摘要
本文针对传统的形状匹配算法的处理计算量过大、消耗时间过长,从而导致无法应用于大量的图像集以及在线的形状匹配场景的问题,在学者提出的距离融合算法的基础上进行了改进,在处理阶段引入无监督学习的方法进行多种聚类。通过引入预处理算法对图像集进行特征提取以及划分,在算法的计算量上做出优化,大幅降低了算法的计算时耗,并且保证其正确率几乎没有降低。
The problem of traditional shape matching algorithm is too large and the consumption time is too long,which can not be applied to a large number of image sets and online shape matching scenes. It is improved on the basis of the distance fusion algorithm proposed by scholars. Introduce unsupervised learning methods in the processing stage to perform multiple clustering. The feature extraction and division of the image set are introduced by introducing the preprocessing algorithm,and the calculation of the algorithm is optimized,which greatly reduces the computational time consumption of the algorithm and ensures that the correct rate is almost not reduced.
引文
[1]刘丽,匡纲要.图像纹理特征提取方法综述[J].中国图象图形学报,2009,14(4):622-635.
[2]Zhao Q,Cao J,Hu Y.Image Retrieval Based on Color-Spatial Distributing Feature[J].Journal of Southern Yangtze University,2007,346:79-86.
[3]Micha???giewka,Korytkowski M,Scherer R.Distributed image retrieval with color and keypoint features[C]//IEEE International Conference on Innovations in Intelligent Systems&Applications.IEEE,2017.
[4]Zhu X.Shape Recognition Based on Skeleton and Support Vector Machines[C]//International Conference on Intelligent Computing.Springer,Berlin,Heidelberg,2007.
[5]张桂梅,蔡报丰.基于内距离形状上下文特征的形状匹配研究[J].南昌航空大学学报(自然科学版),2016,30(2):1-7.
[6]BAI X,YANG X,LATECKI LJ,et al.Learning contextsensitive shape similarity by graph transduction[J].IEEE transactions on pattern analysis and machine intelligence,2010,32(5):861-74.
[7]BAI X,WANG B,YAO C,et al.Co-transduction for shape retrieval[J].IEEE Transactions on Image Processing,2012,21(5):2747-2757.