摘要 |
PROBLEM TO BE SOLVED: To provide a device and a method for separating a sound source, by grasping and tracking an accurate position and the direction of a specific object existing as a plurality of sound sources, by fusing audio and video sensors. SOLUTION: This device has a voice true degree module for determining a voice true degree for expressing a true degree under an object tracked by a plurality of sounds received from the different direction, a picture true degree module for determining a picture true degree for expressing a true degree being the object tracked by a plurality of images arranged in the different direction in a picture, and a recognizing and tracking module for determining that a tracking object does not exist in the sound source or an image source in noncoincidence, by recognizing and tracking the object, by using a pair determined by the voice true degree and the picture true degree, when determining that the voice true degree coincides with the picture true degree, by determining whether or not the voice true degree coincides with the picture true degree. COPYRIGHT: (C)2006,JPO&NCIPI
|