Implementation of the Canny Edge Detection algorithm for a stereo vision system

Conference · Tue Dec 31 04:00:00 EST 1996

OSTI ID:459059

Wang, J R; Davis, T A; Lee, G K ^[1]

---

1. North Carolina State Univ., Raleigh, NC (United States)

There exists many applications in which three-dimensional information is necessary. For example, in manufacturing systems, parts inspection may require the extraction of three-dimensional information from two-dimensional images, through the use of a stereo vision system. In medical applications, one may wish to reconstruct a three-dimensional image of a human organ from two or more transducer images. An important component of three-dimensional reconstruction is edge detection, whereby an image boundary is separated from background, for further processing. In this paper, a modification of the Canny Edge Detection approach is suggested to extract an image from a cluttered background. The resulting cleaned image can then be sent to the image matching, interpolation and inverse perspective transformation blocks to reconstruct the 3-D scene. A brief discussion of the stereo vision system that has been developed at the Mars Mission Research Center (MMRC) is also presented. Results of a version of Canny Edge Detection algorithm show promise as an accurate edge extractor which may be used in the edge-pixel based binocular stereo vision system.

• OSTI ID: 459059
• Report Number(s): CONF-961225--
• Country of Publication: United States
• Language: English

Related Subjects

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATION
FOCUSING
IMAGE PROCESSING
ROBOTS
VISION