Abstract

The proposed method for monitoring of rigid body movement with six degrees of freedom is based on imaging the marked rigid body with a single TV camera. The cues marking the object are extracted out of one of the interlaced fields and stored as a group of pixel clouds. During the scanning time of the consecutive (complementary) interlaced field, the video input to the system is blocked and the pixel clouds are processed for cue identification and accurate localization. The position and orientation of the monitored object are computed by a single iteration (batch) method. The system is self-calibrating and determines automatically the initial position and orientation frame of reference. The location of the cues, as well as the minimal size of the cue clouds, must be discerned by the system. Only 4-6 cues have to be simultaneously detected and individually identified; the number of required cues is specified by the algorithm selected in a specific implementation.

© 1986 Optical Society of America