regions is considered as a fundamental process in computer graphics, since several algorithms use the segmentation results as an initial step, such as, skeleton extraction, shape retrieval, shape correspondence, and compression. In this work, we present a new segmentation algorithm using spectral clustering where the affinity matrix is constructed by combining the minimal curvature and dihedral angles to detect both concave and convex properties of each edge. Experimental results show that the proposed method outperforms some of the existing segmentation methods, which highlight the performance of our approach.

No abstract

No abstract