参数优化的PCNN图像检索
Parameter Optimization in PCNN for Image Retrieval

作者: 郭 成 , 曾 亮 :国防科学技术大学计算机学院,湖南 长沙;

关键词: 脉冲耦合神经网络粒子群算法参数优化图像检索Pulse Coupled Neural Network (PCNN) Particle Swarm Optimization (PSO) Parameter Optimization Content-Based Image Retrieval (CBIR)

摘要:
脉冲耦合神经网络(PCNN)用于图像检索时需人工确定较多参数,参数确定的好坏严重影响检索效果,针对以上问题,提出一种基于进化学习的参数优化方法。通过引入粒子群算法(PSO),构建优化目标函数,提前对图像库中少量图像进行分类训练,对脉冲耦合神经网络的各参数进行优化并用于图像检索。实验表明,提出的算法能有效找到各参数的近似最优解。对图像库中未训练图像进行检索时也取得较好效果,在检索查准率、查全率及主观视觉效果方面本文方法均优于经验参数。

Abstract: When PCNN was used for image retrieval, the manual parameters selection became the difficulties and whether the selected parameters are good or not determines the retrieval results mostly. A novel method based on evolutionary learning for optimizing the parameters of Pulse Coupled Neural Network (PCNN) was proposed to overcome these problems. Firstly we classified some images of the database and trained them in advance by introducing the Particle Swarm Optimization (PSO) and restructured the fitness function to optimize the parameters which were used in image retrieval. Experimental results show that the proposed method can achieve the optimal parameters adaptively, and the retrieval results perform well even in the untrained images. The retrieval results convince that the proposed method was better than experienced parameters on precision ratio, recall ratio and personal visual judgment.

文章引用: 郭 成 , 曾 亮 (2015) 参数优化的PCNN图像检索。 软件工程与应用, 4, 115-120. doi: 10.12677/SEA.2015.46015

参考文献

[1] 贾松敏, 徐涛, 董政胤, 等. 采用脉冲耦合神经网络的改进显著性区域提取方法[J]. 光学精密工程, 2015, 23(3): 819-826.

[2] Wang, Z.B., Ma, Y.D., Cheng, F.Y. and Yang, L.Z. (2010) Review of Pulse-Coupled Neural Networks. Image and Vision Computing, 28, 5-13.
http://dx.doi.org/10.1016/j.imavis.2009.06.007

[3] 王晓飞, 李柏年. 利用脉冲耦合神经网络的纹理图像检索方法[J]. 计算机工程与应用, 2012, 48(7): 201-204.

[4] 朱红伟, 周冬明, 聂仁灿, 等. 利用PCNN实现商标图像检索新方法[J]. 云南大学学报(自然科学版), 2012, 34(3): 276-284.

[5] 张静, 曹林伟. 基于脉冲耦合神经网络与欧拉数的图像检索[J]. 计算机应用与软件, 2014, 31(6): 232-235.

[6] Deng, X.Y. and Ma, Y.D. (2012) PCNN Model Automatic Parameters Determination and Its Modified Model. Acta Electronica Sinica, 40, 955-964.

[7] Zhou, D.G., Zhou, H., Gao, C. and Guo, Y.C. (2015) Simplified parameters model of PCNN and its application to image segmentation. Pattern Analysis & Applications.
http://dx.doi.org/10.1007/s10044-015-0462-6

[8] 曲仕茹, 杨红红. 基于遗传算法参数优化的PCNN红外图像分割[J]. 强激光与粒子束, 2015, 27(5): 32-37.

[9] Yonekawa, M. and Kurokawa, H. (2010) The Parameter Optimization of the Pulse Coupled Neural Network for the Pattern Recognition. Lecture Notes in Computer Science, 6354, 110-113.
http://dx.doi.org/10.1007/978-3-642-15825-4_13

[10] Kennedy, J. and Eberhart, R. (1995) Particle Swarm Optimization. Proceedings of IEEE International Conference on Neural Networks, 4, 1942-1948.
http://dx.doi.org/10.1109/icnn.1995.488968

[11] 马国庆, 李瑞峰, 刘丽. 学习因子和时间因子随权重调整的粒子群算法[J]. 计算机应用研究, 2014, 31(11): 3291- 3294.

[12] 任伟建, 武璇. 一种动态改变学习因子的简化粒子群算法[J]. 自动化技术与应用, 2012, 31(10): 9-11.

[13] 马义德, 李廉, 绽琨, 等. 脉冲耦合神经网络与数字图像处理[M]. 北京: 科学出版社, 2008.

[14] Ma, Y., Liu, L., Zhan, K. and Wu, Y.Q. (2010) Pulse-Coupled Neural Networks and One-Class Support Vector Machines for Geometry Invariant Texture Retrieval. Image & Vision Computing, 28, 1524-1529.
http://dx.doi.org/10.1016/j.imavis.2010.03.006

[15] Johnson, J.L. and Padgett, M.L. (1999) PCNN Models and Applications. IEEE Transactions on Neural Networks, 10, 480-498.
http://dx.doi.org/10.1109/72.761706

[16] 朱永松, 国澄明. 基于相关系数的相关跟踪算法研究[J]. 中国图象图形学报, 2004, 9(8): 963-967.

分享
Top