K.S. Younis, B. Aldawood, G.A. Abandah, S. Alsisan (Jordan), and J. Igual (Spain)
Digital image processing, video processing, pattern recognition, motion detection and estimation, image sequence processing
A system that automates monitoring bird behavior in Jordan is built. This system is designed to help ornithologists in monitoring bird behavior using infra red camera directed on the nest box of the targeted bird; the House Sparrow. The system processes consecutive camera images to detect motion as an indication of bird nest activity. It records video clips of bird activity and saves them with date and time information. The system uses region-of-interest based computation and the moments of the image intensity histogram. We optimize the selection of the threshold for motion detection by using the receiver operating characteristics (ROC) analysis. Obtained results depend on the user’s threshold sensitivity selection. Accuracy of true activity detection ranges from 80% to 92% with false detection percentages from 10% to 44%, respectively. The system can trigger multiple events on motion detection including video capturing and various alerts.
Important Links:
Go Back
