论文部分内容阅读
提出了一种基于Hopfield神经网络(HopfieldNeuralNetwork,简称HNN)优化的图像重建算法.将图像重建问题转化为HNN优化问题,取重建图像熵函数最大以及原始投影与再投影之间的误差平方和最小作为图像重建的优化目标,作为能量函数构造连续型HNN模型,由HNN能量函数极小化可得到重建问题的优化解.这种方法具有简单易行、计算量小、收敛快、便于并行计算等特点.对照ART算法,用计算机模拟产生的无噪声投影数据检验新算法,验证了新算法的优越性
An image reconstruction algorithm based on Hopfield Neural Network (HNN) optimization is proposed.It transforms the image reconstruction problem into the HNN optimization problem, and takes the maximum of the entropy function of reconstructed image and the minimum sum of squares of the error between the original projection and the re-projection The optimal objective of image reconstruction is to construct continuous HNN model as an energy function and to obtain the optimal solution of the reconstruction problem by minimizing the energy function of HNN. This method has the characteristics of simple and easy operation, fast calculation, fast convergence and convenient parallel computation Compared with the ART algorithm, the new algorithm is verified by the noise-free projection data generated by the computer simulation to verify the superiority of the new algorithm