Image Restoration

Abstract

Restoration of a high quality image from a degraded recording is an important problem in early vision processing. Image usually refers to a two-dimensional light intensity function x(a, b). Since light intensity is a real positive quantity and the maximum brightness of an image is restricted by the practical imaging system, x(a, b) is a finite, real and non-negative function, where 0≤x(a,b)≤A with A being the maximum image brightness. Figure 7.1 shows a digital image restoration system containing three subsystems: an imaging system, image digitizer and image restoration system. The imaging system, which consists of an optical system and recording devices, is a major source of degradations. To enable processing by a computer, images are sampled and quantized by the image digitizer. The image digitizer also introduces some degradations because of quantization error. The image restoration system uses some techniques to remove (1) deterministic degradations indexDegradations, deterministic such as blur due to optical system aberrations, diffraction, motion, atmospheric turbulence, film nonlinearities, and (2) statistical degradations such as noise due to electronic imaging sensors, film granularity and atmospheric light fluctuations. The digital image restoration system gives an estimate of the original image in some sense.