多分辨率快速圆检测在抑菌圈测量系统中的应用

Application of fast multi-resolution circle detection in inhibition zone measurement system

本文提出了多分辨率快速圆检测的算法。基于多分辨率的框架,在低分辨率下快速检测圆心和半径的信息,然后映射到高分辨率中,确定圆的边缘和圆心所在的局部区域,减少处理单元的数量,并提出局部单圆验证算法获得精确的圆心和半径信息。本文提出的算法克服了圆检测精度与时空复杂度的矛盾,在高分辨率、多圆、多噪声的情况下,算法的效率下降较少。最后将圆检测算法应用于基于机器视觉的抗生素残留量检测系统中进行抑菌圈检测,结果表明处理时间短、检测精度高。

The paper introduces a high speed circle detection method based on multi-resolution theory. With low resolution, coarse circle center and radius information are detected with less time and space consumed. The result is used to determine the candidate region in which the potential circle exists. The paper proposes a novel local validation algorithm to obtain precise information of each single circle. The method introduced in this paper is designed for the application of detecting circles under high resolution and noisy circumstance. The method was used in antibiotic potency residue measurement system based on machine vision. Test results prove that the proposed method features high precision, fast speed.