基于PCNN的图像二值化及分割评价研究

在介绍传统二值化方法的基础上，对PCNN脉冲耦合神经网络进行了研究，由于PCNN独特的相似神经元同步点火的特性，使其非常适合于各类图像的分割；同时结合图象二值化研究，对目前仍处于难点的图像分割评价作了研究，提出了综合评价思想。并通过实验仿真验证了：PCNN用于图像二值化，能够更好地区分和合理地分割出图像的目标和背景区域，具有分割精度高，适应类型广的特点；同时通过大量的实验数据也验证了所提出的综合评价思想的合理性，较之以往的评价准则更加准确和科学。

Image Binarization Based on PCNN and Segmentation Evaluation Methods

This paper have mainly studied the image binary segmentation based on PCNN， and given a brief introduction of segmentation evaluation criteria. We combine these evaluation criteria with segmentation algorithms and give a quantitative evaluation to the PCNN segmentation algorithm through doing large amounts of experiments on different types of images. The experimental results show that PCNN based image segmentation has a property of high accuracy, strong adaptability features. And its performance obviously superior to the traditional segmentation algorithms; Meanwhile we get that the proposed composite segmentation evaluation method is feasible and effective, and it can reflect the performance of segmentation algorithms.