TM 影像分类算法比较与评价.pdfVIP

Vol No Journal of Heilongjiang Institute of T echnology Feb T M T M KM eansIterative Self Organizing Data A naly sis T echnique ISODA T A M inimum Distance M D Spectral A ngle M apper SA M A rtificial Neural N etw ork A N N Support V ector M achine SV M SV M L M D T M Kappa P A Comparison and evaluation of TM image classification algorithm Y A NG Jie WEI Huaf eng LIU Shiw en College of Geoscience and Surveying Engineering China U niversity of M ining T echnology Beijing Beijing China Abstract Based on the T M images of Qian an city the six classification algorithms named as KM eans Iterative Self Organizing Data A naly sis T echnique ISODA T A M inimum Distance M D Spectral A ngle M apper SA M A rtificial Neural Netw ork A N N Support V ector M achine SV M are compared and analyzed comprehensively It is concluded that the accuracy of SV M is the highest and it also cost s much time With the arrival of the era of big data both precision and efficiency of the algorithm are needed but most of evaluation index of current remote sensing image classification cannot take the time efficiency of the algorithms into consideration So an index L is put f orw ard f or classified comprehensive evaluation w hich is applied successf ully to the six kinds of classification algorithm s finally getting the optimal algorithm M D Key words T M image classification comprehensive evaluation index of classification big data Kappa co eff icient KM eans Iterative Self Organizing Data A naly sis T echnique