基于并行点火PCNN模型的图像分割新方法

A Novel Method of Image Segmentation Based on Parallelized Firing PCNN

提出一种并行点火脉冲耦合神经网络(Parallelized firing pulse coupled neural networks,PFPCNN)模型的图像分割方法.首先用改进的Unit—linking PCNN(ULPCNN)模型对图像进行增强,便于后续的图像分割.然后采用PFPCNN新模型对增强后的图像进行分割,最后用最大香农熵方法判定最佳分割结果.各种复杂场景下的仿真实验及定量评价表明,本文提出的图像分割方法,其效果明显优于常规的PCNN分割方法。

A novel method for image segmentation based on parallelized firing pulse coupled neural networks （PFPCNN） is presented in this paper. At first, the improved unit-linking PCNN （ULPCNN） is used to enhance the image. Then, PFPCNN model is adopted to segment the enhanced image by the improved ULPCNN. Finally, the maximal Shannon entropy is used to determine the optimal result from the segmented images. Experimental results show that the proposed method is more effective than the traditional PCNN and other improved PCNN models by quantitatively evaluating their performance.