一种超模糊熵ULPCNN图像自动分割新方法

为了自动地对图像进行二值分割，提出了一种新的自适应迭代全局阈值图像分割算法．首先对二维超模糊集隶属函数进行了自适应修正，并将其引入到图像超模糊熵概念中; 然后从适应图像分割角度，将传统脉冲耦合神经网络模型改进为具有单调指数上升阈值函数的ULPCNN抑制捕获模型; 最后把ULPCNN与最大超模糊熵判据相结合对图像进行自动分割，并与基于最大香农熵、最小交叉熵及最小模糊熵准则的ULPCNN分割方法作了比较．理论分析和实验结果表明，该方法能自动确定迭代次数和选取最佳阈值，对图像目标划分清晰，细节保持较好，改善了图像的分割性能．

Automated image segmentation using the ULPCNN model with ultra-fuzzy entropy

In order to process the binary segmentation of an image automatically, a new adaptive iterative image segmentation algorithm with the property of global threshold is proposed. Firstly, the two-dimensional ultra-fuzzy sets membership function, which is adaptively modified, is introduced into the concept of image ultra-fuzzy entropy. Secondly, the traditional pulse coupled neural networks (PCNN) model is improved to obtain the restraining capture Unit-Linking PCNN model with the monotony exponential raised threshold function from the point of view of image segmentation. Finally, the improved ULPCNN is combined with the criterion of maximum ultra-fuzzy entropy to process image segmentation automatically. A comparison is made between this algorithm and ULPCNN segmentation methods based on the criteria of maximum Shannon entropy, minimum cross entropy and minimum fuzzy entropy. Theoretical analysis and experimental simulations show that the proposed algorithm automatically determines the number of iterative times, chooses the best threshold, separates the objects in the image clearly, preserves most of the details, and enhances the performance of image segmentation.