四元共空间特征提取算法及其在纸币识别中的应用

doi:10.11959/j.issn.1000-436x.2018285

通信学报 ›› 2018, Vol. 39 ›› Issue (12): 40-46.doi: 10.11959/j.issn.1000-436x.2018285

四元共空间特征提取算法及其在纸币识别中的应用

盖杉

南昌航空大学信息工程学院，江西南昌 330063

修回日期:2018-09-27 出版日期:2018-12-01 发布日期:2019-01-21
作者简介:盖杉（1980–），男，吉林长春人，博士，南昌航空大学副教授，主要研究方向为图像处理、稀疏表示与模式识别。
基金资助:
国家自然科学基金资助项目(No.61563037);江西省杰出青年基金资助项目(No.20171BCB23057);江西省自然科学基金资助项目(No.20171BAB202018)

Feature extraction algorithm based on quaternion common spatial pattern for banknote recognition

Shan GAI

School of Information Engineering, Nanchang Hangkong University, Nanchang 330063, China

Revised:2018-09-27 Online:2018-12-01 Published:2019-01-21
Supported by:
The National Natural Science Foundation of China(No.61563037);The Outstanding Youth Scheme of Jiangxi Province(No.20171BCB23057);The Natural Science Foundation of Jiangxi Province(No.20171BAB202018)

摘要/Abstract

摘要：

针对目前纸币图像特征提取与分析方法缺少对相位结构信息有效描述的问题，提出一种基于四元共空间模式的特征提取算法。该算法首先采用四元矩阵描述纸币图像的相位与幅值信息，并对四元复合厄米特协方差矩阵进行对角化；然后将样本向量输入到复合四元滤波器中，并将分解得到的四元矩阵实部与虚部的方差作为纸币特征向量；最后采用神经网络作为分类器并在识别阶段引入拒识类。实验结果表明所提的算法不仅取得了较高的识别率，而且还能够满足纸币清分系统的实时性要求。

关键词: 四元矩阵, 共空间模式, 特征提取, 神经网络, 纸币识别

Abstract:

New feature extraction algorithm was proposed based on quaternion common spatial pattern in order to solve the lack of effective description of phase information in the banknote feature extraction and analysis. Firstly, the quaternion matrix was utilized to describe the phase information of the banknote image, and made diagonalization of quaternion composite Hermitian matrix. Secondly, the sample vector was input to the composite quaternion filter. The extracted feature vector was obtained by using the variance of the real part and imaginary part. Finally, the neural network was applied as classifier and the reject class was introduced in the banknote recognition. The experimental results illustrate that the proposed algorithm obtains high recognition rate and meets the real-time requirement of the banknote recognition system. The proposed algorithm has already been applied in a resource-constrained embedded system at the same time.

Key words: quaternion matrix, common spatial pattern, feature extraction, neural network, banknote recognition

中图分类号:

TN92

盖杉. 四元共空间特征提取算法及其在纸币识别中的应用[J]. 通信学报, 2018, 39(12): 40-46.

Shan GAI. Feature extraction algorithm based on quaternion common spatial pattern for banknote recognition[J]. Journal on Communications, 2018, 39(12): 40-46.

图/表 7

图1

图2

图3

表1

表2

表3

表4

参考文献 12

[1]	KATO N , OMACHI S . A handwriting character recognition system using directional element feature[J]. IEEE Trans on Pattern Analysis and Machine Intelligence, 1999,21(3): 258-262.
[2]	TAKEDA F , OMATU S . Bank note recognition system using neural network with random masks[C]// World Congress an Neural Networks. 1993.241-244.
[3]	VIOLA P , JONES M . Rapid object detection using a boosted cascade of simple features[C]// IEEE International Conference on Computer Vision and Pattern Recognition. 2001.511-518.
[4]	VIOLA P , JONES M , SNOW D . Detecting pedestrians using patterns of motion and appearance[C]// International Conference on Computer Vision and pattern recognition. 2003.1324-1331.
[5]	刘家锋，刘松波，唐降龙 . 一种实时纸币识别方法的研究[J]. 计算机研究与发展， 2003,40(7): 1057-1061.
	LIU J F , LIU S B , TANG X L . An algorithm of real-time paper currency recognition[J]. Journal of Computer Research and Development, 2003,40(7): 1057-1061.
[6]	CHOI E , LEE J , YOON J . Feature extraction for bank note classification using wavelet transform[C]// International Conference on Pattern Recognition. 2006,934-937.
[7]	盖杉，刘鹏，刘家锋，等. 新的纸币图像特征提取方法[J]. 通信学报， 2010,31(4): 128-133.
	GAI S , LIU P , LIU J F , et al. New method of banknote feature extraction[J]. Journal on Communications, 2010,31(4): 128-133.
[8]	GINZBERG P , WALDEN A . Matrix-valued and quaternion wavelets[J]. IEEE Transactions on Signal Processing, 2013,61(6): 1357-1367.
[9]	XU D , MANDIC D P . The theory of quaternion matrix derivatives[J]. IEEE Transactions on Signal Processing, 2015,63(6): 1543-1556.
[10]	ENSHAEIFAR S , CHEONG C , PARK C , et al. Quaternion common spatial patterns[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2017,25(8): 1278-1286.
[11]	TOOK C C , MANDIC D P , ZHANG F . On the unitary diagonalisation of a special class of quaternion matrices[J]. Applied Mathematics Letters, 2011,24(11): 1806-1809.
[12]	PARK C , TOOK C , MANDIC D P . Augmented complex common spatial patterns for classification of noncircular EEG from motor imagery tasks[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2014,22(1): 1-10.

人民币(面值/元)	DWT算法+BP神经网络	MASK算法+BP神经网络	GRID算法+BP神经网络	CFFE算法+BP神经网络	QCSP算法+BP神经网络
100	78.56%/2.97%	86.24%/2.08	94.57%/1.64%	97.24%/1.26%	98.75%/0.95%
50	82.49%/1.65%	88.17%/1.27%	92.56%/0.92%	98.04%/0.83%	99.15%/0.74%
20	80.57%/2.47%	89.24%/2.28%	93.17%/1.75%	97.06%/1.19%	98.85%/0.97%
10	85.24%/3.06%	89.07%/2.72%	93.26%/2.18%	98.18%/1.65%	99.24%/1.08%
5	88.56%/2.04%	92.17%/1.62%	95.64%/1.18%	98.84%/0.86%	99.49%/0.72%

美元(USD)	DWT算法+BP神经网络	MASK算法+BP神经网络	GRID算法+BP神经网络	CFFE算法+BP神经网络	QCSP算法+BP神经网络
100	68.29%/2.24%	74.15%/2.07%	80.26%/1.82%	86.54%/1.37%	93.27%/0.86%
50	72.13%/2.95%	78.54%/2.53%	89.25%/1.62%	95.64%/1.08%	98.32%/0.75%
20	80.06%/3.05%	86.49%/2.84%	92.16%/2.16%	97.45%/1.75%	99.05%/1.08%
10	83.14%/1.96%	87.63%/1.37%	94.28%/1.05%	96.57%/0.84%	98.45%/0.57%
2	76.29%/2.05%	79.57%/1.84%	82.15%/1.49%	88.47%/0.97%	94.52%/0.78%
1	80.06%/1.82%	86.27%/1.37%	90.29%/0.92%	95.78%/0.83%	98.42%/0.47%

欧元(EUD)	DWT算法+BP神经网络	MASK算法+BP神经网络	GRID算法+BP神经网络	CFFE算法+BP神经网络	QCSP算法+BP神经网络
500	71.28%/2.45%	75.19%/2.07%	78.24%/1.68%	84.59%/1.05%	89.27%/0.87%
200	80.45%/1.92%	84.27%/1.47%	88.19%/0.95%	92.57%/0.74%	94.85%/0.53%
100	86.94%/2.17%	88.27%/2.05%	94.19%/1.72%	97.28%/1.38%	99.05%/0.96%
50	78.25%/1.82%	84.17%/1.73%	89.26%/1.29%	92.87%/1.05%	95.46%/0.84%
20	68.29%/2.95%	73.84%/2.62%	78.75%/1.84%	86.27%/1.49%	92.19%/0.98%
10	81.57%/1.34%	84.89%/1.18%	88.27%/0.87%	93.59%/0.64%	97.42%/0.45%
5	79.06%/1.45%	86.27%/1.19%	90.15%/0.86%	95.72%/0.73%	98.85%/0.62%

纸币数据库	DWT算法+BP神经网络	MASK算法+BP神经网络	GRID算法+BP神经网络	CFFE算法+BP神经网络	QCSP算法+BP神经网络
D1	96.28	35.74	27.31	25.19	18.64
D2	124.38	53.87	39.27	30.42	25.16
D3	78.84	53.69	28.42	22.37	19.82

[1]	陈晋音, 熊海洋, 马浩男, 郑雅羽. 基于对比学习的图神经网络后门攻击防御方法[J]. 通信学报, 2023, 44(4): 154-166.
[2]	李建锋, 刘哲宇, 荣洋, 李展, 廖柏林, 屈林曦, 刘志杰, 林琨煌. 用于线性噪声时变凸二次规划的归零神经网络[J]. 通信学报, 2023, 44(4): 226-233.
[3]	林云, 徐怀韬, 王森, 张思成, 庄龙. 基于特征融合的通信语音干扰效果客观评估[J]. 通信学报, 2023, 44(3): 105-116.
[4]	江沸菠, 彭于波, 董莉. 面向6G的深度图像语义通信模型[J]. 通信学报, 2023, 44(3): 198-208.
[5]	杨宏宇, 杨海云, 张良, 成翔. 基于特征依赖图的源代码漏洞检测方法[J]. 通信学报, 2023, 44(1): 103-117.
[6]	何世文, 袁军, 安振宇, 张敏, 黄永明, 张尧学. 基于图神经网络的联合用户调度与波束成形优化算法[J]. 通信学报, 2022, 43(7): 73-84.
[7]	冷涛, 蔡利君, 于爱民, 朱子元, 马建刚, 李超飞, 牛瑞丞, 孟丹. 基于系统溯源图的威胁发现与取证分析综述[J]. 通信学报, 2022, 43(7): 172-188.
[8]	李昂, 陈建新, 魏昕, 周亮. 面向6G的跨模态信号重建技术[J]. 通信学报, 2022, 43(6): 28-40.
[9]	兰巨龙, 朱棣, 李丹. 面向多模态网络业务切片的虚拟网络功能资源容量智能预测方法[J]. 通信学报, 2022, 43(6): 143-155.
[10]	王晓丹, 李京泰, 宋亚飞. DDAC：面向卷积神经网络图像隐写分析模型的特征提取方法[J]. 通信学报, 2022, 43(5): 68-81.
[11]	廖育荣, 王海宁, 林存宝, 李阳, 方宇强, 倪淑燕. 基于深度学习的光学遥感图像目标检测研究进展[J]. 通信学报, 2022, 43(5): 190-203.
[12]	张帆, 黄赟, 方子茁, 郭威. 卷积神经网络的损失最小训练后参数量化方法[J]. 通信学报, 2022, 43(4): 114-122.
[13]	朱政宇, 侯庚旺, 黄崇文, 孙钢灿, 郝万明, 梁静. 基于并行CNN的RIS辅助D2D保密通信系统资源分配算法[J]. 通信学报, 2022, 43(3): 172-179.
[14]	霍俊彦, 王丹妮, 马彦卓, 万帅, 杨付正. 基于轻量级全连接网络的H.266/VVC分量间预测[J]. 通信学报, 2022, 43(2): 143-155.
[15]	龙华, 黄张衡, 邵玉斌, 杜庆治, 苏树盟. 基于改进CFCC特征提取的语种识别算法研究[J]. 通信学报, 2022, 43(12): 211-221.

四元共空间特征提取算法及其在纸币识别中的应用

Feature extraction algorithm based on quaternion common spatial pattern for banknote recognition

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 12

相关文章 15

Metrics

推荐阅读 0