Considering the problems(complicated computation and huge data) of SEBVHOS(surface extraction from binary volumes with high-order smoothness),we propose a parallel algorithm to accelerate the SEBVHOS execution.Firstly,SEBVHOS is parallelized.Secondly,optimization techniques are applied to improve performance of the algorithm.Meanwhile,matrix compression is applied to improve performance of memory space.Experiments show that the average speed-up ratio achieves 1.87 in a dual-core system.

(Abstract )Feature extraction in the multi(view environment is an important step in 3D reconstruction. However, it is a very time(consuming task. To accelerate the speed of extracting feature, this paper presents a parallel method to extract feature in the multi(view environment. The key views are selected by the texture feature discrepancy which is computed based on the grey level grows matrix. Harris corner detection algorithm and Blob detection algorithm are adopted to extract feature of the key view images. The method is parallelized by exploring the inherent parallelism of proposed procedures. Experimental results show that the method can select the key views efficiently, and the average speedup achieves 1.88 on a dual(core system. (Key words )feature extraction; parallel algorithm; multi(view image; 3D reconstruction; texture feature

This paper proposes a three-dimensional Reeb graph calculation of human body.Vertex coordinates of three-dimensional mesh data are employed to calculate vertex geodesic distance and construct Morse function,and then human body three-dimensional Reeb graph can be extracted according to vertex triangular relationship.It adds the key joint nodes,which can more accurately describe the topology of three-dimensional model,by calculating the arc of Reeb graph nodes.Experimental results show that the method is simple and of wide applicability.

In this paper, a parallel Surface Extraction from Binary Volumes with Higher-Order Smoothness (SEBVHOS) algorithm is proposed to accelerate the SEBVHOS execution. The original SEBVHOS algorithm is parallelized first, and then several performance optimization techniques which are loop optimization, cache optimization, false sharing optimization, synchronization overhead optimization, and thread affinity optimization, are used to improve the implementation’s performance on multi-core systems. The performance of the parallel SEBVHOS algorithm is analyzed on a dual-core system. The experimental results show that the parallel SEBVHOS algorithm achieves an average of 1.86x speedup. More importantly, our method does not come with additional aliasing artifacts, comparing to the original SEBVHOS algorithm.

Human activity analysis and identification now becomes a hot issue in the area of computer vision.3D body segmentation plays an important role in body analysis and identification.In this paper,a body segmentation algorithm based on Reeb graph is proposed,which could effectively achieve human body segmentation.Reeb graph is extracted from a reconstructed human body,and then the Reeb graph skeleton is segmented.The corresponding relationship between reconstructed human body and Reeb graph framework is calculated,and inverse mapping is conducted,thus to achieve 3D body segmentation.