Quality metrics of coherent ladar intensity images

The problem of how to accurately evaluate the quality of coherent ladar restoration images has never been solved. In this paper the intensity image of coherent ladar which included two simple targets was used and was denoised by six different kinds of filters, such as Lee filter and Non-local Means filter. Mean structural similarity (MSSIM) index, peak-signal-to-noise ratio (PSNR) and speckle index (SI) were used to evaluate the restoration intensity images of coherent ladar. The values of each metric were analyzed, and the accuracies were compared. The results show that PSNR isn´t in accordance with visual effects sometimes and SI has certain limitations to reflect the quality of images. MSSIM is more suitable to precisely evaluate the quality of restoration intensity images and basically in accordance with visual effects.