基于随机Hough变换的复杂条件下圆检测与数目辨识  

Circle detection and its number identification in complex condition based on random Hough transform

作　　者：周封[1] 杨超[1] 王晨光[1] 王丙全[1] 刘健[1] 

机构地区：[1]哈尔滨理工大学,哈尔滨150080

出　　处：《仪器仪表学报》2013年第3期622-628,共7页Chinese Journal of Scientific Instrument

摘　　要：在复杂背景和光照条件中检测不规则分布的圆,并且准确辨识其数目,可为一些专用的智能仪器提供理论依据,有着广泛的工业应用价值。针对需要检测的圆数量大、光照不均匀、光照强度变化、分布不规则、相互间有遮挡及边界模糊和灰度接近等复杂条件,在随机Hough变换的基础上进行改进,利用圆的特性和梯度算法对圆进行判定,可以在复杂条件下对圆进行准确的检测判定和数目辨识。通过对钢管储存现场图像检测的计算与分析,证明此算法在较高的干扰下可以准确地对钢管进行定位和数目辨识,对于相互遮挡的钢管也可以准确识别。相对于目前通用的圆检测算法,本算法可以达到更好的识别效果,完全满足工业实际应用的要求。Detecting irregularly distributed circles in complex background and illumination condition and accurately identifying the number of the circles have a wide industrial application value, which provides the theoretical basis for some special intelligent instruments. Aiming at the problems of large number of circles to be detected, uneven illumination,light intensity variation,irregular circle distribution, circle mutual occlusion, blur boundary and adjacent grey degree, this paper proposes an improved gradient algorithm based on random Hough transform, which can accurately detect the circles and identify the number of circles under complex conditions based on the characteristics of the circles. The steel tube images in the storage field were detected and analyzed. The results prove that the proposed algorithm can accurately locate the steel tubes and identify the number of the steel tubes under high disturbance condition, and also can accurately identify the occluded steel tubes. Compared with the existing commonly used detection algorithm, this algorithm can achieve higher recognition effect and fully satisfies the requirements of practical industrial applications.

关 键 词：图像处理 复杂条件 圆检测 随机HOUGH变换 数目识别 梯度算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]