摘要
以基于离散Hopfield神经网络的图像恢复模型为基础,研究基于神经网络解决降质图像恢复问题.从Toeplitz循环矩阵和模拟退火算法等二个方面对该算法进行改进,使时间复杂度和空间复杂度大大降低,同时提高了图像的恢复效果.实验表明,改进算法对于降质图像恢复是有效的,比传统的逆滤波、维纳滤波方法具有更好的峰值信噪比.
Based on the discrete Hopfield neural network model, a problem of degraded image restoration using a neural network was researched, and a new practical algorithm was presented by improving degraded image restoration algorithm, which consisted of circulant Toeplitz matrices and simulated annealing. The new algorithm reduced the space and time complexities of image restoration sighificantly and improved the quality of degraded image restoration. The new algorithm made degraded image restoration better on a conventional computer. The experiments shows that this algorithm is effective for image restoration and has a higher PSNR than traditional restoration algorithms of inverse filter and wiener filter.
出处
《湖南文理学院学报（自然科学版）》
CAS
2005年第3期59-63,共5页
Journal of Hunan University of Arts and Science(Science and Technology)
基金
湖南文理学院2004年院级科研资助项目(JJYB0410)
