Study On Principle And Implementation Method Of 3D Precision Positioning In Large Field Working Space

According to the disadvantage that it isn't able to implement large-scale measurement with high precision in field condition with the restriction of vision sensor's resolution and viewing field in conventional vision measurement, a kind of measuring method of 3D precision positioning in large field working space is studied. It provides an effective way to solve the potential contradiction between measuring range and measuring accuracy in conventional vision measurement. Visual characters which are called global control points are placed at the position of points in large-sized space. The position information of global control points are taken by special metrical camera. It is processed using high precision image processing technique, automatic matching of global control points technique and ray bundle adjustment technique. Then the three-dimensional information of global control points is able to be retrieved. It can accomplish the measurement flexibly, freely, rapidly and accurately.This dissertation's main work is as following:1. A method of 3D precision positioning in large-scale field condition space has been studied. It takes ray bundle adjustment technique as technological core. It can accomplish measuring process only using a CCD camera and some assistant measuring equipments such as global control points and reference bar.2. Scheme of global control points has been designed. A few coded points and plenty of uncoded points are used together. Size, shape, material of global control points and coding method, coding capacity of coded points are designed.3. Detection algorithm of global control points has been studied. It includes image preprocessing algorithm, precision extracting algorithm of uncoded points' center and automatic recognition algorithm of coded points' coded value.4. Automatic accurate registration algorithm of corresponding uncoded points in different measurement images has been studied. Coarse registration algorithm based on epipolar geometry constraint has been studied. The error of epipolar slope in coarse registration has been analyzed. Then precise registration algorithm using three images has been studied based on the analysis on epipolar slope error.5. Method of establishing and solving ray bundle constraint equations has been studied in theory based on close-range photogrammetry.6. System experiment has been designed. It includes detection experiment of global control points, automatic accurate matching experiment of uncoded points and verification experiment of whole system measurement precision.