University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Graduation Talks


Perceptual Video Quality Assessment and Applications to Videos from Digital Camcorder System


Kongfeng Zhu, University Konstanz
Konstanz, Germany

date & place

Wednesday, 25.04.2012, 15:15 h
Room C252


Images and videos are subject to a wide variety of distortions during acquisition, digitizing, processing, restoration, compres- sion, storage, transmission and reproduction, any of which may result in degradation in visual quality perceived by human users. To characterize an image or video system, it is important to as- sess the quality of output image or video. The most reliable way of assessing the quality of an image or a video is subjective eval- uation, where a number of human users are asked to evaluate the perceived quality to get the so-called mean opinion score (MOS). However, this approach is too cumbersome, slow and expensive for most applications. To automatically assess the quality of an image or a video as seen by an average human observer in real time, a great amount of efforts have been made on objective quality evaluation. The ob- jective image/video quality assessment (IQA/VQA) metrics (or algorithms) can be classified into full-reference (FR), reduced- reference (RR), and no-reference (NR) depending on whether a reference, partial information about a reference, or no reference is available in assessing the quality, respectively. The challenge is that the objective quality assessment system would be able to measure image or video impairments like how a human being perceives them. NR quality assessment generally reduces the storage require- ments of the algorithm. This leads to considerable savings, espe- cially in the case of video signals. Also, in certain applications such as evaluating the performance of digital camera and cam- corder systems, the original uncorrupted images or videos are often unavailable because the imaging (sensing) and recording system is a black box one cannot probe into. However, due to the lack of any information about the distortion-free references, NR VQA is the hardest among the three VQA paradigms. My thesis will focus on no-reference video quality assessment, aiming to assess the perceptual quality of videos which are dis- torted by consumer-level digital camcorder system (DCS), espe- cially lossy video compression.