| Summary: | 碩士 === 國立中興大學 === 機械工程學系所 === 100 === In computer vision, get a useful information directly from the one containing the image of the three-dimensional information has been difficult, the range image segmentation is a very important initial at the object surface in the original image is dividedbasic geometric elements and geometric elements with plane, spherical, cylindrical and conical surface.
A mechanical components often contain many different geometric components and if the direct analysis to the image all the pixels likely to cause long computation time, so in this paper,we use of the hierarchical clustering , the first image is divided into small blocksby block to find the components.
In this paper,the first step use sign the the Gaussian curvature and mean curvature of the surface morphology,than we use RANSAC algorithm to identify all the surface elements from the sign image.
|
|---|
