Multiframe blind deconvolution using wavelength diversity

The resolution of images acquired with ground-based telescopes is generally limited by the blurring effects of phase distortions caused by atmospheric turbulence. A number of methods have been developed to combat this problem. One method, based on post-processing of blurred imagery, is blind deconvolution. A blind deconvolution method in current use at several astronomical observing facilities is phase diversity. In phase diversity, two images of the same object are simultaneously acquired, with one image in-focus and the other image defocused by a known amount. These image pairs are jointly processed to estimate the original undistorted object. Phase diversity has been extensively researched.; An alternative to phase diversity, called wavelength diversity, has received far less attention to date. In wavelength diversity, images of the same object are simultaneously acquired at two separate wavelengths. These images are jointly processed to estimate the intensity of the original undistorted object at each wavelength. Previous work on wavelength diversity required an assumption that the object intensity was common, up to a scale factor, at the two diverse wavelengths. Previously presented results on wavelength diversity were also based on simulation only, and did not include any results with real data.; This thesis presents a new algorithm for the blind deconvolution of wavelength-diverse image sequences that does not require the common object assumption of previous work. The algorithm is tested using both synthetic and real data. This thesis presents the first successful results of wavelength-diverse blind deconvolution using real images. Results with simulated images are excellent, demonstrating that the algorithm is capable of recovering high-frequency object detail that is obscured in the blurred images. Results with real data show that the algorithm is capable of providing demonstrable improvements in image fidelity.