Wide Field Multi-Objects Position Detection Through Digital Close-Range Photogrammetry
Yi Jin,Chao Zhai,Yonggang Gu
DOI: https://doi.org/10.1117/12.835051
2009-01-01
Abstract:Digital close-range measurement technique is developed from the geodesy surveying and the photogrammetry surveying, at present it is mainly applied to wide field detection with high relative accuracy, but its absolute accuracy is not high, about one micron to one millimeter. In this paper we apply this technique to wide field multi-objects optical fibers' space coordinates detection, a set of non-contact on-line detection system is designed and the accuracy of position detection is less than 0.03mm in the field of 600mm×600mm, that means this system has high relative accuracy and high absolute accuracy. In order to build this system, two aspects are mainly researched: CCD measurement error is the basis of photogrammetry, in order to control measurement error, the influence to measurement error affected by speckle recognition algorithm, the light source and camera space position is researched from experiments and theory. The result shows some significant conclusions: the detection error of Gravity Method is about 0.03 pixel; The uniformity of light source is important; The position detection of static goals through general photogrammetry can achieve high accuracy, about several microns, but when objects are moving, F number, lamp-house, speckle status and imaging size will probably cause additional measurement error, It is about a dozen microns to tens of microns. How to detect objects in wide field of view is a critical problem in photogrammetry, for the normal single frame photo-field is about hundreds of millimeter and the measuring field is several meters. In this paper, surface measuring through laser ranging device and Triangle Intersection Method are used to obtain fibers' positions in wide field of view. And some key technologies are adopted such as precise calibration of CCD camera, light rays adjustment method, subpixel image processing technology and corresponding image points matching. Multi-objects can be detected simultaneously in the wide field of view through theses technologies and the detection accuracy is increased. The experiments result shows that this system is stable and reliable, and it has the potential in precision measurement, industry measurement and other application areas.