基于迭代匹配的橡胶栓图像自适应旋转研究

Research on the image adaptive rotation based on the template matching

为解决橡胶栓检测中图像的自适应旋转这一难题,将图像识别中的最小距离匹配方法引入图像旋转,并利用迭代法实现了旋转匹配过程。首先,将旋转角度为0°的标准图像存储为模板,然后计算待检测图像与标准模板间的距离,再利用迭代运算使图像不断转动,减小图像与模板间的距离,直到该距离小于某一给定值。在原始的迭代算法的基础上,还提出了一种简化的迭代算法,在保证旋转精度的前提下具有更快的运算速度。实验结果表明,基于迭代匹配的橡胶栓图像自适应旋转方法精度高,计算速度快,能满足橡胶栓实时检测的要求。

Considering the difficulty of the image adaptive rotation in the surface inspection of rubber bolt, a new approach was proposed based on the template matching and iterative method. A standard image which rotated 0° was stored as the template. Then the distance between the sampled image and the template was calculated. The image was rotated by using the iterative method until the distance between the sampled image and the template decreased to a certain value. In order to rotate the image faster, a simplified iterative method was also presented based on the original iterative method. The experiment result shows that the approach based on the template matching and iterative method can assure the precision and the rapidity of image adaptive rotation.