Many surveillance systems record in a multiplexed format.
The two most common styles appear as multiple images or alternating images on the screen. We can convert both so that you can view only the needed video. These videos are often not in real time. We can adjust the speed for easier viewing and presentation. We can then perform enhancement and still capture techniques on the video.
Time-multiplexed video recordings must be demultiplexed for effective analysis by investigators: a task often made difficult by the proprietary nature of the recording equipment. In addition to a lack of standards, some systems may record scenes from different cameras with a time-varying sequence.
To address this issue, a content-based image analysis technique for extracting a specified camera view from a multiplexed video sequence is presented. A single frame from the specified view is used as a reference to which subsequent frames are compared. Frames matching the reference within a specified similarity threshold are passed on to the demultiplexed output. An optional region of interest may be used to select an area of the scene that is unique to that camera view but relatively consistent over time. The software implementation of this technique allows multiple camera views to be demultiplexed simultaneously from a digitized stream of multiplexed video.