距离保持水平集在汽车门锁图像分割中的应用研究

Research on the application of distant preserving level set algorithm in car door lock image segment

针对基于机器视觉的汽车门锁装配尺寸检测系统中摇臂距离的检测问题,研究了距离保持水平集的图像分割算法。该算法是将轮廓表示为一个符号函数,符号函数的零水平面与实际轮廓相对应,并在能量泛函中加入了内部能量泛函,表示两个函数之间的差异。实验表明,与一般的基于边缘检测的图像分割算法相比较,该算法能够很好的解决汽车门锁图像中摇臂外弧线分割问题,提高了汽车门锁检测系统的精度。同时,该算法也可应用到其他类似工业装配件局部图像分割中。

This paper focus on the research of image segment algorithm based on distant preserving level set, which is used in rocker distance measurement in the car door lock assembly detection system, which is based on machine vision. The algorithm regard profile as a sign function, and the zero horizontal contour of a sign function is in correspondence with the actual outline, more over, the algorithm add a internal energy functional to the energy functional, which represents the difference between the two functions. Experiments show that compared with the general image segmentation algorithm based on edge detection, the algorithm can solve the problem of image segmentation in car door lock＇s rocker outside arc, and improve the accuracy of car door lock auto detection system. Moreover, the algorithm can also be applied to other similar industrial product image segmentation.