一种新的PCNN模型参数估算方法

A New Method of PCNN's Parameter's Optimization

PCNN在图像处理领域得到广泛的应用,对其展开研究具有重要的理论意义及应用价值.在对PCNN的研究应用中,其模型参数的合理确定是一个难点.本文提出用灰度-信息量直方图来表征图像特征,通过对信息量直方图的分析,提出了估算PCNN时间衰减参数的自适应算法.该算法可以仅在PCNN的一个运行周期中以最少的迭代次数有效地完成图像分割,并且解决了对多目标进行分割时容易丢失目标的问题.

Recent research in neuroscience has proposed a new artificial neural networks model, the pulse coupled neural networks, PCNN, Several PCNN structures for image proeessing have been proposed depending on the model＇ s potential, In the research of the theories and the applications of PCNN,it isn＇t a trivial task to define the relative parameters properly,and people usually get the values by experience with many experiments.As a contribution to this research field,this paper presents a new method for image processing based on the image histogram of gray-level and amount of information. The histogram is used as a new tool to describe the image features and furthermore to define the decay time constant of PCNN. With this new algorithm, we can get perfect segmentation result with the fewest iteration times only in one computation period of P CNN, and also can resolve the problem in image segmentation that it is liable to loss some object while many objects exist. Experiments show that the new algorithm has good performance in image processing.