面向多目标透镜的连通域标记检测算法  

Connected Domain Label Detection Algorithm for Multi-target Lens

作　　者：谢智烜 姚红兵 范宁 陈枫 XIE Zhixuan;YAO Hongbing;FAN Ning;CHEN Feng(School of Mechanical Engineering,Jiangsu University,Zhenjiang 212000,China;School of Science,Hohai University,Nanjing 210098,China)

机构地区：[1]江苏大学机械工程学院,江苏镇江212000 [2]河海大学理学院,江苏南京210098

出　　处：《电子科技》2020年第4期50-54,共5页Electronic Science and Technology

基　　金：国家自然科学基金(51775253,61505071);江苏省自然科学基金(BK20150526);江苏大学引进人才科研启动项目(15JDG036,15JDG037)。

摘　　要：文中提出了一种适合FPGA实现的多目标连通域标记算法。该算法通过初次扫描时在图像的边框处做特殊的标记值,提取出环形具有内外双层轮廓的目标内部区域,进而将紧挨在一起的目标分离开以实现多目标的检测。该方案采用单次逐像素扫描的方法,对图像进行一次逐像素扫描即可完成连通域的标记和等价表的整理。实验结果表明,对一幅大小为M×N的图像,在不计算等价表递归时间的情况下,经过M×N×2个时钟周期后可识别出多个目标,且得到了每个目标对应区域的面积和外接矩形等参数。A multi-target connected domain labeling suitable algorithm for FPGA was proposed in the study.By making a special mark value at the edge of the image during the first scanning,the algorithm extracted ring inner region with inner and outer double-layer contours object to separate the close objects for multi-target detection.The scheme adopted a single pixel-by-pixel scanning method to perform one scanning of the image,and could accomplish the mark of connected regions and the sorting of equivalent tables.Experimental results showed that for an image with the size of M×N,without calculating the recursive time of the equivalent table,multiple targets could be identified after M×N×2 clock cycles.In addition,parameters including the area and the outer rectangle corresponding to each target were finally obtained.

关 键 词：图像处理 多目标图像 连通域标记 FPGA 机器视觉 二值图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]