摘要
由于退化图像的点扩散函数难以准确确定,提出一种基于Fourier正交基函数的前向神经网络图像复原模型,该模型以一组Fourier正交基为隐层神经元的激励函数,根据误差传递算法进行权值修正,达到收敛目标。给出Fourier神经网络及其相应的衍生算法的图像恢复实现步骤。实验表明,该方法能较好地实现图像的复原。
According to the fact that PSF （ point spread function） of the degraded image can＇ t obtain accurately, this paper constructed a feed-forward neural network for image restoration based on the Fourier orthogonal function. The hidden-layer neurons were activated by a series of Fourier orthogonal functions, updated its weight by the error back-propagation training algorithm and finally reached convergence target. This paper applied the Fourier neural network and its hidden-neuron growing algorithm to recover the fuzzy image. Experiments show they have better performances on image restoration.
出处
《计算机应用研究》
CSCD
北大核心
2012年第3期1143-1145,共3页
Application Research of Computers
基金
山西省自然科学基金资助项目(2008011030)
山西省回国留学人员科研资助项目(2011-075)
太原市大学生创新创业专项项目(2010
2011)
太原科技大学校UIT项目(XJ2010040)
关键词
图像复原
傅里叶正交基函数
傅里叶神经网络
衍生算法
image restoration
Fourier orthogonal function
Fourier neural network
hidden-neuron growing algorithm
