基于广义点摄影测量的圆和圆角矩形三维重建

Reconstruction of circles and round rectangles by generalized point photogrammetry

为了解决目前工业制造中圆和圆角矩形高精度误差检测方面所面临的难题,采用广义点摄影测量理论进行高精度三维重建与误差检测.论述了广义点摄影测量的数学模型,建立二维影像与待测目标参数间的函数关系,根据切线的斜率确定所需使用的数学关系式,并基于已知的像片内外方位元素,采用最小二乘模板匹配和物方直接解方法相结合进行圆和圆角矩形的三维重建和误差检测.实际图像数据的实验结果表明,由于采用了大量冗余观测值和相应约束条件按照最小二乘法求取整体最优解,可以获得0.03mm左右的三维重建和误差检测精度.

To resolve the faced problems of automatic inspection of circles and round rectangles in industrial applications, a new three-dimensional reconstruction technique is proposed based on generalized point photogrammetry and least squares template matching. A model of generalized point photogrammetry is addressed. Since coordinates of generalized image point are result of image template matching and also functions of circle and round rectangle, the parameters of circle and round rectangle are related with image coordinates directly. Thus parameters of circle and round rectangle can be obtained from several images by least squares template matching and immediate object space solution with known camera parameters. The proposed approach was tested with real image data. A large number of redundant observations and constraints were used during the least squares adjustment process, so 0. 03 mm precision of three-dimensional reconstruction and visual inspection was obtained, which satisfies the industrial applications.