南京林业大学学报(自然科学版)
南京林業大學學報(自然科學版)
남경임업대학학보(자연과학판)
JOURNAL OF NANJING FORESTRY UNIVERSITY(NATURAL SCIENCE EDITION)
2015年
3期
18-24
,共7页
白金婷%张晓丽%王书涵%吴石磊
白金婷%張曉麗%王書涵%吳石磊
백금정%장효려%왕서함%오석뢰
多源数据融合%IHS变换%小波变换%IHS小波变换
多源數據融閤%IHS變換%小波變換%IHS小波變換
다원수거융합%IHS변환%소파변환%IHS소파변환
multi-source data fusion%IHS transform%wavelet transform%wavelet transform integrated with IHS
选择北京密云水库及周边地区为实验区,在分析多源数据融合层次和各种融合方法的基础上,针对TM与SPOT5数据运用IHS变换融合法、小波变换融合法以及IHS小波变换融合法分别进行融合实验。在小波变换以及IHS小波变换融合法中,分别采用了二维连续与离散的小波变换方式,研究了小波变换层数以及低频与高频分量系数的选择结果对于融合效果的影响。最后运用主观定性以及客观定量评价方法,经对不同指标的权重计分建立详细的等级评定,对融合结果进行评判。结果表明:在所选小波方法中,二维连续2层小波效果较好,而IHS结合二维连续2层小波方法效果较为突出。
選擇北京密雲水庫及週邊地區為實驗區,在分析多源數據融閤層次和各種融閤方法的基礎上,針對TM與SPOT5數據運用IHS變換融閤法、小波變換融閤法以及IHS小波變換融閤法分彆進行融閤實驗。在小波變換以及IHS小波變換融閤法中,分彆採用瞭二維連續與離散的小波變換方式,研究瞭小波變換層數以及低頻與高頻分量繫數的選擇結果對于融閤效果的影響。最後運用主觀定性以及客觀定量評價方法,經對不同指標的權重計分建立詳細的等級評定,對融閤結果進行評判。結果錶明:在所選小波方法中,二維連續2層小波效果較好,而IHS結閤二維連續2層小波方法效果較為突齣。
선택북경밀운수고급주변지구위실험구,재분석다원수거융합층차화각충융합방법적기출상,침대TM여SPOT5수거운용IHS변환융합법、소파변환융합법이급IHS소파변환융합법분별진행융합실험。재소파변환이급IHS소파변환융합법중,분별채용료이유련속여리산적소파변환방식,연구료소파변환층수이급저빈여고빈분량계수적선택결과대우융합효과적영향。최후운용주관정성이급객관정량평개방법,경대불동지표적권중계분건립상세적등급평정,대융합결과진행평판。결과표명:재소선소파방법중,이유련속2층소파효과교호,이IHS결합이유련속2층소파방법효과교위돌출。
IHS transform can retain the complete information of high spatial resolution, and the wavelet transform fusion imaging has the better spectral information, therefore, the combination of the two transform image fusion can take advan?tage of their respective strengths to provide high?quality images for the subsequent information extraction. Based on the analyses of multi?sources data fusion methods at different levels, Beijing Miyun Reservoir and its surroundings were se?lected as the study area in this research. The data fusion methods including IHS transform, wavelet transform and inte?grated wavelet?IHS transform were firstly used to experiment with Landsat TM multispectral image and SPOT5 panchro?matic image. Furthermore, in the fusion methods of wavelet transform and IHS wavelet transform, the two?dimensional continuous and discrete wavelet transform were used respectively, and the effects of the wavelet transformation layer numbers as well as the selection of the low?frequency coefficients and the high?frequency coefficients on the resulting fused images were discussed. Finally, the fusion results were evaluated by using subjectively qualitative method and ob?jectively quantitative indices, after establishing elaborated ratings to each weight of the index. The results showed that the two?dimensional continuous wavelet method with two layers had better comprehensive effects among the three methods of wavelet, while IHS transform fusion method combined with the two?dimensional continuous wavelet method with two lay?ers could maintain better spectral and spatial resolution, which mean the image quality could be improved significantly by fusing the data of Landsat TM and SPOT5 by using wavelet transform integrated with IHS, providing data security for image classification,information extraction and vegetation monitoring based on the remote sensing technique.