基于脉冲耦合神经网络的红外图像分割算法研究  

Research on Infrared Image Segmentation AlgorithmBased on Pulse-Coupled Neural Network

作　　者：赵亮 杨凯 姚兴 王标 袁鹏喆 ZHAO Liang;YANG Kai;YAO Xing;WANG Biao;YUAN Pengzhe(Wang Jia Ling Coal Mine,China Coal Hua Jin Group Co.,Ltd.,Hejin 043300,China;Taiyuan Research Institute of China Coal Technology Engineering Group,Taiyuan 030006,China;China Coal(Tianjin)Intelligent Research Institute Co.,Ltd.,Tianjin 300000,China)

机构地区：[1]中煤华晋集团有限公司王家岭矿,山西河津043300 [2]中国煤炭科工集团太原研究院有限公司,山西太原030006 [3]中煤(天津)地下工程智能研究院有限公司,天津300000

出　　处：《自动化仪表》2024年第7期50-54,共5页Process Automation Instrumentation

摘　　要：红外图像噪声多、结构复杂,传统的分割方法难以对红外图像实现完整的分割。脉冲耦合神经网络(PCNN)参数众多,而人工调整网络参数耗时耗力。为了避免人工调整网络参数、实现对红外图像较完整的分割,将粒子群优化(PSO)算法和改进的PCNN相结合,提出了一种新的红外图像分割算法。首先,对红外图像进行预处理。其次,为了减少网络参数,对PCNN进行了化简。然后,将预处理图像输入网络进行循环迭代,利用PSO算法计算每次迭代的分割结果的适应函数值,从而确定群体和个体最佳参数。最后,通过种群最佳参数得到分割结果。为了验证算法的分割性能,试验使用了不同的自然风景图像和红外图像。试验结果表明,所提算法可以对红外和自然风景图像实现较完整的分割;无论是分割效果还是定性分析,所提算法都要优于传统算法。Infrared images are noisy and complex in structure,and it is difficult for traditional segmentation methods to achieve complete segmentation of infrared images.Pulse-coupled neural network(PCNN) has numerous parameters,and manual adjustment of network parameters is time-consuming and labor-intensive.In order to avoid the manual adjustment of network parameters while achieving a more complete segmentation of infrared images,a new infrared image segmentation algorithm is proposed by combining the particle swarm optimization (PSO) algorithm and the improved PCNN.Firstly,the infrared images are preprocessed.Secondly,the PCNN is simplified in order to reduce the network parameters.Then,the preprocessed images are input to the network for cyclic iterations,and the PSO algorithm is used to calculate the adaptation function values of the segmentation results for each iteration,so as to determine the population and the individual optimal parameters.Finally,the segmentation results are obtained by the population optimal parameters.In order to verify the segmentation performance of the algorithm,different natural landscape images and infrared images are used in the experiments.The experimental results show that the proposed algorithm can achieve a more complete segmentation of infrared and natural landscape images;the proposed algorithm is better than the traditional algorithms in both segmentation effect and qualitative analysis.

关 键 词：红外图像分割 脉冲耦合神经网络 粒子群优化算法 循环迭代 适应度函数 种群最佳参数 个体最佳参数 

分 类 号：TH-39[机械工程]