Recursive Bayesian Calibration of Depth Sensors With Non-Overlapping Views

Provided by: Karlsruhe Institute of Technology
Topic: Mobility
Format: PDF
Setting up a network of multiple depth sensors introduces several issues that need to be addressed. For example, the popular Microsoft Kinect TM device acquires depth information using an active measurement system which projects an infrared pattern. This has important consequences in setups with overlapping fields of view, where the interference caused by overlaying patterns becomes an important factor for measurement quality. One solution for this problem is a time multiplexing approach where measurements are taken only by a single sensor at any given time. However, this approach becomes ineffective for applications such as tracking, where it may be necessary to switch between sensors very quickly, requiring the projectors to be rapidly powered on and off.

Find By Topic