Shift-and-add is an approach employed to mitigate the phenomenon of resolution degradation in images acquired through a turbulent medium. Using this technique, a large number of consecutive short exposures is registered below the coherence time of the atmosphere or other blurring medium. The acquired images are shifted to the position of the brightest speckle and stacked together to obtain high resolution and high signal-to-noise frame. In this Letter, we present a highly efficient method for determination of frame shifts, even if in a single frame the object cannot be distinguished from the background noise. The technique utilizes our custom genetic algorithm, which iteratively evolves a set of image shifts. We used the maximal energy of stacked images as an objective function for shifts estimation and validate the efficiency of the method on simulated and real images of simple and complex sources. Obtained results confirmed that our proposed method allows for the recovery of spatial distribution of objects even only 2% brighter than their background. The presented approach extends significantly current limits of image reconstruction with the use of shift-and-add method. The applications of our algorithm include both the optical and the infrared imaging. Our method may be also employed as a digital image stabilizer in extremely low light level conditions in professional and consumer applications.