基于图像分割的目标尺寸特征测量-《计算机技术与发展》

文章信息/Info

Author(s):: WANG Ying; WANG Peng－fei; YANG Yu－wang; School of Computer Science and Engineering，Nanjing University of Science and Technology，Nanjing 210094，China

Keywords:: image segmentation; feature measurement; Hough transform; minimum enclosing rectangle

摘要:: 主要研究了两种基于图像分割的工业元件尺寸测量方法，分别是基于霍夫变换和基于最小外接矩形算法的测量算法，并研发了图像测量系统。系统中针对不同目标形状，分别应用了两种方法，对于矩形度高于阈值的元件使用改进的霍夫变换检测出元件的四条边，而后组合边检测出矩形，最后应用几何数学解析出元件尺寸;针对矩形度低于阈值的元件，使用新的最小外接矩形算法提取最小外接矩形，然后应用几何数学解析得到元件尺寸。对于两种方法分别进行实验，对比它们的实验结果并进行误差分析。结果验证了这两种方法在特征测量上的有效性及互补性。

Abstract:: In this paper，we mainly study the two methods of measuring dimensions of industrial components based on image segmentation，including the ones based on Hough transform and based on the minimum bounding rectangle algorithm，and develope an image measurement system．Two methods are applied respectively according to different target shapes in the system．When the rectangular degree is higher than the component threshold we use Hough transform to detect four edges of the component，and then integrate them for detection of the rectangle．Finally the component size is analyzed by geometry math．For the component with its rectangular degrees below threshold，we extract a minimum bounding rectangle by the minimum bounding rectangle algorithm，and then utilize the geometry math for its size．Two methods are experimented and compared in the error analysis，which shows that they are both effective and complementary．