基于支持向量机的实时智能微生物识别技术  

Real-time and Intelligent Microbe Recognition Technology Based on Support Vector Machine

作　　者：李乔[1] 万薇 张海明[3] Li Qiao;Wan Wei;Zhang Haiming(The 28th Research Institute of China Electronics Technology Group Corporation,Nanjing 210007,China;Jiangxi Electric Power Company Economic Research Institute,Nanchang 330043,China;No.73228 Troop of PLA,Hangzhou 310002,China)

机构地区：[1]中国电子科技集团公司第28研究所 [2]江西省电力公司经济技术研究院 [3]中国人民解放军73228部队

出　　处：《信息化研究》2018年第5期44-50,共7页INFORMATIZATION RESEARCH

摘　　要：文章根据目前海洋微生物探测的需求,提出了一种实时智能微生物识别技术。该技术运用于拖曳式水下显微镜,可用来探测海洋中微生物的浓度。识别过程主要分为三个步骤,第一步获取背景图片,选取多张不含微生物目标的图片,做平均处理后作为背景图片;第二步是确定目标区域,将含有目标的图片与背景图片作差,采用顶帽变换和OTSU等算法进行处理,筛选出含目标并且面积适中的八连通区域;第三步是对目标进行识别,采用支持向量机的方法对筛选出来的含目标的区域进行识别,由此即可判断出一定体积海水中所要探测的微生物的数量,从而计算出浓度。验证结果表明,该方法可以做到实时智能识别,准确率可以达到93.4%。This paper proposes a real-time and intelligent microbial recognition technology to meet the requirements of microbial detection.This technology is applied to polled underwater microscope,and can be used to measure the concentration of microbe in the ocean.Three steps are included.First step,several pictures that don't have microbe targets are picked and averaged to be the background pictures.Second step is to select target areas,subtract the pictures that have targets with the background pictures,and then use top-hat transformation and OTSU algorithm to get 8-adjacent connection areas which have the targets and moderate area.The third step is to recognize targets,the target areas are recognized by support vector machine.So that the number of microbe targets in the ocean can be counted to calculate the concentration.Experimental results show that this technology realized real-time and intelligent recognition,and the accuracy rate is as high as 93.4%.

关 键 词：微生物识别 顶帽变换 OTSU 支持向量机 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]