基于多项式外推的Richardson-Lucy算法加速研究

Acceleration of Richardson-Lucy algorithm based on polynomial extrapolation

提出一种新的可加速Richardson-Lucy(R-L)图像迭代恢复算法的方法。该方法基于阻尼R-L算法,通过存储阻尼R-L算法的前若干次(n次)迭代运算结果,利用多项式外推法分析这前n次迭代运算结果,并用一个多项式函数近似描述各结果之间的关系。通过该多项式的外推,预测以后的迭代结果,从而减少迭代运算的次数,取得了较好的加速效果。该方法可以实现几乎没有图像失真的复原,并能应用于其他类型的算法。

A new method for the acceleration of Richardson-Lucy （R-L）iterative image restoration algorithms is given, which is based on the damped R-L algorithm and polynomial extrapolation. By storing the latest several （n） results of damped R-L algorithm, a polynomial function is used to analyze these results, and to approximate the relationship among them. Then, this polynomial function is utilized to predict the later results in the iterative process. This method can simplify calculation and accelerate the image restoration algorithms. The method can offer nearly perfect reconstructed image, and is applicable to other relevant algorithms.