一种新的基于PCNN的图像自动分割算法研究

A Study of a New Image Segmentation Algorithm Based on PCNN

脉冲耦合神经网络(PCNN)非常适合图像分割.在参数确定的情况下,分割效果随迭代次数呈周期性变化.因此确定最佳迭代次数是运用PCNN进行图像自动分割的关键.本文提出一种基于连通域计算的边缘统计算法,用于评价迭代结果的有效边缘.最大有效边缘值所对应的迭代输出即为最佳分割.实验证明,该算法比基于图像熵和基于边缘算子的算法灵敏度高,抗噪声能力强.

Pulse-Coupled Neural Networks （PCNN） is very suitable for image segmentation. In condition of certain parameters, the result of segmentation will periodically change with the iteration times. Therefore, how to decided the best iteration times is the key of applying PCNN image auto-segmentation. In this paper, an edge statistic algorithm based on calculation of connected region is provided. This algorithm calculates the valid edge of the segmentation result, and it means that the max is accordant with the best segmentation. It has been proved by experiments that the algorithm has much better sensitivity than those methods based on entropy of image or on edge operator, and also has stronger robustness of noise.