山东农业大学学报(自然科学版)
山東農業大學學報(自然科學版)
산동농업대학학보(자연과학판)
JOURNAL OF SHANDONG AGRICULTURAL UNIVERSITY(NATURAL SCIENCE)
2015年
4期
600-606
,共7页
黄安%王旭红%杨联安%杜挺%王元元%刘建红
黃安%王旭紅%楊聯安%杜挺%王元元%劉建紅
황안%왕욱홍%양련안%두정%왕원원%류건홍
OLI影像%融合算法%适应性%应用研究%乡镇尺度
OLI影像%融閤算法%適應性%應用研究%鄉鎮呎度
OLI영상%융합산법%괄응성%응용연구%향진척도
OLI images%merging algorithms%suitability%application research%township scale
本文利用OIF因子选择乡镇尺度下Landsat8 OLI影像MS最优波段组合,在此基础上,研究OLI影像MS波段与PAN波段对6种融合算法:Brovey法、PCA法、Daubechies小波变换法、Coifet小波变换法、HIS与小波相结合的变换法、PCA与小波相结合的变换法融合的适应性,并对融合前后影像进行SVM分类,以验证融合结果在实际生产应用中的有效性。结果表明:B456为7波段35种组合方式中最佳波段组合,其OIF值为27.842;对融合前后影像进行定性和定量精度评价,OLI影像对PCA算法融合适应性最强,各精度指标均占优;Daubechies小波算法光谱扭曲度最小;HIS-wavelet算法清晰度最高;PCA-wavelet算法相关系数最高,融合结果信息含量最大;适应性最差为Brovey算法。土地利用分类精度验证结果表明:OLI影像经PCA算法融合后有助于提高分类精度。
本文利用OIF因子選擇鄉鎮呎度下Landsat8 OLI影像MS最優波段組閤,在此基礎上,研究OLI影像MS波段與PAN波段對6種融閤算法:Brovey法、PCA法、Daubechies小波變換法、Coifet小波變換法、HIS與小波相結閤的變換法、PCA與小波相結閤的變換法融閤的適應性,併對融閤前後影像進行SVM分類,以驗證融閤結果在實際生產應用中的有效性。結果錶明:B456為7波段35種組閤方式中最佳波段組閤,其OIF值為27.842;對融閤前後影像進行定性和定量精度評價,OLI影像對PCA算法融閤適應性最彊,各精度指標均佔優;Daubechies小波算法光譜扭麯度最小;HIS-wavelet算法清晰度最高;PCA-wavelet算法相關繫數最高,融閤結果信息含量最大;適應性最差為Brovey算法。土地利用分類精度驗證結果錶明:OLI影像經PCA算法融閤後有助于提高分類精度。
본문이용OIF인자선택향진척도하Landsat8 OLI영상MS최우파단조합,재차기출상,연구OLI영상MS파단여PAN파단대6충융합산법:Brovey법、PCA법、Daubechies소파변환법、Coifet소파변환법、HIS여소파상결합적변환법、PCA여소파상결합적변환법융합적괄응성,병대융합전후영상진행SVM분류,이험증융합결과재실제생산응용중적유효성。결과표명:B456위7파단35충조합방식중최가파단조합,기OIF치위27.842;대융합전후영상진행정성화정량정도평개,OLI영상대PCA산법융합괄응성최강,각정도지표균점우;Daubechies소파산법광보뉴곡도최소;HIS-wavelet산법청석도최고;PCA-wavelet산법상관계수최고,융합결과신식함량최대;괄응성최차위Brovey산법。토지이용분류정도험증결과표명:OLI영상경PCA산법융합후유조우제고분류정도。
In this study, we used the OIF factor to choose the best MS band combination for Landsat8 OLI image at the township scale, aiming to study the suitability of 6 kinds of fusion algorithms including standard color variation method (Brovey method), the principal component transformation method (PCA method), Daubechies transformation method for wavelet, Coifet wavelet transformation method, transformation method combining HIS with wavelet and PCA combining with wavelet transformation method for merging of MS and PAN brands of OLI images, and classified the image before and after merging to verify the validity of the fusion results in actual production application with SVM method. Results showed that the B456 brand which OIF value was 27.842 was the best band combination among 35 kinds of combinations of 7 bands. Qualitative and quantitative accuracy assessment before and after merging image showed that each index of OLI image was the dominated for PCA algorithm, which had the highest merging adaptation. And the spectral distortion degree of Daubechies wavelet algorithm was the smallest; HIS-wavelet algorithm had the highest sharpness; PCA-wavelet algorithm had the highest correlation coefficient and merging information contents compared with others. Brovey was the worst adaptive algorithm among 6 kinds of fusion algorithms;Accuracy verification of land use classification demonstrated that OLI image which merged by PCA algorithm would contribute to improve the classification accuracy.