Space-Variant Blurring Image Restoration And Its Application

Space-variant blurring image restoration is an important aspect of image restoration field. Based on the existing space-variant restoration methods, a constraint Least-Squares restoration is realized to restore space-variant blurs. The study of this paper starts from the polynomial approximation to the inverse filtering of images. Regularizing the filter with energy constraint may get energy-constrained polynomial restoration, which makes the polynomial approximation to the inverse filtering of images well posed. But restoration error from Taylor expansion degrades restoration. So polynomial coefficients are modified to suppress the restoration error. Based on these studies, abstract function and operators are used to describe the blurring model and the restoration model. Then a general description of the image restoration problems is achieved. Decomposing the space-variant operators to the linear combination of the space-invariant operators, a restoration model with Least-Squares constraint, which includes direct model and iterative model, is proposed to restore space-variant blurs. The key point of the restoration model is the decomposition of the space-variant operators. The effectiveness of the decomposition is proved by some examples, and the choice conditions of sub-operator are presented. The space-variant restoration model is used to restore space-variant blurring images of opto-electronic imaging systems. Aimed at the opto-electronic imaging system with a specific point spread function, says Gaussian distribution, the polynomial fitting is used to decompose the operator. Based on the choice conditions of sub-operator, the primary function is obtained to decompose operator. Then the direct and iterative constrained Least-Squares restoration methods are realized to restore space-variant blurring images. The restoration equations for each method are derived. According to the direct equation, the restoration image is the linear combination of the blurring image's even-order differences. The combining coefficients are specified by the point spread function of imaging systems, constraint operators, constraint coefficients and decomposition primary function. According to the iterative equation, the update restoration image is the linear combination of the blurring image's even-order differences and constraint item. Combining coefficients of differences are specified by the point spread functions of imaging systems and decomposition primary function. The analysis experiments indicate that the restoration equation can improve the frequency response of the imaging system. After restoration, the point spread functions of the imaging system are driven to impulse functions. And the restoration equation with specific constraint can restore space-variant blurring image effectively. Simulations and restorations of space-variant blurring images are presented to demonstrate the efficiency of the proposed methods. And the images obtained from an stv680 digital camera can be restored with the methods. Restoration methods can improve the definition of the whole image. The results of the restoration demonstrate the efficiency of the methods.