为了解决现有单分子定位算法中定位速度慢和对噪声有评估依赖性的问题，基于补零快速傅里叶变换和相位梯度算子提出一种新型的噪声自由的频率域非迭代荧光单分子定位算法。计算机模拟结果表明该算法定位精度可达纳米量级，而定位速度与解线性方程组法在同一个数量级。进而在传统实验参数条件下，对不同间隔分子带模型进行了模拟超分辨成像。模拟结果表明，可以区分中心相隔30 nm的两个分子带。此外，将该算法用于HeLa细胞突起中微丝束结构的荧光超分辨成像，从重构获得的图像中可以看到微丝束的直径为75~200 nm，验证了该算法的实用性。

In order to improve the computational speed and prior background noise-dependent problem of single molecule localization algorithm, a Fourier domain localization scheme based on zero-padded fast Fourier transform and phase gradient operators is used to obtain a powerful mathematical model for localizing a single molecule without numerical fitting. Numerical simulations indicate that the proposed method exhibits nanometer scale localization precision while executing almost as fast as the fluoroBancroft algorithm. Furthermore, a sample consisted of several lines of molecules are simulated imaged. The results demonstrate that two lines of molecules separated by 30 nm can be resolved. Finally, super-resolution images of filopodia in HeLa cells are reconstructed based on the method, in which filopodia with diameter of 75~200 nm are resolved.