基于模糊调整的径向基函数网络图像恢复算法

为了解决传统图像恢复中存在的建模难的问题,提出了一种基于RBF神经网络的图像恢复算法,该算法利用RBF神经网络的非线性映射能力和适应性,通过记录退化过程的逆过程来恢复图像。首先改进RBF网络中心参数的确定过程,提出基于模糊调整的中心参数学习算法,然后用模糊调整后的网络进行图像恢复。仿真结果表明,改进的RBF网络可对典型退化图像进行令人满意的恢复。

Image restoration with radial basis function network based on fuzzy adjustment

To overcome the difficulties of creating mathematics model in traditional image restoration, a image restoration algorithm based on a RBF neural network was given. The algorithm could restore similarly degenerated image by learning the reverse process of a degenerated process by using no-linear map and adaptability of RBF network. The method of determining RBF network central parameters was improved by fuzzy adjustment at first and then the image restoration algorithm was designed. The simulation shows that the improved RBF network can restore the typically degenerated image satisfactorily.