摘要
针对视觉目标中存在的遮挡现象,提出一种基于谱聚类实现深度图像遮挡边界检测的方法。首先定义一种新的遮挡相关特征——有效标准差特征,基于相关特征利用均卡方集距抽取部分像素点,构建相似矩阵;然后基于相似矩阵利用Nystrom逼近方法近似估算全部像素点的拉普拉斯矩阵与逼近特征向量,对得到的逼近特征向量进行聚类分析,把深度图像中的全部像素点划分为遮挡边界点和非遮挡边界点两大类;最后可视化遮挡边界点得到深度图像中的遮挡边界。实验结果表明,本文方法无需标记样本,且在深度图像中目标物体的遮挡边界检测方面具有较好的有效性和普适性。
Aiming at the occlusion phenomenon in the visual object,we propose a novel occlusion boundary detection approach for deep images based on the spectral clustering.Firstly,a new occlusion-related feature,effective standard deviation feature,is defined.Secondly,some pixels are extracted by using mean chi-square set distance,and the similarity matrix is constructed based on the occlusion-related feature.Thirdly,the Laplacian matrix of all the pixels and approximation eigenvectors are approximated by Nystrom approximation method based on the similarity matrix.Then,the obtained approximation eigenvectors are clustered to divide all the pixels in the depth image into two categories,namely the occlusion boundary points and non-occlusion boundary points.Finally,the occlusion boundary of the depth image is obtained by visualizing occlusion boundary points.Experimental results show that the proposed method which does not need any labeled samples has good effectiveness and generality for occlusion boundary detection of the object in the depth image.
引文
[1]Chen P G,Zhang X M,Yuen P C,et al.Combination of spatio-temporal and transform domain for sparse occlusion estimation by optical flow[J].Neurocomputing,2016,214:368-375.
[2]Chen X W,Li Q,Zhao D Y,et al.Occlusion cues for image scene layering[J].Computer Vision and Image Understanding,2013,117(1):42-55.
[3]Chen C,Corso J J.Joint occlusion boundary detection and figure/ground assignment by extracting commonfate fragments in a back-projection scheme[J].Pattern Recognition,2017,64:15-28.
[4]Cheng H M,Tseng C Y,Hsin C H,et al.Singleimage 3-D depth estimation for urban scenes[C].2013 20th IEEE International Conference on Image Processing,2013:2121-2125.
[5]Fu H,Wang C,Tao D,et al.Occlusion boundary detection via deep exploration of context[J].Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016:241-250.
[6]Zhang S H,Zhang Y J,Kong L F.Self-occlusion detection approach based on depth image[J].Journal of Chinese Computer Systems,2010,31(5):964-968.张世辉,张煜婕,孔令富.一种基于深度图像的自遮挡检测方法[J].小型微型计算机系统,2010,31(5):964-968.
[7]Zhang S,Gao F,Kong L.A self-occlusion detection approach based on range image of vision object[J].ICIC Express Letters,2011,5(6):2041-2046.
[8]Zhang S H,Liu J X.A self-occlusion detection approach based on depth image using SVM[J].International Journal of Advanced Robotic Systems,2012,9(6):230.
[9]Zhang S H,Liu J X,Kong L F.Using random forest for occlusion detection based on depth image[J].Acta Optica Sinica,2014,34(9):0915003.张世辉,刘建新,孔令富.基于深度图像利用随机森林实现遮挡检测[J].光学学报,2014,34(9):0915003.
[10]Zhang S H,Pang Y C.OccIusion boundary detection method for depth image based on ensemble learning[J].Acta Metrologica Sinica,2014,35(6):569-573.张世辉,庞云冲.基于集成学习思想的深度图像遮挡边界检测方法[J].计量学报,2014,35(6):569-573.
[11]Zhang S H,Zhang Y C,Zhang H Q,et al.Occlusion boundary detection using graph-based semisupervised learning[J].Acta Metrologica Sinica,2016,37(6):576-581.张世辉,张钰程,张红桥,等.基于图的半监督学习的遮挡边界检测方法[J].计量学报,2016,37(6):576-581.
[12]Liu G,Wang X.Adaptive semi-supervised spectral clustering based on nystr9m method[C].2010 3rd International Congress on Image and Signal Processing,2010:524-528.
[13]Bai X D,Cao Z G,Wang Y,et al.Image segmentation using modified SLIC and Nystr9m based spectral clustering[J].Optik,2014,125(16):4302-4307.
[14]Fu X,Martin S,Mills S,et al.Improved spectral clustering using adaptive mahalanobis distance[C].2013 2nd IAPR Asian Conference on Pattern Recognition(ACPR),2013:171-175.
[15]Fowlkes C,Belongie S,Chung F,et al.Spectral grouping using the Nystrom method[J].IEEETransactions on Pattern Analysis and Machine Intelligence,2004,26(2):214-225.
[16]Dong A G,Li J X,Zhang B,et al.Hyperspectral image classification algorithm based on spectral clustering and sparse representation[J].Acta Optica Sinica,2017,37(8):0828005.董安国,李佳逊,张蓓,等.基于谱聚类和稀疏表示的高光谱图像分类算法[J].光学学报,2017,37(8):0828005.
[17]Rosten E,Drummond T.Machine learning for highspeed corner detection[C].European Conference on Computer Vision,2006:430-443.