DCT Domain Best Shot System

Abstract

Digital video sequences may lead to visible degradation due to different kind of artefacts. Among them, blurring degradation is easily visible and annoying also for image and video processing non-experts. Blurring distortion is commonly caused by the object motion, camera shake or out of focus. To avoid this kind of degradation, also due to video compression, quality metrics for measuring sharpness are important to suppress worst images, to take the best shot in a burst of subsequent pictures. In this way, we obtain better looking video for the final user. In this paper, we present a low cost DCT-based single step algorithm, which requires no additional memory, so it is suitable for real time processing. It is a simple but robust method and it obtains similar visual results compared to more sophisticated algorithm (also not DCT-based). Moreover, the measure proposed takes into consideration not only blurring, but also other visual quality attributes, like brightness and orientation.

Keywords

• discrete cosine transform
• image quality
• blurring
• brightness
• sharpness