Robust Algorithm for Large-Scale Gaussian Patterns Localization

Efficient accurate Gaussian localization is an important topic in many applications, e.g. based super-resolution microscopy and image scanning microscopy, which requires large-scale patterns for reconstruction. Existing methods usually require high signal-to-noise the existing standard fitting algorithm manually inputting a good initial value all parameters, could be not convenient to use difficult guarantee robustness localizations with computer. It would even more challenge detect high-dynamic-range of amplitudes, as well estimate parameters efficient low ratio data strong background. In this paper, we propose detection technique robust method without estimation. our technique, fast Pearson correlation proposed improve efficiency calculation normalized cross large scale object template matching. By introducing blind background estimation, modified iterative least-squares initials estimation noisy The simulation shows that performance SNR improvement 27% can achieved; capable calculating resulting accuracy very close exiting methods, indicates higher than 10dB required obtain subpixel accuracy.