A Denoising Filter Design based on No-Reference Image Content Metric

DIGITAL images are subject to a wide variety of distortions during acquisition, processing, compression, storage, transmission and reproduction. Any of these may result in degradation of their visual quality. Hence, there has been an increasing need to develop quality measurement techniques that can predict perceived image/video quality automatically. These methods are useful in various image/video processing applications such as compression, communication, printing, display, analysis, registration, restoration, and enhancement. Subjective quality metrics are considered to give the most reliable results since, it is the end user who is judging the quality of the output in many applications. Subjective quality metrics are costly, time-consuming and impractical for real-time implementation and system integration. On the other hand, objective metrics like full-reference, reduced-reference, and no-reference metrics are most popular. This paper proposes an ideal no-reference measure that is useful for the parameter optimization problem and it takes care of both noise and blur on the reconstructed image into account. The experimental results have shown that the technique works well with images with various kinds of noise. Keywords-Image Visual quality, Objective Metrics, No-Reference Metrics, Image noise, Parameter Optimization,Steering Kernel Regression Algorithm