| 摘要 |
<P>PROBLEM TO BE SOLVED: To enhance accuracy of position estimation of an object when tracking the object. <P>SOLUTION: Position predicting means 14 predicts a position of an object in object frame images to be processed based on a movement of the object among more than two frame images. Position map generating means 15 generates a correspondence between a plurality of positions positioned at a periphery of the predicted position and a value indicating an existence probability of the object as a position map. Feature amount calculating means 16 calculates a feature amount, which indicates similarities of the object, of each of the plurality of positions positioned in the vicinity of the predicted position to generate an appearance map. Position estimating means 17 combines the position map and the appearance map to estimate the position of the object in the object frame images to be processed. <P>COPYRIGHT: (C)2012,JPO&INPIT |
