Temporal Capture and Sequence Reconstruction for Evaluation of Display Systems and Video Content

摘要

It is being increasingly recognized that temporal measures are an extremely important addition to conventional measures such as brightness, color, and contrast. In addition to flicker, color breakup, and visible flashes, inaccuracies in the portrayal of motion can greatly affect user acceptance of a system. The project team has previously reported on the use of image capture systems that can capture many images of the display for analysis to determine image characteristics and to reconstruct the temporal behavior of the display. This paper will describe how this technique can be extended to produce useful observation of multi-frame sequences. Issues for proper use of this technique include: 1) frame identification by cues in the frame image or other means, 2) externally generated image timestamps for both intra-frame and inter-frame timing, 3) tying image capture to the display's pixel transition timing to avoid erroneous measures taken during pixel transitions, 4) formatting test material to ensure repeatable display performance, and 5) proper sequence reconstruction and analysis to ensure that the reconstructed sequence is accurately representative of system behavior. Finally, the paper will discuss ways in which measurements taken using these techniques can be calibrated by comparison using human subjects and conventional measures.