Highlights
OCT 6, 2011 — PSCR delivered its PSVQ group’s latest research report, “Recorded-Video Quality Tests for Object Recognition Tasks,” to DHS. As with previous PSVQ reports, DHS will post this paper to its SAFECOM site.
The paper (available here) was written to provide public safety agencies with bit rate recommendations related to recorded video where object recognition, such as in a forensic application, is important.
The paper discusses:
How lighting, motion, and the size of the target play a role in determining bit rate recommendations.
Defining use cases for video applications, and then categorizing use cases that share certain important aspects, which is the concept behind the generalized use classes explained in the user guide developed by the VQiPS working group.
An experiment to explore the effects of network conditions upon video quality for public safety applications. Specifically, H.264 compression combined with resolution reduction to VGA—or 640x480 pixels—and CIF—or 352x288 pixels. These types of conditions are representative of video that has been transmitted to mobile devices.
Bit rate recommendations related to recorded video.
The most significant finding reported is that recognition rates were almost all statistically equivalent to recognition rates in the PSVQ group’s previous live video test. It is hypothesized that the particular recognition task under study is being performed at a cognitively low level—perhaps even at a subconscious level.
. . .
