基于机器视觉的汽车门锁自动检测系统研究

摘要

In order to solve the problems of labor-intensive,low efficiency and accuracy in car door manual detection,designed a car door lock automatic detection system based on machine vision which consisted of hardware and software.Due to the low con-trast surface,the number of assembly parts and the complex of structure in car door lock,studied the improved distance preserving level set image segmentation algorithm and improved SIFT image matching algorithm,programmed the image processing software and system interface software.The test results show that the system hardware selection is rational,the system software is highly stable and efficient which improved the efficiency and accuracy in detection process of the car door lock with practical signifi-cance and high economic value.Meanwhile,the system can also be extended to other industrial assembly quality detection.%为了解决汽车门锁人工检测劳动强度大、效率和精度低的问题，设计了一套基于机器视觉的汽车门锁自动检测系统，该系统由硬件系统和软件系统组成。针对门锁表面对比度低、装配件多和结构复杂的问题，研究了改进的距离保持水平集图像分割算法和改进的SIFT图像匹配算法，最后将这些算法在计算机中编程实现，完成图像处理软件与系统界面软件设计。测试结果表明，该系统硬件选型合理，软件稳定性好、效率高，提高了汽车门锁检测的效率与精度，具有很高的实际意义和经济价值。同时，该系统也可推广到其他工业装配件的质量检测中。