Roadside camera motion detection for automated speed measurement

In this paper, we present an algorithm to detect scene change that indicates that camera re-calibration is necessary. This work is in support of using un-calibrated traffic management roadside cameras for automated speed estimates. We construct an activity region using moving vehicle edges, and small differences in the activity region in consecutive images are used as a decision criteria for recalibrating the camera. The camera motion detection uses training sets of 10-second video sequences and detects moving vehicles using frame differencing. We present a validation of our algorithm using real-world traffic scenes.