Multi-target detection with rotations

<p style='text-indent:20px;'>We consider the multi-target detection problem of estimating a two-dimensional target image from large noisy measurement that contains many randomly rotated and translated copies image. Motivated by single-particle cryo-electron microscopy, we focus on low signal-to-noise regime, where it is difficult to estimate locations orientations images in measurement. Our approach uses autocorrelation analysis rotationally translationally invariant features We demonstrate that, regardless level noise, our technique can be used recover when sufficiently large.</p>