基于计算机视觉的位移测试方法研究与实现

Development of vision-based displacement test method

针对Hough变换、最优圆拟合两种圆检测算法进行了圆检测效果、算法应用范围及计算效率的对比分析;研发了基于圆检测算法的非接触性动态位移测试方法并应用于大型振动台实验。结果表明:Hough变换圆检测及最优圆拟合圆检测均具有较好的圆参数检测效果。Hough变换圆检测方法更适用于测试背景简单且需单次识别多个标靶圆的工况;最优圆拟合圆检测方法更适用于测试背景复杂的工况。应用非接触性位移测试方法于大型振动台试验中,测试了模型地基土的位移反应。应用基于机器视觉技术的位移测试方法可以有效解决复杂工况下的位移测试问题。

The circle detection effect, algorithm application range and calculation efficiency of different circle detection algorithms were analysed comparatively. Base on the displacement test method was developed and applied to large-scale circle detection algorithms, a vision-based dynamic shaking table tests. The results show that： both Hough transform circle detection algorithm and optimal fitting circle detection algorithm have good performance in circle parameters detection. The Hough transform circle detection algorithm is more suitable for testing under such operation conditions in which the test background is simple or with the need of identifying multiple target circles at once. The optimal fitting circle detection method is more suitable for testing under such operation conditions in which the test background is complex. By applying the vision-based displacement testing method to large-scale deformation of a model of soil foundation was tested. The method proposed can effectively displacement testing in complex test conditions. shaking solve table tests, the the problem of