| 摘要 |
For a tracking of a target object in a time series of frames of image data, a tracking object designation acceptor accepts a designation of a tracking object, a target color setter sets a color of the designated tracking object as a target color, and a particle filter processor employs particles for measurements to determine color likelihoods by comparison between the target color and colors in vicinities of particles, works, as the color likelihoods meet a criterion, to estimate a region of the tracking object in a frame of image data in accordance with results of the measurements, and as the color likelihoods fails to meet the criterion, to use particles, for measurements to determine luminance likelihoods based on luminance differences between frames of image data in a time series of frames of image data, and estimate a region of the tracking object in a frame of image data in accordance with results of the measurements, and updates the target color by a color in either estimated region.
|
