ii
ABSTRACT
In order to improve the imaging quality, conventional methods such as ameliorating
the design of optical system, using special optical material, complicating the optical
structure and so on, were mainly focused on optics field. Although these methods improved
the imaging quality, they also increased the cost of designing and processing of imaging
system greatly. Therefore, many researchers advised to use algorithms of digital image
processing to restore the degraded images, improve the imaging quality in electrics field.
The research of this paper is just in this field.
Firstly, the paper analyzed aberration and put forward the mathematical model of
optical imaging system’s Space-Variant Point-Spread Function (SVPSF). According to the
model, the imaging system was simulated. In succession, according to the analysis of
noises and moving blurring, the degradation model produced by these two reasons was also
simulated in this paper.
To the degradation produced by aberration, two algorithms were brought forward
through analyzing theory, validating experimental data and referring to the emulational
model. In the first algorithm, all SVPSFs were analyzed through Fourier transform.
According to the results, this paper got every inverse filters and designed a group of filters.
Using these inverse filters, the degraded images were restored. The experimental data
improved this algorithm had good restoration results. In the other algorithm, Referring to
the principle of inverse convolution and using the algorithm of solving extremum, the
restoration filters which made the blur circle smallest was gotten. The restoration results
proved that this algorithm was also efficient.
To the degraded model produced by noises and moving aberration, on the basis of
summing up the common image restoration algorithms, a new algorithm associated with
spatial domain and frequency domain was brought up. In this algorithm, the impulse noise
was removed in the space domain firstly. In the followed, an improved Constrained Least
Squares restoration algorithm was thought out in the terms of the properties of Wiener
Filter and Constrained Least Squares Filter. The result of experiment proved it was an
efficient algorithm.
Key words: Aberration, Space-variant point-spread function
Image restoration, Constrained least squares