Title :
Acquisition of accurate distance information in stereovision using move-parallel method for teleoperation
Author :
Mizukami, Manami ; Hongte, Kim ; Takeno, Junichi
Author_Institution :
Robot & Sci. Lab., Meiji Univ., Kanagawa, Japan
fDate :
6/27/1905 12:00:00 AM
Abstract :
The authors have developed a new stereo image shooting method, named the move-parallel method. There are two conventional stereo image shooting methods: the parallel-view method and the cross-eyed method. The parallel-view method allows the user to shoot and supply images that are free from geometric distortion, and thus is suitable for shooting distant views in consideration of the significant visual burden. The cross-eyed method uses the cross point to adjust the parallax of the object, but the perception of space is inevitably distorted in terms of binocular parallax. The new move-parallel method generates geometric distortion-free images by using the same shooting technique as the parallel-view method and offers easy-to-view images equivalent to those obtained with the cross-eyed method. This is achieved by moving the captured right and left images until the parallax of the object to be fixated disappears by image processing. It is important in teleoperation to capture correct information on the distance to the object to be fixated and to other objects. The authors conducted several experiments to evaluate the features of the move-parallel method. This paper reports on the method, which reduces distortion of space perception in terms of binocular parallax, provides correct distance information to the observer, and reduces the visual burden of viewing stereo images.
Keywords :
data acquisition; stereo image processing; telecontrol; visual perception; accurate distance information acquisition; binocular parallax; cross point; cross-eyed method; easy-to-view images; geometric distortion-free images; image processing; move-parallel method; parallel-view method; space perception; stereo image shooting method; Cameras; Eyes; Focusing; Glass; Humans; Image generation; Image processing; Liquid crystal displays; Remote monitoring; Robots;
Conference_Titel :
Mechatronics and Automation, 2005 IEEE International Conference
Print_ISBN :
0-7803-9044-X
DOI :
10.1109/ICMA.2005.1626723