Abstract:

BACKGROUND: Super-resolution reconstruction has been extensively studied and used in many fields, such as video and remote sensing.
OBJECTIVE: To reconstruct a high-resolution image from the low-resolution image sequence, an adaptive algorithm of super-resolution reconstruction is proposed.
METHODS: We use a constant regularize parameter (λ=2/3) and adaptive step size as scheme Ⅰ. The scheme Ⅱ takes into account inaccurate estimates of the registration parameter, the point spread function and the additive Gaussian noise in the low resolution image sequence. We structure a novel adaptive regularization functional, and analyze experimentally the convexity of the nonlinear cost function. Based on the convex of the cost function, we get the adaptive step size by the mathematical theory, which improves the spatial resolution of the image and the rate of convergence.
RESULTS AND CONCLUSION: Optical images are used to test the proposed method. The scheme Ⅱ performs better than scheme Ⅰ, in the sense of the enhanced peak signal to noise ratio. Compared to scheme Ⅱ, the computational cost of scheme Ⅰ is twice or more slower. The average computational cost of scheme Ⅱ is 68.25 seconds. The results show that the spatial resolution of the image and the rate of convergence are significantly improved. The experiment also proves that the stability of the algorithm is good with the progress of the iterative process.