Omnitor logo
Start Products Projects Support About Omnitor Contact us
Allan eC decoration picture Blue color Communication for all
Video quality
Measuring video quality

Technical supervision and advice for producing video in sign language with the help of videoediting and videocompression technology.

We have a special video for measuring image frequency, sharpness and image quality during use of picture phones of different brands. It is used internationally and can be ordered from Omnitor, in a swedish or an english version.

Above: An example for measuring the frame-rate on the other side by the test tape.

There are nine video sequences collected on the video.

  1. "Silent Voice", an international reference video sequence with American Sign Language, used by the video coding standards groups in their work. Can be used in discussions with designers. No specific measurement is done on this sequence, but a general judgement of  smoothness of movements and sharpness of details can be done.

  2. "Fast" . A sequence, rapidly signed . A sign language user will get a first impression on how useful the system is by looking at the reproduction of this sequence.

  3. "Slow". A sequence, slowly signed . This demonstrates that it is possible to reduce signing speed with some effort. Also this is meant to give the sign language user a first impression if the system at least can carry sign language at a reduced rate.

  4. "Frame rate". A picture update rate measurement sequence. The user shall judge or count how many times per second the picture is changed. It is important for smooth sign language transmission to have a picture update rate of at least 12 pictures per second. Improvements in perception can be observed if the rate is higher, up to 25 pictures per second. It is very important to measure this factor with sign language in the picture. The measurement tool is a pattern of 25 squares, with a dot moving from one square to next on every picture. During one second, all squares get one visit each of the dot. The resulting rate is evaluated simply by counting or judging how many squares the dot visits at the receiving end during one second with vivid signing in the picture marked "signing".

  5. "Static resolution". A collection of still striped patterns for static resolution measurement (see figure 3 below). The user checks which striped pattern can be seen at the receiving side. The number beside that pattern gives an indication on the resolution of the picture while it is still, measured in vertical lines over the picture. There are also still pictures of a person in different sizes, in order to give a chance to judge if the resolution is enough for viewing one, two or four signing persons.

  6. "Blur". A sequence with three signs containing more and more motion, for judging blur of details in motion . The resolution of the picture is often reduced during motion. The user shall judge if the fingers, eyes and mouth can be seen in a sign with little, medium and a lot of movement. For good perception of vivid sign language, the details should be perceivable with vivid signing.

  7. "Sync". A sequence to measure the synchronism between sound and video . A flashing square and a sound is sent at the same time. The user shall check on a timescale the difference in time between sound and image at the receiving end. For lip-reading supported by hearing this value should probably not be higher than 0.1 seconds. This value is not relevant for pure sign language use, and we have not verified it in this project.

  8. "Lip-reading". A sequence of a talking person, for a brief observation on the usefulness for lip-reading . A user, interested in using lip-reading, can get a brief judgement of the movements of the lips. We have only provided the sequence here for reference, and not provided any evaluation method for this factor.

  9. "Delay". A sequence for measuring the delay of the picture from video input in one end to display in the other end. If the picture is too much delayed, the turn-taking in the dialogue is disturbed. A flashing square and a sharp sound is sent simultaneously. The sound should be sent through a direct cable to be played or displayed at the receiving end. The time difference is observed. The time should be less than 0.5 seconds for a smooth conversation. At a value of more than 1.2 seconds it is disturbing.

Notes:

The quality requirements for sign language have been published in a
publication form the ITU. ITU-T [H.Supp-1] Supplement 1 (05/99) to Series H - Application profile - Sign language and lip-reading real-time conversation using low bit-rate video communication. It is available from http://www.itu.int/itudoc/itu-t/rec/h/index.html

Hellström Gunnar, Quality Measurement on Video Communication for Sign Language, Proceedings of the 16th International Symposium on Human Factors in Telecommunications - Oslo, Norway, May 12-16, 1997, pp 217-224. ISBN 82-994236-0-0.