Adaptive Image Estimation Using Reduced Update Filters

摘要

When an image is estimated from noisy data using a linear shift-invariant (LSI) filter, the subjective improvement is relatively poor at low signal-to-noise ratios. This occurs for at least two reasons: first, the statistics of the image are markedly space-variant and second, the eye is very sensitive to blurring of edges. However, adaptive filtering techniques can be applied to improve the subjective quality of noisy images even at low signal-to-noise ratios. This is accomplished in the present work by using multiple models to match the space-invariant statistics and by using oriented edge models to prevent edge blurring in the filtered result. (Author)