﻿ 基于联合多特征字典稀疏表示的步态识别算法 Gait Recognition Algorithm Based on Sparse Representation of Joint Multi-Feature Dictionary

Computer Science and Application
Vol.07 No.04(2017), Article ID:20452,9 pages
10.12677/CSA.2017.74048

Gait Recognition Algorithm Based on Sparse Representation of Joint Multi-Feature Dictionary

Xin Hu, Xiaohong Wu, Xiang Lei, Xiaohai He

School of Electronic Information, Sichuan University, Chengdu Sichuan

Received: Apr. 13th, 2017; accepted: Apr. 25th, 2017; published: Apr. 30th, 2017

ABSTRACT

Most of the existing gait recognition algorithms extract the single feature using model features or global features. However, these algorithms usually have a poor robustness and a low recognition rate in practical situations such as multi-angle. To solve this problem, a gait recognition algorithm based on joint sparse representation of multi-feature dictionaries is proposed in this paper. In this algorithm, three characteristics in different particle size are selected: Procrustes Mean Shape, Gait Energy Image and Region Area Sequence which is structured in this article. Feature training dictionaries are constructed and a multidisciplinary sparse representation to feature samples is done. Finally, the test sample category is obtained by calculating the minimum cumulative residual and achieves the integration of feature layer. Experimental results show that the multi-feature joint recognition method used in this paper has a higher recognition rate and a certain robustness at multiple angles compared to single feature extraction and recognition. This paper basically fulfills the complementary information between features.

Keywords:Gait Recognition, Joint Sparse, RAS

1. 引言

2. 多特征提取

2.1. PMS特征提取

Procrustes均值形状(PMS)分析法是方向统计学中一种广泛使用的方法，通常适用于编码二维形状，并提供了一种好的工具来寻找一组图形的均值形状，因此它可以用来描述人体步态特征 [8] 。

Step 1. 采用Canny算子提取图像步态轮廓，将所有轮廓点表示为以人体质心为坐标原点的极坐标值，并以等间隔的角度对轮廓线进行采样，得到轮廓的一维极坐标向量表示。

Step 2. 通过极坐标到直角坐标的逆变换将每个形状表示为一个复数向量。其中为采样后的轮廓坐标，为采样频率。为将形状置于坐标空间中心位置，定义中心向量，其中

Step 3. 当序列中含有帧图像时，可得到个中心配置向量，计算配置矩阵，得到矩

Step 4. PMS的均值形状对应的最优配置，即最大特征值对应的特征向量，将其作为步态序列的统计特征用于识别。

2.2. GEI特征提取

Figure 1. The mean shape of the three goals

 (1)

2.3. RAS特征提取

(2)

Figure 2. Gait sequences and energy maps

Figure 3. Dimensionality reduction by PCA

Figure 4. Feature extraction by RAS

，则第副步态能量图的区域面积特征向量，对于包含幅图像的步态序列，其RAS特征向量表示为

3. 基于联合多字典稀疏表示的步态识别

(3)

(4)

Figure 5. RAS features

Figure 6. The RAS feature corresponding to gait sequence

Figure 7. The framework of gait recognition algorithm based on joint multi- dictionary sparse representation

4. 实验步骤与结果分析

4.1. 实验步骤

Step 1. 将测试样本图像规整到150 * 90大小，提取RMS、GEI与RAS特征；

Step 2. 采用PCA方法对GEI与RAS特征进行降维处理，对每种特征采用最近邻分类器计算单一特征下的识别率；

Step 3. 采用联合多特征字典稀疏表示方法，计算三个特征融合下的识别率，与单一特征下的识别结果进行比较；

Step 4. 调整PCA降维数，测试维数对识别结果的影响；

Step 5. 改变稀疏度，测试稀疏度对识别结果的影响；

Step 6. 对比三个视角下的识别性能。

4.2. 实验结果与分析

Table 1. Comparison of single feature recognition and multi-feature joint recognition results at 90 degree perspective

Table 2. The influence of PCA dimension reduction on recognition results at 90 degree single view

Table 3. Influence of different sparse degree on multi-feature joint recognition

Table 4. Recognition results at different perspectives

5. 小结

Gait Recognition Algorithm Based on Sparse Representation of Joint Multi-Feature Dictionary[J]. 计算机科学与应用, 2017, 07(04): 398-406. http://dx.doi.org/10.12677/CSA.2017.74048

1. 1. Murray, M.P. (1967) Gait as a Total Pattern of Movement. American Journal of Physical Medicine, 46, 290-333.

2. 2. Chai, Y., Ren, J., Han, W., et al. (2011) Human Gait Recognition: Approaches, Datasets and Challenges. Proceedings of 2011 4th IEEE International Conference on Imaging for Crime Detection and Prevention, London, 3-4 November 2011, 1-6.

3. 3. Yoo, J. and Marker, N. (2011) Automated Markerless Analysis of Human Gait Motion for Recognition and Classification. ETRI Journal, 33, 259-266. https://doi.org/10.4218/etrij.11.1510.0068

4. 4. Boulgouris, N.V., Hatzinakos, D. and Plataniotis, K.N. (2005) Gait Recognition: A Challenging Signal Processing Technology for Biometric Identification. IEEE Signal Processing Magazine, 22, 78-90. https://doi.org/10.1109/MSP.2005.1550191

5. 5. Boulgouris, N.V. and Chi, Z.X. (2007) Gait Recognition Using Radon Transform and Linear Discriminant Analysis. IEEE Transactions on Image Processing, 16, 857-860. https://doi.org/10.1109/TIP.2007.891157

6. 6. Xu, D., Huang, Y., Zeng, Z. and Xu, X. (2012) Human Gait Recognition Using Patch Distribution Feature and Locality-Constrained Group Sparse Representation. IEEE Transactions on Image Processing, 21, 316-326.

7. 7. Nagendraswamy, H.S. (2014) Change Energy Image for Gait Recognition: An Approach Based on Symbolic Representation. International Journal of Image, Graphics and Signal Processing, 6, 1. http://www.mecs-press.org/

8. 8. Wang, L., Tan, T., Hu, W., et al. (2003) Automatic Gait Recognition Based on Statistical Shape Analysis. IEEE Transactions on Image Processing, 12, 1120-1131. https://doi.org/10.1109/TIP.2003.815251

9. 9. Han, J. and Bhanu, B. (2006) Individual Recognition Using Gait Energy Image. IEEE Transactions on Pattern Analysis and Machine Intelligence, 28, 316-322. https://doi.org/10.1109/TPAMI.2006.38

10. 10. Liu, N., Lu, J. and Tan, Y. (2011) Joint Subspace Learning for View-Invariant Gait Recognition. IEEE Signal Processing Letters, 18, 431-434. https://doi.org/10.1109/LSP.2011.2157143

11. 11. Online CASIA Database Information. http://www.cbsr.ia.ac.cn/english/Gait%20Databases.asp