Quality assessment of H.264/AVC encoded sequences

Methodology

The Degradation Category Rating (DCR) subjective test methodology has been used - see ITU-T P.910 for additional details.

In short, pairs of original/impaired sequences have been presented to each test participant. For each pair, participants were asked to judge the quality of the impaired sequence with respect to the original.

Participants

42 individuals have participated in this test. The participants were non-expert male and female with ages in the range 23 to 32 (most of them were students).

The subjects were screened for normal visual acuity and color blindness. One of them showed lack of visual acuity and another one was a color blind person.

Test room environment

Two LCD displays of the same model have been calibrated. The display and room characteristics used in the subjective tests are listed below:

  • Height of the picture shown in the screen: 8 cm
  • Viewing distance: 64 cm
  • Background room illumination: 13.45 lux
  • Peak luminance of the LCD screen: 95.8 lux
  • Luminance of inactive screen: 2.23 lux
  • Luminance of background behind the display: 10.15 lux
  • Ratio of luminance of inactive screen to peak luminance: 0.023
  • Ratio of luminance of background behind the display to peak of luminance: 0.14

Encoded sequences

The original sequences were encoded using the H.264/AVC JM 12.4 reference software.

The main encoder parameters were as follows:

  • Main profile, level 2.0
  • GOP structure: 15 frames length, with two B frames inserted between I/P frames
    IBBPBBPBBPBBPBBIBBP...
  • 8x8 DCT not allowed
  • Maximum of 3 reference frames used for motion estimation
  • Low complexity RD optimization mode
  • Original JM rate control algorithm

A template for the encoder configuration file can be found HERE.

Data

You can get all bitstreams for this test HERE.  

You can decode all bitstreams using the batch file provided (just place the JM H.264 decoder and bitstreams on the same directory).

MOS data is HERE.

Notes:

  • Sequence "Australia" 
    Only the first 250 frames were actually used (and displayed at 25 frames per second);

    •  
  • Sequence "Mobile"
    A sequence of 900 frames     that result from merging Mobile sequence in the forward direction, then backward, and then forward again has been encoded. From the resulting sequence, only the last part (last 300 frames) has been used for subjective quality assessment.