Fference water index (NDWI) [63] to acquire index features. The formulas of NDVI and NDWI are as follows. NDV I =( N IR – Red) ( B – B14 ) = 24 ( N IR Red) ( B24 B14 ) ( Green – N IR) ( B – B23 ) = 7 ( Green N IR) ( B7 B23 )(13)NDW I =(14)exactly where NIR, Red, and Green represent the near-infrared band, red band, and green band, respectively. As shown in Table 3, band 24 and band 14 in the OHS information are chosen for NDVI, whereas band 7 and band 23 are suitable for NDWI. Figure 7 presents the two kinds of OHS hyperspectral index functions. Both the NDVI worth for land and NDWI worth for water are constructive, which can fundamentally represent the spatial Compound 48/80 custom synthesis distribution of land vegetation and water.Remote Sens. 2021, 13,12 ofFigure 7. OHS hyperspectral index attributes in the YRD. (a) NDVI (b) NDWI.two.3.3. synergetic Classification GF-3 polarization and texture characteristics (eight m) and OHS spectral and index features (10 m) derived from the above methods had been applied to carry out synergetic classification. Ahead of classification, the spatial resolution in the two sorts of data should be consistent by means of resampling, which was set to 10 m in this study. Following ortho-rectification and image coregistration, the above characteristics had been classified by way of three classical supervised classification solutions, which includes maximum likelihood (ML) [25], Mahalanobis distance (MD) [26], and assistance vector machine (SVM) [21]. Within this study, to obtain the fusion datasets of GF-3 PolSAR and OHS hyperspectral data for coastal wetland classification, the layer stacking process was utilised to combine 11 GF-3-derived polarization and texture capabilities and seven OHS derived spectral and index attributes into one multiband image in the function level. This new multiband image consists of a total of 18 bands. The classifiers represent 3 distinctive classification principles, as shown below.The ML classifier is amongst the most well-known approaches of classification in remote sensing, in which a pixel with all the maximum likelihood is classified in to the corresponding class. The likelihood Lk is defined because the posterior probability of a pixel belonging to class k. L i = p ( i | x ) = p ( x| i ) p ( i ) = p (x) p ( x| i ) p ( i )i =1 M(15)p ( x| i ) p ( i )where p( i ) and p(x| i ) are the prior probability of class i as well as the conditional probability density function to observe x from class i , respectively. Ordinarily, p( i ) is assumed to be equal, and p(x| i )p( i ) is also widespread to all classes. For that reason, L i is dependent upon the probability density function p(x| i ). The MD classifier can be a direction-sensitive distance classifier that makes use of statistics for every class. It is related towards the ML classifier, however it assumes that all classes have equal covariances, and is, hence, much less time-consuming. The MD of an observation x = (x1 , x2 , x3 , . . . , xn )T from a set of observations with imply = ( , , , . . . , )T and covariance matrix S is defined as [26]: D M ( x ) ==( x – ) T S -1 ( x – )(16)Remote Sens. 2021, 13,13 ofThe SVM classifier is often a supervised classification system that usually yields fantastic classification results from complex and noisy information. It is actually derived from statistical learning theory that separates the classes having a decision surface that maximizes the margin amongst the classes. The surface is often known as the optimal hyperplane, plus the data GLPG-3221 Autophagy points closest towards the hyperplane are referred to as help vectors. When the training information are linearly separable, any hyperplane is usually written as the set of points x sa.