二維影像於三維輪廓重建之研究

• Main Author: 馮信榮
• Other Authors: 楊宏智
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/59060980147467978510

碩士 === 國立臺灣大學 === 機械工程學研究所 === 89 === In visual system, the CCD (Charge Coupled Device) camera captures a two-dimensional image. The geometric relationship between this image and its 3-D object has a deficiency in depth data. If the information on depth (z) is combined with the data points (x, y) on the 2-D plane, then reconstruction of the 3-D contours of an object can be made. In this thesis, an on-line non-contact image system is presented. This on-line image-measuring system consists of a CCD camera with an optical lens, a frame grabber, and an automatic moving platform as the hardware system. In this research, the Sum-Modified-Laplacian (SML) operator is utilized to compute local measures of the quality of image focus. The discrete Gaussian filter and the histogram equalization are used to equalize the distribution of illumination and gray level values effectively, and thus reduce the effect of noise simultaneously. The image system utilizes a CCD camera with a fixed focal length. The CCD camera is moved along the z direction, and so is the focus plane. The region, where the focus plane intersects the object, is the best focused area. The quadratic equations are used to model the result and interpolate the focused measure values of sequential images captured by this system to obtain the most accurate depth estimates.