Spectral Clustering Bipartite Graph Segmentation Method Based on Color Invariant Features

doi:10.3969/j.issn.1671-1122.2018.12.002

Abstract

Abstract:

The segmentation results of segmentation methods based on spectral clustering are significantly affected by the performance of superpixels clustering. However, the performance of superpixels clustering mostly depends on the construction of affinity model. Bipartite graph segmentation framework with cross-affinity matrix makes superipixels clustering more efficient, but its affinity model uses simple color features without considering the effect of illuminant changes such as highlights, shading et al, which may result in failed object segmentation. To improve the coherence of superpixels clustering, this paper uses color descriptor with color invariant features and Ridge feature which reflects physical reflection of imaging surface to construct cross-affinity model. Based on the validation in Berkeley database, the spectral clustering segmentation method based on color invariant features achieves better performances compared to existing segmentation techniques.

Key words: spectral clustering, superpixels, bipartite graph segmentation, color invariance

CLC Number:

TP309

Wei ZHAO, Na ZHAO, Yixing ZHANG. Spectral Clustering Bipartite Graph Segmentation Method Based on Color Invariant Features[J]. Netinfo Security, 2018, 18(12): 8-14.

Figures/Tables 7

References 17

[1]	COMANICIU D, MEER P.Mean shift: A Robust Approach toward Feature Space Analysis[J]. Pattern Analysis and Machine Intelligent, 2002,5(7): 603-619.
[2]	ZHAO Feng, FAN Jiulun, ZHI Xiaobin, et al.Spatial and Gray Feature-based Spectral Clustering for Image Segmentation[J].Journal of Xi’an Institute of Posts and Telecommunications, 2012, 17(1):53-57.
	赵凤,范九伦,支晓斌,等.基于灰度和空间特性的谱聚类图像分割[J]. 西安邮电学院学报, 2012, 17(1):53-57.
[3]	LI Zhengguo, WU Xiaoming, CHANG S F.Segmentation Using Superpixels: A Bipartite Graph Partitioning Approach[C]// IEEE.2012 IEEE Conference on Computer Vision and Pattern Recognition, June 16-21, 2012, Providence, RI, USA.NJ:IEEE, 2012: 789-796.
[4]	KIM T H, LEE M, LEE S U.Learning Full Pairwise Affinities for Spectral Segmentation[J]. Pattern Analysis and Machine Intelligent, 2013, 35(7):1690-1730.
[5]	YIN Fang, CHEN Deyun, WU Rui.Improved Method of Image Segmentation Using Spectral Clustering[J]. Computer Engineering and Applications,2011,47(21):185-187.
	尹芳,陈德云,吴锐.改进的谱聚类图像分割方法[J].计算机工程与应用,2011,47(21):185-187.
[6]	GAO Shangbing, ZHOU Jingbo, YAN Yunyang.A New Superpixel-based Spectral Clustering for Image Segmentation[J].Journal of Nanjing University: Natural Sciences,2013,49(2):169-175.
	高尚兵,周静波,严云洋.一种新的基于超像素的谱聚类图像分割算法[J].南京大学学报:自然科学版,2013,49(2):169-175.
[7]	BOYKOVY,FUUKA-LEA G. Graph Cuts and Efficient N-D Image Segmentation[J].Journal of Computer Vision, 2006, 70(2): 109-131.
[8]	GAO Hongtao, LU Wei, YANG Yuwang.An Adaptive Adjusting Kernel Function-Based Extraction Method for Image Salient Area[J].Netinfo Security,2018,18(2):54-60.
	高洪涛,陆伟,杨余旺.基于自适应调节核函数的图像显著区域提取方法[J].信息网络安全,2018,18(2):54-60.
[9]	SERRA M, PENACCHIO O, BENAVENTE R, et al.Names and Shades of Color for Intrinsic Image Estimation[C]// IEEE.2012 IEEE Conference on Computer Vision and Pattern Recognition, June 16-21, 2012, Providence, RI, USA.NJ:IEEE, 2012: 278-285.
[10]	KIM T H, Lee K M, LEE S U.Learning Full Pairwise Affinities for Spectral Segmentation[C]// IEEE. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 13-18, 2010, SanFrancisco, CA, USA.NJ:IEEE,2010: 789-796.
[11]	VAZQUEZ E, Baldrich R, WEIJER J V D,et al. Describing Reflectances for Colour Segmentation Robust to Shadows, Highlights and Textures[J]. Pattern Analysis and Machine Intelligent, 2007, 6(1):1-15.
[12]	NG A Y, JORDAN M I, WEISS Y.On Spectral Clustering: Analysis and an Algorithm[C]//NIPS. The 14th International Conference on Neural Information Processing Systems: Natural and Synthetic, December 3 - 8, 2001, Vancouver, British Columbia, Canada. Cambridge: MIT Press , 2001:849-856.
[13]	YAO Xiaoyu, MA Hui, LIAN Zhe.A Method of Anti-Antivirus Based on Feature Decomposition[J]. Netinfo Security, 2012,12(4):43-46.
	姚晓宇,马辉,廉喆.基于特征分解的免杀方法研究[J].信息网络安全, 2012,12(4):43-46.
[14]	BERLIN B, KAY P.Basic Color Term: Their University and Evolution [M]. Berkeley: University of California Press, 1969.
[15]	UNNIKRISHNAN R, PANTOFARU C, Hebert M.Toward Objective Evaluation of Image Segmentation Algorithms[J]. Pattern Analysis and Machine Intelligent, 2007, 29(6):929-944.
[16]	MARTIN D, FOWLKES C, TAL D, et al.A Database of Human Segmented Natural Images and Its Application to Evaluating Segmentation Algorithms and Measuring Ecological Statistics[C]// IEEE. The 8th IEEE International Conference on Computer Vision, July 7-14, 2001,Vancouver, BC, Canada.NJ:IEEE,2001: 416-423.
[17]	COUR T, BENEZIT F, SHI Jianbo.Spectral Segmentation with Multiscale Graph Decomposition[C]// IEEE. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 20-25, 2005, San Diego, CA, USA.NJ:IEEE,2005:24-32.

评估指标算法	PRI	GCE
FH	0.7139	0.1746
FNcut	0.8146	0.1809
MNcut	0.7559	0.1925
SAS	0.8266	0.1868
本文方法	0.8315	0.1754