重庆教育学院学报
重慶教育學院學報
중경교육학원학보
JOURNAL OF CHONGQING COLLEGE OF EDUCATION
2012年
6期
27-30
,共4页
自适应ISODATA聚类%色差分析%烟草异物%在线识别
自適應ISODATA聚類%色差分析%煙草異物%在線識彆
자괄응ISODATA취류%색차분석%연초이물%재선식별
self-adaptive ISODATA clustering%color aberration analysis%the foreign in tobacco%real-time recognition
提出一种利用聚类和色差分析的烟草异物在线识别算法,采用自适应ISODATA聚类取单元颜色特征,并分析烟叶和异物各分量的分布差异,将单元特征各分量根据直方图分布量化到HIS颜色空间中,从而得到样本学习结果并进行识别。实验结果表明,该算法的平均识别率达到95%,误识率在5%以下。
提齣一種利用聚類和色差分析的煙草異物在線識彆算法,採用自適應ISODATA聚類取單元顏色特徵,併分析煙葉和異物各分量的分佈差異,將單元特徵各分量根據直方圖分佈量化到HIS顏色空間中,從而得到樣本學習結果併進行識彆。實驗結果錶明,該算法的平均識彆率達到95%,誤識率在5%以下。
제출일충이용취류화색차분석적연초이물재선식별산법,채용자괄응ISODATA취류취단원안색특정,병분석연협화이물각분량적분포차이,장단원특정각분량근거직방도분포양화도HIS안색공간중,종이득도양본학습결과병진행식별。실험결과표명,해산법적평균식별솔체도95%,오식솔재5%이하。
This paper presents a real-time method of recognizing the foreign in tobacco based on self-adaptive clustering and color aberration analysis.Through this method,self-adaptive ISODATA clustering is used to obtain the unit color character,the distributional differences of the color components of tobacco and the foreign are analyzed,the color components of unit character are quantized into the HIS color space according to the distribution of histogram so as to gain and recognize the sampling result.The experimental result shows that the recognition rate can be 95% and the mis-recognition rate can be less than 5% by using this method.