Effect of pan-sharpening multi-temporal Landsat 8 imagery for crop type differentiation using different classification techniques This study evaluates the potential of pan-sharpening multi-temporal Landsat 8 imagery for the differentiation of crops in a Mediterranean climate. Five Landsat 8 images covering the phenological stages of seven major crops types in the study area (Cape Winelands, South Africa) were acquired. A statistical pansharpening algorithm was used to increase the spatial resolution of the 30m multispectral bands to 15m. The pan-sharpened images and original multispectral bands were used to generate two sets of input features at 30 and 15 metre resolutions respectively. The two sets of spatial variables were separately used as input to decision trees (DTs), k-nearest neighbour (k-NN), support vector machine (SVM), and random forests (RT) machine learning classifiers. The analyses were carried out in both the object-based image analysis (OBIA) and pixel-based image analysis (PBIA) paradigms. For the OBIA experiments, three image segmentation scenarios were tested (good, over and under segmentation). The PBIA experiments were carried out at 30m and 15m resolutions. The results show that pan-sharpening led to dramatic (~15%) improvements in classification accuracies in both the PBIA and OBIA approaches. Compared to the other classifiers, SVM consistently produced superior results. When applied to the pansharpened imagery SVM produced an overall accuracy of nearly 96% using OBIA, while PBIA's overall accuracy was 1.63% lower. We conclude that pansharpening Landsat 8 imagery is highly beneficial for classifying agricultural fields whether an objector pixel-based approach is used.

No abstract

No abstract

This study examined the value of automated and manual feature selection, when applied to machine learning and object-based image analysis (OBIA), for the differentiation of crops in a Mediterranean climate. Five Landsat8 images covering the phenological stages of seven major crops types in the study area (Cape Winelands, South Africa) were acquired and processed. A statistical image fusion technique was used to enhance the spatial resolution of the imagery. The pan-sharpened imagery was used to produce a range of spectral features, textural measures, indices and colour transformations, after which it was segmented using the multi-resolution (MRS) algorithm. The entire set of 205 features (41 per image capture date) was then subjected to different feature selection and reduction methods. The feature selection and reduction methods included manual feature removal (i.e. grouping into semantic themes), filter methods (such as classification and regression trees (CART) and random forest (RF)), and statistical principal components analysis (PCA). The experiments were carried out in two scenarios, namely 1) on all input images in combination; and 2) on each individual image date. The feature subsets were used as input to decision trees (DTs), k-nearest neighbour (k-NN), support vector machine (SVM), and random forest (RF) machine learning classifiers. In order to assess the value of each feature reduction method (comprising feature reduction and selection techniques), overall accuracy, kappa coefficient and McNemar's test were employed to assess classification accuracy and compare the results. The results show that feature selection was able to improve the overall crop identification accuracy for the DT, k-NN, and RF classifiers, but was unable to do so for SVM. SVM scored the highest overall accuracy and kappa coefficient, even without applying feature reduction or selection. Based on these results it was concluded that, although feature selection can aid the crop differentiation process, it is not a necessity.