The Research Of Measuring Technologies For Geometric Size Based On Depth From Defocus

When use the camera to obtain a single image and do non-contact measurement for geometric size, as the measured object exist the height difference in the axis orientation, and it will result that the different sections can not focus at the same time or the different fuzzy degrees. So it will affect the measurement accuracy. For these parts, the extracting depth information is the key of the high-precision measurement. Stereo vision is the basic tool and instrument in the modern 3D measurement. Binocular vision and multi-vision have complex models, and have the problem of corresponding points'matching. This dissertation studies these problems, uses the depth from defocus of the monocular vision to gain the depth, and measures the high-precision's geometric size of different sections'according to the depth.Three kinds of the DFD algorithms are discussed in the dissertation. They are the DFD algorithm proposed by Pentland, Subbarao and the DFD with the telecentric lens. Point out the limitation of three methods in the application according to their principles. On the basis of this, a new method of depth from defocus is advanced. It is the depth measuring method that based on using edge spread level to describe defocusing degrees of images.The edge is the most basic characteristic of an image. The greater blur degree, the more serious edge diffusion, and reflecting in the spatial frequency are the attenuation of the high frequency components. The dissertation obtains the depth of object according to the relationship between the specific MTF values of the modulation transfer function curve and the degree of defocusing. Through the calibration of the whole measurement system, we gain the relation between distance and the degree of defocusing, and measure the depth. Finally, the dissertation tests the mentioned method, reunifies and completes the different sections' geometric size measured of single image based on depth from defocus.