冷凝器管口图像多目标搜索与定位算法  

Searching and locating algorithm for multi-tube orifices similar to circle

作　　者：刘畅[1] 王耀南[1] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082

出　　处：《仪器仪表学报》2011年第11期2515-2522,共8页Chinese Journal of Scientific Instrument

基　　金：国家863计划(2008AA04Z214);国家自然科学基金重点项目(60835004;60775047)资助

摘　　要：提出一种基于图像行列像素值累加统计规律的多管口识别与定位算法,可实现对冷凝器管板平面图像中多管口的轮廓识别与质心定位。通过行列像素值累加和求均值进行质心粗略定位,求取管口横纵轴径长,绘制矩形窗口初步筛选出多个待定位的类圆管口目标;针对不同的定位精度要求,采用采样密度可调整的轮廓拟合定位算法,实现对多管口数量统计、采样网格可调的轮廓近似、链码编码和质心定位。实验证明该算法能快速识别图像中的多个管口,通过链码编码完成图像数据压缩,具有较好的准确性。A searching and locating algorithm for multi-tube orifices similar to circle in a tubeplate image is presented,which is based on the statistical law of the row and column pixel cumulative values on the binary image.The proposed algorithm can realize boundary fitting,calculate the diameter arguments of orifices in horizontal and vertical directions and locate the approximate centroids of multi-tube orifices similar to circle.Through preprocessing and analyzing the statistical law of the row and column pixel values,the multi-tube orifices can be screened out using rectangular windows with appropriate sizes.Aiming at different demands of locating precision,an adaptive boundary fitting algorithm is applied,which can identify the multi-tube orifices from background,count the number of orifices,fit boundaries with adjustable sampling grid,encode the boundaries in chaincode and locate the centroids of the orfices.Experiment results show that this algorithm can recognize multi-tube orifices in an image quickly and finish data compression with chain coding,which shows a good accuracy.

关 键 词：多管口 轮廓拟合 采样网格 质心定位 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]