基于Chebyshev神经网络的图像复原算法

Image Restoration Algorithm Based on Chebyshev Neural Network

退化图像的点扩散函数难以准确确定,为此,提出一种基于Chebyshev正交基函数的前向神经网络图像复原算法。该算法以一组Chebyshev正交基为隐层神经元的激励函数,采用BP算法对权值进行修正,达到收敛目标。给出2类Chebyshev神经网络的实现步骤及其相应衍生算法的图像恢复实现步骤。实验结果表明,该算法能较好地实现图像复原。

According to the fact that Point Spread Function（PSF） of the degraded image cannot be obtained accurately, a feed-forward neural network for image restoration is constructed based on the Chebyshev orthogonal function in this paper. The hidden-layer neurons are activated by a series of Chebyshev orthogonal functions. It updates its weights by the error Back Propagation（BP） gaining algorithm and finally reaches convergence target. This paper applies the two types of Chebyshev neural networks and their hidden-neuron growing algorithms to recover the fuzzy image. Experimental results show they have better performance on image restoration.