<title>Online database updating by change detection</title>

Abstract

Synthetic vision systems render artificial images of the world based on a database and position/attitude information of the aircraft. Due to both its static nature and inherent modelling errors, the database introduces anomalies in the synthetic imagery. Since it reflects at best a nominal state of the environment, it often requires updating via online measurements. The latter can vary from correction of pose and geometry to more complex operations such as marking the locations of detected obstacles. This paper presents an approach for detecting database geometric anomalies online. Since range sensors have a low update rate, they cannot be used for quick validation. Instead of range data, the proposed technique employs an imaging sensor, which can be of any type. It takes advantage of the fact that given a geometric model of the scene and known motion of the observer, the sensor image warping can be exactly predicted. If the geometry of the database is incorrect, the sensor image will not be correctly predicted and geometric differences will thus be detected. The algorithm is tested against simulated imagery and results show that it can correctly identify geometric anomalies. It can cope with known misalignment of the database and pose estimation errors. The technique is shown to be quite robust in low visibility conditions, given that the validated features are at least partially visible. It also automatically detects if a motion is sufficient for a given sensitivity.

<title>Database verification using an imaging sensor</title>

“…Information about the correctness of the geometry can thus be directly included in the database. Mathematical details on the anomaly detection algorithm can be found elsewhere 19 .…”