The Accuracy Analysis And Control Of Digital Close Range Photogrammetry

For the high agility and applicability, digital close range photogrammetry has become a focus in optical non-contact measurement area. How to assure the measure precision is the most pivotal matter of close range photogrammetry. On the basis of a digital close range photogrammetry prototype system called AutoLocator, the accuracy-related aspects such as image processing, edge-detection, coded target center location, lens distortion compensation and three-dimensional coordinate computation are systematically investigated to reduce the measurement error as much as possible. The main achievements of this thesis are as follows:Edge detection of the coded targets is the very beginning of the process pipeline of AutoLocator system. First, several representative edge-detection methods are discussed. We adopt Canny edge detector and local surface fitting to extract edges with sub-pixel accuracy.The precision of the image coordinates of the coded target centers is one of the crucial factors for the system. In this paper, an improved intensity barycenter method is presented. The experimental results show that this method increases the measurement precision and result stability.The distortion error of perspective projection is discussed in depth. Base on the theory of perspective projection and the theory of spatial analytic geometry, a mathematic model of distortion error is established. Through simulation experiment, the rule of the distortion error is obtained. The camera lens distortion is another crucial factor for the system precision. In this paper, a new lens distortion correction method is established based on the model of lens distortion. Prior calibration is needless. The lens distortion correction is accomplished simultaneously with the camera pose estimation and the coded target reconstruction. This method not only improves the applicability of the photogrammetry system in industrial applications, but also increases the measurement precision.The methods presented in this paper have been applied to AutoLocator system. Within the experiments done, the 3D location precision is around 0.02mm/m。...