Blur estimation methods for system of audiovisual monitoring of meeting participants

Abstract

The problem of analysis and development of mathematical methods and software for estimation of blur and other quality metrics of digital images is considered. The classification of modern methods of blur estimation used for real-time systems is presented. During experiments several methods of image part segmentation were applied for enhancement of the processing speed and reliability of image quality assessment. The proposed method of preliminary extraction of face area on the image and estimation of its blur was successfully used for elimination of distorted images at the automatic registration of participants in the intelligent meeting room.