אירועים והרצאות בפקולטה למדעי המחשב ע"ש הנרי ומרילין טאוב

We address the open problem of blindly separating single-path position varying image mixtures, without having prior information about the sources. We assume that the mixing system's spatial distortion and attenuation change with position. A staged method for estimating the mixing models is used in turn to recover source signals from such mixtures. Our method is based on a Staged Sprase Component Analysis (SSCA) of the mixtures. Our method consists of three stages: aligning the signals to estimate the spatial distortion component of the mixing system; classifying the sparse signal samples to their estimated sources and estimating the spatial attenuation component of the mixing system; and finally inverting the mixing system to recover the sources. Small error in the spatial distortion component leads to a significant degradation in separation quality. However, in practice, uncertainty is associated with the model estimation stage, due to: noisy samples; bad spatial spread of samples; mixed samples in the sparse representation and more. We propose a solution by adding a step of model refinement that is based on simple image quality measures that would allow us to improve separation results. We test some standard methods and propose a new aproach based on Phase Congruency measure. M.Sc. thesis under the supervision of Prof. Zeevi