论文部分内容阅读
研究目的:利用数字图像测量技术对河流模型实验中的河床地形测量研究。创新要点:以高质量的图像径向畸变校正为基础,依据多幅图像间映射换算以及单幅图像灰度信息变换两方面的研究,来提取三维地形信息。研究方法:使用考虑残差的一阶误差传播的最大似然估计方法求解图像与地形坐标的映射变换参数;借助漫反射光照模型,通过适当假设,推导出地形高程与图像灰度、灰度梯度及太阳方向的关系式。重要结论:1.以Devernay基于直线的图像畸变校正方法为基础,通过使用更高精度的Canny边缘检测算子提取图像边缘,黄金分割法和二次插值法相结合的最优化方法求取畸变系数等,可以得到较好的校正效果,求算的畸变系数精度更高,求算过程更快;2.利用图像与地面间的透视投影关系方法求解地形高程坐标,均匀并尽可能多地将控制点布置在控制区域内,使计算值更接近实测值,并满足河流演变实验的精度要求;3.通过推导关系式,理论上可以根据图像灰度、灰度梯度、太阳方向向量算得地形高程。
Research purposes: The study of riverbed topography measurement in river model experiment by using digital image measurement technology. Key points of innovation: Based on the high-quality image distortion correction, the 3D terrain information is extracted based on the mapping conversion between multiple images and the gray information transform of single image. Research methods: The maximum likelihood estimation method based on the first-order error propagation considering the residuals is used to solve the mapping transformation parameters between the image and the topographic coordinates. With the help of diffuse reflection model, the topography and the grayscale, And the relationship between the direction of the sun. The main conclusions are as follows: 1. Taking Devernay linear distortion correction method as the basis, using the more accurate Canny edge detection operator to extract the image edge, the golden section method and the quadratic interpolation method, the optimization method to get the distortion coefficient and so on , You can get a better correction effect, calculated the distortion coefficient higher accuracy, faster calculation process; 2 using the image and the ground perspective projection relationship between the method of solving terrain elevation coordinates, and as much as possible the control point Which is arranged in the control area so that the calculated value is closer to the measured value and meets the accuracy requirements of the river evolution experiment. 3. By deriving the relational formula, the terrain elevation can be calculated theoretically based on the image gray level, the grayscale gradient and the solar direction vector.