基于Harris矩阵的脉冲耦合神经网络模型
A Pulse Coupled Neural Network Model Based on Harris Matrix

DOI: 10.12677/CSA.2021.118211, PP. 2064-2069

Keywords: 脉冲耦合神经网络模型，动态阈值，衰减时间常数，Harris矩阵
Pulse Coupled Neural Network Model, Dynamic Threshold, Decay Time Constant, Harris Matrix

Full-Text   Cite this paper   Add to My Lib

Abstract:

分割含有弱边界、对比度低的图像目标时，传统的脉冲耦合神经网络模型难以得到有效的分割效果，主要原因是动态阈值的衰减时间常数固定，衰减速度固定，分割弱边界时，输出的脉冲信号难以准确描述目标区域，产生误分割现象。为了解决这一问题，本文采用Harris矩阵获取图像梯度信息，特别是弱边界的梯度信息，提出动态阈值的衰减速度与图像的梯度信息有关，当梯度信息大时，动态阈值的衰减快，当梯度信息小时，动态阈值的衰减速度慢，给出动态阈值的新定义。通过对弱边界、对比度低的图像进行仿真对比实验，说明本文算法的分割效果优于传统脉冲耦合神经网络模型。
The traditional pulse coupled neural network model can not get an effective segmentation effect when the image object with weak boundary and low contrast is segmented. The main reason is that the decay time constant of the dynamic threshold is fixed and the decay speed is fixed, when the weak boundary is segmented, it is difficult for the output pulse signal to accurately describe the target region, resulting in false segmentation. In order to solve this problem, the Harris matrix is used to obtain the gradient information of the image, especially the gradient information of the weak boundary. It is proposed that the decay velocity of the dynamic threshold is related to the gradient information of the image. When the gradient information is large, the decay velocity of the dynamic threshold is fast, when the gradient information is small, the decay velocity of the dynamic threshold is slow, and a new definition of the dynamic threshold is given. Simulation experiments on weak edge and low contrast images show that the proposed algorithm is superior to the traditional pulse coupled neural network model.

References

[1]	Eckhom, R., Reiboeck, H.J., Arndt, M., et al. (1990) Feature Linking via Synchronization among Distributed Assemblies: Simulation of Results from Cat Cortex. Neural Computation, 2, 293-307. https://doi.org/10.1162/neco.1990.2.3.293
[2]	Johnson, J.L. and Padgett, M.L. (1999) PCNN Models and Applications. IEEE Transactions on Neural Networks, 10, 480-498. https://doi.org/10.1109/72.761706
[3]	周东国, 高潮, 郭永彩. 一种参数自适应的简化PCNN图像分割方法[J]. 自动化学报, 2014, 40(6): 1191-1197.
[4]	Wu, C.D., Liu, Z.G. and Jiang, H. (2018) Catenary Image Segmentation Using the Simplified PCNN with Adaptive Parameters. Optik, 157, 914-923. https://doi.org/10.1016/j.ijleo.2017.11.171
[5]	Zhou, D.G. and Chi, M. (2019) Pulse Coupled Neural Network and Its Optimization for Segmentation of Electrical Faults with Infrared Thermography. Applied Soft Computing Journal, 77, 252-260. https://doi.org/10.1016/j.asoc.2018.10.056
[6]	Chen, Y.L., Park, S., Ma, Y.D., et al. (2011) A New Automatic Parameter Setting Method of a Simplified PCNN for Image Segmentation. IEEE Transactions on Neural Networks, 22, 880-891. https://doi.org/10.1109/TNN.2011.2128880
[7]	Wei, S., Hong, Q. and Hou, M.S. (2011) Automatic Image Seg-mentation Based on PCNN with Adaptive Threshold Time Constant. Neurocomputing, 74, 1485-1491. https://doi.org/10.1016/j.neucom.2011.01.005
[8]	Gao, C., Zhou, D.G. and Guo, Y.C. (2013) Automatic Iterative Algorithm for Image Segmentation Using a Modified Pulse Coupled Neural Network. Neurocomputing, 119, 332-338. https://doi.org/10.1016/j.neucom.2013.03.025
[9]	He, F.L., Guo, Y.C. and Gao, C. (2019) A Parameter Estima-tion Method of the Simple PCNN Model for Infrared Human Segmentation. Optics and Laser Technology, 110, 114-119. https://doi.org/10.1016/j.optlastec.2018.05.042
[10]	Li, J.J. (2011) Harris Corner Detection Algorithm Based on Improved Contourlet Transform. Procedia Engineering, 15, 2239-2243. https://doi.org/10.1016/j.proeng.2011.08.419

Full-Text