Perceptual Video Quality Assessment Tool

Abstract: Compressing the video data is necessary for video transmission over networks with limited resources. Video quality at the end is based on the compression techniques used. In this research work, we mainly focused on compressing video files using H.264 encoding by considering some parameters and conducting a subjective test on the resulting compressed videos in a specially developed Perceptual Video Quality Assessment (PVQA) tool. This PVQA tool was developed as part of the thesis project. For the compression techniques, we have considered spatial resolutions (i.e. VGA and QVGA), temporal resolutions (i.e. 25fps and 12.5fps) and bit rate (i.e. 300kbps and 600kbps). By considering these parameters for three different test sequences, we down scaled and encoded to transmit over lossless channel. The received data is decoded and up scaled using open source software’s. For encoding and decoding of videos in H.264, we used JM reference software. For scaling the videos spatially and temporally, both FFMPEG and Virtual Dub are used. For up scaling the decoded video in spatial resolution (QVGA to VGA) two filter methods are used i.e. Bicubic Interpolation and Nearest Neighbor technique in Virtual Dub. For up scaling in temporal resolution (12.5fps to 25fps) the repeat frame method of Virtual Dub is used. These generated videos were evaluated by conducting subjective tests to determine the best technique that is suitable for transmission over a bit rate limited network. The developed PVQA tool satisfies the requirements of International Telecommunication Union (ITU) standard ITU-R BT.500-12. The PVQA tool will ask the subjects to grade the videos quality according to his/her experience and these grades are saved in excel sheets. The PVQA tool was developed under Java environment, with the VLC player embedded. The tool was designed to support the video. Results of the subjective video quality survey are summarized and finally conclusions are given.