采用多形状特征融合的多视点目标识别

李平; 魏仲慧; 何昕; 何丁龙; 何家维; 梁国龙; 凌剑勇

doi:doi:10.3788/ope.20142212.3368

光学精密工程, 2014, 22 (12): 3368, 网络出版: 2015-01-13

采用多形状特征融合的多视点目标识别

Object recognition based on shape feature fusion under multi-views

论文大纲

李平 ^1,2,*魏仲慧 ¹何昕 ¹何丁龙 ¹何家维 ¹梁国龙 ¹凌剑勇 ¹

作者单位

¹ 中国科学院长春光学精密机械与物理研究所, 吉林长春 130033

² 中国科学院大学, 北京 100039

摘要

研究了多视点下三维目标的识别问题。针对传统的采用单一特征的方法在目标描述方面的不足, 提出了一种融合多种特征的识别算法。首先, 利用各向异性高斯方向导数相关矩阵提取目标角点, 采用骨架约束提取特征角点, 将各特征角点到目标质心的归一化距离作为角点描述子。接着, 分别提取目标的几何矩不变量、仿射矩不变量、目标边界的傅里叶描述子; 计算4种特征的类内和类间散布矩阵; 以样本散布矩阵的迹作为权重, 加权融合4种特征。然后, 对融合后的特征向量进行独立成分分析(ICA), 得到相互独立的特征分量。最后, 采用支持向量机的分类方法进行分类。实验结果表明, 本文提出的方法比采用单一特征的方法的正确识别率平均提高10%以上, 且在小训练样本(10%总体样本)情况下仍能获得80%以上的识别率, 可满足经纬仪实时目标识别系统的要求。

Abstract

Three dimensional(3D) object recognition was researched under multi-view points. For the shortages of traditional signal feature description for 3D object recognition under multi-view points, a new recognition algorithm fusing multiple features was proposed. Firstly, the object corners were extracted by using the correlation matrixes of anisotropic Gaussian directional derivatives, the particular corners were selected by the skeleton constraint, and the normalized distance between particular corners and the object centroid was taken as the corner descriptor. Then, the geometric moment invariants, affine moment invariants, and the Fourier descriptor of object boundary were extracted, respectively , and the scatter matrixes within and between classes for the four features were calculated. By taking the trace of sample scatter matrix as the weight, the four features were fused. Furthermore, the Independent Component Analysis (ICA) was carried out on the fused vector to obtain independent features. Finally, a Support Vector Machine (SVM) was adopted to complete the whole classification of the experiments. Experimental results show that the recognition accuracy of the proposed approach is higher than that of the signal feature approach by 10% averagely and that in the small training sample (10% of the total samples) condition still achieves more than 80% .It concludes that proposed algorithm meets the demand of theodolites for real-time object recognition.

参考文献

[1] 曹健.显著角点特征在三维模型检索中的应用［J］. 计算机仿真, 2012, 29(12): 321-324.

CAO J. Application of salient corner features in 3D model retrieval ［J］. Computer Simulation, 2012, 29(12): 321-324. (in Chinese)

[2] ALYUZ N, GOKBERK B, AKARUN L. 3-D Face recognition under occlusion using masked projection ［J］. IEEE Transactions on Information Forensics and Security, 2013, 8(5): 789-802.

[3] 朱旭峰, 马彩文, 刘波. 基于特征级融合和支持向量机的飞机识别［J］. 光电子·激光, 2011, 22(11): 1710-1713.

ZHU X F, MA C W, LIU B. Aircraft recognition scheme based on feature fusion and support vector machine ［J］. Journal of Optoelectronics·Laser, 2011, 22(11): 1710-1713. (in Chinese)

[4] 丁昊, 李旭东, 赵慧洁. 基于SVDD的三维目标多视点视图建模［J］. 北京航空航天大学学报, 2012,38(11): 1517-1521.

DING H, LI X D, ZHAO H J. Method of multi-view modeling for 3D target based on SVDD ［J］. Journal of Beijing University of Aeronautics and Astronautics, 2012,38(11): 1517-1521. (in Chinese)

[5] 朱旭锋, 马彩文. 基于多不变量和多分类器融合的飞机识别［J］. 仪器仪表学报, 2011,32(7): 1621-1627.

ZHU X F, MA C W. Aircraft recognition based on multiple classifier fusion with multiple invariants［J］. Chinese Journal of Scientific Instrument, 2011, 32(7): 1621-1627. (in Chinese)

[6] 薛鹏飞, 胡玉兰. 小波和仿射不变矩的目标特征提取方法研究［J］. 沈阳理工大学学报, 2012, 31 (3): 11- 14.

XUE P F, HU Y L. Method of feature extraction of the target based on wavelet moment and affine invariant moment［J］. Journal of Shenyang Ligong University, 2012, 31(3): 11-14. (in Chinese)

[7] SINGH CH D,POOJA. Local and global features based image retrieval system using orthogonal radial moments［J］. Optics and Lasers in Engineering, 2012,50: 655-667.

[8] 易盟, 郭宝龙, 张旭. 基于复合Zernike矩相角估计的图像配准［J］. 光学精密工程, 2012, 20(5): 1117-1125.

YI M, GUO B L, ZHANG X. Image registration based on complex Zernike moment phase angle estimation ［J］. Opt. Precision Eng., 2012, 20(5): 1117 - 1125. (in Chinese)

[9] 杨云, 岳柱. 基于融合图像轮廓矩和Harris角点方法的遮挡人体目标识别研究［J］. 液晶与显示, 2013, 28(2): 273-277.

YANG Y, YUE ZH. Human body target recognition under occlusion based on fusion of image contour moment and harris angular points ［J］. Chinese Journal of Liquid Crystals and Displays, 2013,28(2): 273-277. (in Chinese)

[10] 许强, 马登武. 傅里叶描述子与角点相结合的形状匹配［J］. 光电工程, 2013,40(6): 123-128.

XU Q, MA D W. Shape matching combining corner with Fourier descriptor ［J］. Opto-Electronic Engineering, 2013, 40(6): 123-128. (in Chinese)

[11] 陈晓飞. 基于骨架的目标表示和识别技术研究［D］. 长沙: 国防科技大学, 2004.

CHEN X F. Research on object representation and recognition techniques based on skeleton ［D］. Changsha: National University of Defense Technology, 2004. (in Chinese)

[12] 王巍, 赵红蕊. 面向影响匹配的SUSAN角点检测［J］. 遥感学报, 2011,15(5): 948-956.

WANG W, ZHAO H R. The improvement of SUSAN for image matching ［J］. Journal of Remote Sensing, 2011,15(5): 948-956. (in Chinese)

[13] 储珺, 郭卢安政, 赵桂花. 采用环形模板的棋盘格角点检测［J］. 光学精密工程, 2013, 21(1): 189-196.

CHU J, GUOLU A ZH, ZHAO G H. Chessboard corner detection based on circular template［J］. Opt. Precision Eng., 2013, 21(1): 189-196.(in Chinese)

[14] SHUI P L,ZHANG W CH. Corner detection and classification using anisotropic directional derivative representations［J］. IEEE Transactions on Image Processing, 2013, 22(8): 3204-3218.

[15] HU M K. Visual Pattern recognition by moment invariants［J］. IRE Transaction on Information Theory, 1962, 8(2): 179-187.

[16] FLUSSER J,SUK T. Pattern recognition by affine moment invariants［J］. Pattern Recognition, 1993,26(1): 167-174.

[17] WU Y Y,WU Y Q. Shape-based image retrieval using combining global and local shape features［C］. CISP’09, 2nd International Congress on Image and Signal Processing Tian Jin, China, 2009: 1-5.

[18] SWETS D, WENG J. Using discriminant eigenfeatrues for image retrieval ［J］. IEEE Trans. on PAMI, 1996, 18 (8): 831-836.

[19] HYVARINEN A. Fast and robust fixed-point algorithms for independent component analysis［J］. IEEE Transactions on Neural Networks, 10(3): 626-634.

[20] EGGERT D,BOWYER K. Computing the perspective projection aspect graph of solids of revolution ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1993, 15(2): 109-128.

[21] SHILANE P, MIN P, KAZHAN M, et al.. Princeton shape benchmark［S］.2004, Available: http://shape.cs.princeton.edu/benchmark/.

[22] CHANG CH CH,LIN C J. LIBSVM: a library for support vector machines［EL］.2001, Software available at http://www.csie.ntu.edu.tw/-cjlin/libsvm

李平, 魏仲慧, 何昕, 何丁龙, 何家维, 梁国龙, 凌剑勇. 采用多形状特征融合的多视点目标识别[J]. 光学精密工程, 2014, 22(12): 3368. LI Ping, WEI Zhong-hui, HE Xin, HE Ding-long, HE Jia-wei, LIANG Guo-long, LING Jian-yong. Object recognition based on shape feature fusion under multi-views[J]. Optics and Precision Engineering, 2014, 22(12): 3368.

采用多形状特征融合的多视点目标识别

关于本站 Cookie 的使用提示

全站搜索

采用多形状特征融合的多视点目标识别

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索