介绍了总变分的基本理论和基于总差分约束的Richardson—Lucy算法.该算法在处理射电图像时进行了近似计算,导致迭代过程中累计误差逐步增大,图像恢复结果不理想.为此进行了改进,在计算总差分导数时采用了非线性投影取代原始算法的线性近似,提出了基于非线性投影的总变分正则项的Richardson-Luey算法,并成功地应用到射电日像仪图像重建中.通过仿真实验表明,改进的算法在日像仪图像重建中更加有效地恢复出了测得的图像.
This article simply introduces the basic theory of total variation and the Richardson-Lucy algorithm based on Total Variation. When this algorithm is applied to radio image reconstruction, accumulated discrepancy rises gradually during the iterations as the result of the approximation in derivation; hence the recovered images are often unsatisfactory. Aiming at this problem, this article comes up with a new Richardson-Lucy algorithm based on nonlinear projection and it is successfully applied to the heliograph image reconstruction. The simulation experiment shows that it turns out to be more effective to restore the image measured by using the improved Richardson-Lucv algorithm.