Using speckle imaging techniques as a starting point for MFBD scene reconstruction from long horizontal-path, turbulence-degraded imagery

Recent works indicate that both MFBD and speckle imaging methods are e ective in recovering images of scenes from sets of turbulence-degraded imagery acquired over long horizontal paths. In this work, a prototype scene estimate, generated using speckle-imaging methods, is used in place of the multi-frame ensemble average, to initialize the iterative MFBD algorithm. Available performance improvements are described quantitatively by examining the improvement in Mean Squared Error (MSE) compared to a di raction-limited image. When speckle image estimates initialize the MFBD algorithm residual MSE is reduced by 16% on average compared the case where the multi-frame average is used as a starting point. Similarly, residual MSE is reduced another 8% beyond what is available using speckle imaging method alone when the number of iterations is not constrained. We also nd that the variation in reconstruction MSE is reduced signi cantly using only a limited number of iterations when subject to low to moderated image degradation compared to speckle imaging alone