In this paper, we propose a method for automatic segmentation of subject regions using disparity maps for image retrieval by real-world objects. If one wants to look up a particular object in image database, the subject region should be properly extracted from the image as a key for image retrieval. In our approach, we assume that we have the actual object in the real world to be looked up in image database. Taking advantage of disparity maps using a commercially available stereo range finder, subject region extraction is automated. We assume that the x-coordinate, y-coordinate, and disparity obtained from disparity map should yield multimodal Gaussian probability distribution, and subject regions are extracted by properly selecting particular modes of Gaussian densities. We construct a prototype system to retrieve images by real-world objects and show that adaptive and quick automatic extraction of the subject region could be achieved. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 35(1): 47–57, 2004; Published online in Wiley InterScience (). DOI 10.1002sscj.10389
[1]
Dragutin Petkovic,et al.
Query by Image and Video Content: The QBIC System
,
1995,
Computer.
[2]
Sameer A. Nene,et al.
A simple algorithm for nearest neighbor search in high dimensions
,
1997
.
[3]
Giuseppe Riva,et al.
Treating body-image disturbances
,
1997,
CACM.
[4]
Keinosuke Fukunaga,et al.
Introduction to statistical pattern recognition (2nd ed.)
,
1990
.
[5]
Alex Pentland,et al.
Pfinder: Real-Time Tracking of the Human Body
,
1997,
IEEE Trans. Pattern Anal. Mach. Intell..
[6]
Hiroshi Murase,et al.
Visual learning and recognition of 3-d objects from appearance
,
2005,
International Journal of Computer Vision.