Algorithm and Experiment of Continuous Contour Extraction for Gesture without Redundant Background

Yu ZHANG,Weibin XU,Jingwei LU,Guangyi WANG
DOI: https://doi.org/10.3969/j.issn.1006-7167.2017.10.003
2017-01-01
Abstract:Continuous contour extraction plays a significant role in pattern recognition and reconstruction of 3D image.However,the existing contour extraction methods cannot obtain the continuous contour,so it will influence the subsequent image quality seriously.The continuous contour extraction algorithm of gesture is proposed in this paper.Firstly,the Gaussian mixed model is utilized to locate the gesture region in the video image.Then Sobel operator is employed to extract the rough contour of gesture in the selected region.Next,the depth-first algorithm is used to realize contour traversal,and ascertain the longest contour as the gesture contour.Finally,by calculating the gradient and analyzing the relative position of the breaking point in gesture contour,the local Sobel operator or line connection will be adopted to connect the breaking point.The experimental results have shown that the proposed algorithm can get better subjective visual effect and objective performance standard than the traditional methods.Contour integrity can reach to 98%,while contour purity is close to 0.
What problem does this paper attempt to address?