Search

The development of spaceborne remote sensing has greatly facilitated the land cover mapping at various spatial scales. Classification accuracy, however, is usually affected by the heterogeneous spectra of different land cover types for medium–low-spatial-resolution images. The study is aimed at improving the classification accuracy at a city scale by proposing a hierarchical classification method. Time-series Landsat-5 and Landsat-8 Operational Land Imager remote sensing images of 4 years were used as the classified images. A total of six first-class land cover types were determined, namely woodland, grassland, cropland, wetland, artificial surface and others. The object-based image analysis was chosen over pixel-based approaches. More specifically, the nearest-neighbor (NN) classification and SEparability and THresholds (SEaTH) algorithm were combined to produce a hierarchical classification method (NN-SEaTH). SEaTH algorithm was first used to extract the wetland after performing image segmentation in eCognition Developer. Then, the non-wetland was further classified to vegetation and non-vegetation by using a normalized difference vegetation index image. Finally, the other types were then obtained using the NN classification. To validate the proposed method, the NN classifier and NN-SEaTH method were compared. The proposed technique is shown to increase the overall accuracy (OA) and kappa coefficient (k) for the 4 years. The OA and k are, respectively, 96.46% and 0.9231, 96.63% and 0.9269, 96.88% and 0.9394, 95.22% and 0.9239 that are much larger than 88.13% and 0.7503, 88.83% and 0.7660, 88.64% and 0.7630, 87.33% and 0.7371 derived from the NN approach. The study provides a reference for medium-resolution-based land cover mapping by a hierarchical classification. [ABSTRACT FROM AUTHOR]

Several well-developed approaches to inductive learning now exist, but each has specific limitations that are hard to overcome. Multi-strategy learning attempts to tackle this problem by combining multiple methods in one algorithm. This article describes a unification of two widely-used empirical approaches: rule induction and instance-based learning. In the new algorithm, instances are treated as maximally specific rules, and classification is performed using a best-match strategy. Rules are learned by gradually generalizing instances until no improvement in apparent accuracy is obtained. Theoretical analysis shows this approach to be efficient. It is implemented in the RISE 3.1 system. In an extensive empirical study, RISE consistently achieves higher accuracies than state-of-the-art representatives of both its parent approaches (PEBLS and CN2), as well as a decision tree learner (C4.5). Lesion studies show that each of RISE's components is essential to this performance. Most significantly, in 14 of the 30 domains studied, RISE is more accurate than the best of PEBLS and CN2, showing that a significant synergy can be obtained by combining multiple empirical methods. [ABSTRACT FROM AUTHOR]

Several well-developed approaches to inductive learning low exist, but each has specific limitations that are hard to overcome. Multi-strategy learning attempts to tackle this problem combining multiple methods in one algorithm. This article describes a unification of two widely-used empirical approaches: rule induction and instance-based learning. In the new algorithm, instances are treated as maximally specific rules, and classification is oerformed using a best-match strategy. Rules are learned by gradually generalizing instances until no improvement in apparent accuracy is obtained. Theoretical analysis shows this approach to be efficient. It is implemented in the RISE 3.1 system. In an extensive empirical study, RISE consistently achieves higher accuracies than state-of-the-art representatives of both its parent approaches (PEBLS and CN2), as well as a decision tree learner (C4.5). Lesion studies show that eachoof RISE's components is essential to this performance. Most significantly, in 14 of the 30 domains studied, RISE is more accurate than the best of PEBLS and CN2, showing that a significant synergy can be obtained by combining multiple empirical methods. [ABSTRACT FROM AUTHOR]

A conceptually simple way to classify images is to directly compare test-set data and training-set data. The accuracy of this approach is limited by the method of comparison used, and by the extent to which the training-set data cover configuration space. Here we show that this coverage can be substantially increased using coarse-graining (replacing groups of images by their centroids) and stochastic sampling (using distinct sets of centroids in combination). We use the MNIST and Fashion-MNIST data sets to show that a principled coarse-graining algorithm can convert training images into fewer image centroids without loss of accuracy of classification of test-set images by nearest-neighbor classification. Distinct batches of centroids can be used in combination as a means of stochastically sampling configuration space, and can classify test-set data more accurately than can the unaltered training set. On the MNIST and Fashion-MNIST data sets this approach converts nearest-neighbor classification from a mid-ranking- to an upper-ranking member of the set of classical machine-learning techniques. [ABSTRACT FROM AUTHOR]