Three-dimensional Measurement Technology Research Based On Multiple View Geometry

The visual measurement technology has been widely used in such fields as industrial production, measurement and monitoring, etc. Camera calibration is a very important link of visual measurement, because it directly determines the accuracy and precision of visual measurement. The camera is an essential equipment of visual measurement and camera calibration is one of the most important steps of visual measurement. The camera calibration precision will severely affect the visual measuring results. Usually camera is calibrated by using reference points or calibration targets. However, it will reduce the efficiency of the visual measurement. Moreover, the calibration target would deform, which will influence the measurement accuracy seriously.As a result, how to achieve accuracy three-dimensional measurement and rotation angle measurement by using the single CCD camera without any calibration target is researched in this dissertation from following several aspects:1. Camera self-calibrationCalibrating the CCD camera by self-calibration technology can get rid of the dependence on the calibration target. The camera self-calibration methods are multitudinous. A camera self-calibration method based on absolute quadric is proposed in this dissertation.2. Multiple view geometry based on monocular visionMultiple view geometry in computer vision is combined with monocular vision measurement in this dissertation. And a three-dimensional measurement method based on single camera has been developed.3. Rotation angle measurementOn the basis of three-dimensional measurement based on single camera self-calibration, an accurate, efficient and stable rotation angle measurement method is developed in this dissertation.Because the three-dimensional measurement technology researched in this dissertation doesn’t depend on the calibration target and just need a single camera to complete the measurement. As a result, this measurement technique is more convenient, flexible and practical. Furthermore, the three-dimensional measurement based on vision proposed in this dissertation has higher measurement precision and can satisfy the requires of industrial manufacture and engineering measurement completely.