基于KNN有向复杂网络的图像轮廓识别
|
摘要
图像的目标识别是模式识别的研究领域之一,现已广泛应用于视频监控、交通运输和动作识别等。受图像采集过程中光照变化、形状和噪声等因素影响,基于区域或轮廓的方法往往会出现若干错误。图像的复杂网络特征具有较强的稳定与抗噪能力,因此,提出一种图像的有向复杂网络表示模型,利用K近邻(KNN)确定有向复杂网络的演化序列,并利用复杂网络的度平均与熵等参数完成图像的轮廓识别。图像检索实验结果表明,该方法在查全率与查准率上均获得较好结果。
The object recognition in image is one of the research fields of pattern recognition, which is widely used in video surveillance, transportation and motion recognition. There are some errors in the region or contour based methods due to the influence of illumination change, shape and noise during the image acquisition. The characteristics of complex network for image have the abilities of strong stability and anti-noise. In this paper, a directed complex network representation model of image is proposed. The evolutionary sequence of directed complex network has been determined with K-nearest neighbor(KNN), and the contour recognition of image is completed via the parameters of average degree and entropy of complex network. The results of image retrieval experiment show that the proposed method obtains the better results in terms of recall and precision.
The object recognition in image is one of the research fields of pattern recognition, which is widely used in video surveillance, transportation and motion recognition. There are some errors in the region or contour based methods due to the influence of illumination change, shape and noise during the image acquisition. The characteristics of complex network for image have the abilities of strong stability and anti-noise. In this paper, a directed complex network representation model of image is proposed. The evolutionary sequence of directed complex network has been determined with K-nearest neighbor(KNN), and the contour recognition of image is completed via the parameters of average degree and entropy of complex network. The results of image retrieval experiment show that the proposed method obtains the better results in terms of recall and precision.
引文
[1] Backes A R, Casanova D,Bruno O M. A complex network-based approach for boundary shape analysis[J].Pattern Recognition,2009.42(1):54-67
[2] Gao X B, Xiao B, Tao D C, et al. Image categorization:Graph edit distance edge direction histogram[J].Pattern Recognition,2008.41(10):3179-3191
[3] Neuhaus M, Bunke H. A probabilistic approach to learning costs for graph edit distance[C]//Proceedings of the17th International Conference on Pattern Recognition,2004:389-393
[4] Xiao B, Li J, Gao X B. An HMM-based cost function free algorithm for graph edit distance[C]//Proceedings of International Conference on Visual Information Engineering,2008:286-291
[5] Luo B, Wilson R C, Hancock E R. Spectral embedding of graphs[J].Pattern Recognition,2003.36(10):2213-2230
[6] Tang J, Jiang B, Chang C, et al. Graph structure analysis based on complexnetwork[J]. Digital Signal Processing,2012.22(5):713-725
[7] Bai X,Wang B, Wang X G, et al. Co-transduction for shape retrieval[J].IEEE Transactions on Image Processing,2012.21(5):2747-2757
[7] Bhowmik M K, Shil S, Saha P. Feature Points Extraction of Thermal Face using Harris Interest Point Detection[J].Procedia Technology,2013.10:724-730
[8] Shu X'Wu X J. A novel contour descriptor for 2D shape matching and itsapplication to image retrieval[J].Image and Vision Computing,2011.29(4):286-294
[2] Gao X B, Xiao B, Tao D C, et al. Image categorization:Graph edit distance edge direction histogram[J].Pattern Recognition,2008.41(10):3179-3191
[3] Neuhaus M, Bunke H. A probabilistic approach to learning costs for graph edit distance[C]//Proceedings of the17th International Conference on Pattern Recognition,2004:389-393
[4] Xiao B, Li J, Gao X B. An HMM-based cost function free algorithm for graph edit distance[C]//Proceedings of International Conference on Visual Information Engineering,2008:286-291
[5] Luo B, Wilson R C, Hancock E R. Spectral embedding of graphs[J].Pattern Recognition,2003.36(10):2213-2230
[6] Tang J, Jiang B, Chang C, et al. Graph structure analysis based on complexnetwork[J]. Digital Signal Processing,2012.22(5):713-725
[7] Bai X,Wang B, Wang X G, et al. Co-transduction for shape retrieval[J].IEEE Transactions on Image Processing,2012.21(5):2747-2757
[7] Bhowmik M K, Shil S, Saha P. Feature Points Extraction of Thermal Face using Harris Interest Point Detection[J].Procedia Technology,2013.10:724-730
[8] Shu X'Wu X J. A novel contour descriptor for 2D shape matching and itsapplication to image retrieval[J].Image and Vision Computing,2011.29(4):286-294