摘要 |
PROBLEM TO BE SOLVED: To more precisely attain tracking and image pickup of an object to be tracked, by discriminating pixels to track from background pixels by an AdaBoost discriminator, and using the likelihood calculated from the reply value of the AdaBoost discriminator by a particle filter. SOLUTION: In an automatic tracking device, a processing part 2 includes: a tracking processing means for performing tracking processing of tracking an object to be tracked, based on an image picked up by a camera 1; and a control means for controlling the pan, tilt, and zoom of the camera 1 so that the object to track can be tracked by the camera 1 according to the result of the tracking processing. The tracking processing means includes a position estimation means for estimating the position of an object to track by a particle filter configured of a plurality of particles by using the positions of pixels as states based on the image picked by the camera 1. Regarding the particles, the particle filter uses the calculated likelihood from the reply value of an AdaBoost discriminator, constructed to discriminate whether the pixels are the pixels to be tracked or are background pixels, based on one or more featured values related with the pixels, that is, a reply value based on the featured values related to the pixels of the positions of the particles. COPYRIGHT: (C)2010,JPO&INPIT
|