改进型纯相位相关亚像素级安瓿药液图像配准研究

Research on ampoule liquid medicine sub-pixel image registration based on modified phase-only correlation

提出了一种基于改进的纯相位相关亚像素级安瓿药液图像配准算法,有效解决了安瓿药液可见异物视觉检测系统中,由于机械零部件加工装配精度、受力偏载等因素引起的平台振动以及跟踪摆臂与图像传感器的不同步,使获取的安瓿药液序列图像出现不同程度随机方向上的偏移造成的图像处理算法难度大的难题,大幅降低了运算量.实验数据表明,文中方法在保持同等配准精度和误差的情况下,执行时间由最高的9.1843 s减少到0.1331 s,提高了图像配准的实时性,为后续可见异物的精确提取奠定了坚实的基础.

Mechanical parts＇ manufacturing and assembling accuracy, offset load force, asynchrony between tracing arm and CCD cameras can cause random displacements at different degree during ampoule sequential image acquisition in vision-based foreign particle image inspection system. These displacements will bring great challenges to the recognition of the visible particles in the liquid. Therefore, a simplified phase-only correlation based on sub-pixel image registration algorithm is proposed which can greatly reduce the computation complexity. According to the experiments, the running time is reduced sharply from maximal 9.1843 s to 0.1331 s under the condition of keeping the same registration accuracy and error. The proposed algorithm improves the registration real-time and lays a solid foundation for the following foreign particleextraction.