No abstract

No abstract

This paper reports on a series of experiments designed to examine the subjective quality of HDTV encoded video at a distance of 4H from a 46 inch widescreen LCD TV. A common set of video sequences varying in spatial and temporal complexity were encoded at different H.264 encoding profiles. Each video source used in the tests was available in 720p at 50 fps (720p/50) and in 1080i at 25 fps (1080i/25) HDTV formats. Each clip was then encoded in CBR mode using H.264 at bit rates ranging from 1 up to 20Mbit/s with all the other encoding parameters identical. This approach enabled direct comparison across bit rates and formats most relevant to HDTV broadcast services. Two subjective tests were performed, one per HDTV format. The single-stimulus subjective test method with the ACR rating scale was employed, with the clips played in random order. A total of 15 non-expert subjects participated in each test, a different subject sample was used for each test, making 30 participants in total. Using a mean opinion score performance benchmark of 4 corresponding to the rating "Good" on the five-point ACR scale, it was found that for 720p/50 content on average a bit rate of 5Mbit/s was need to meet the performance benchmark. For demanding 720p/50 content a bit rate of 8Mbit/s was needed to achieve a MOS of 4, whereas a bit rate of 2Mbit/s was adequate for easy-to-encode 720p/50 content. For 1080i/25 content, an average bit rate of 8Mbit/s was needed to attain an average MOS of 4. For difficult 1080i/25 content a bit rate of 11Mbit/s was necessary, this bit rate dropped to 3Mbit/s for less demanding 1080i/25 content.

No abstract

No abstract