条盒三维视觉检测中结构光光条中心提取方法

摘要

In order to improve the accuracy of case three-dimensional measuring system based on structured light, the extraction method of light stripes center based on Hough transform and least square was proposed. Firstly, the target structured light was segmented by the median filter and threshold segmenting method which uses the difference of R, G and B in color model. Secondly, multiple iterative thinning was used to deal with the green light area and red light area respectively to obtain their skeleton. Thirdly, the points were classified according to the lines which are obtained by Hough transform. Finally, least square was used to fit the points sectionalized and obtain the center line. A series of experiments have been performed to test the proposed method. It is proven that the proposed automatic extraction method is effective and advanced.%为了提高基于结构光的条盒三维视觉检测中尺寸测量的准确度，研究和提出了霍夫变换和最小二乘法相结合的结构光中心线提取方法。首先，依据图像颜色模型的R、G、B分量差值并结合中值滤波对目标结构光分进行分割提取；其次，分别对绿光与红光区域进行多次迭代细化处理获取骨架；然后，根据霍夫变拟合的直线段对目标点集进行分类；最后，使用最小二乘法分别对不同的点集进行分段拟合，获取结构光中心线。实验结果表明，该方法能够准确地提取出结构光光条中心，具有良好的稳定性，为条盒三维视觉检测工作提供了准确的数据依据。