When object-based analysis is applied to very high-resolution imagery, pixels within the segments reveal large spectral inhomogeneity; their distribution can be considered complex rather than normal. When normality is violated, the classification methods that rely on the assumption of normally distributed data are not as successful or accurate. It is hard to detect normality violations in small samples. The segmentation process produces segments that vary highly in size; samples can be very big or very small. This paper investigates whether the complexity within the segment can be addressed using multiple random sampling of segment pixels and multiple calculations of similarity measures. In order to analyze the effect sampling has on classification results, statistics and probability value equations of non-parametric two-sample Kolmogorov-Smirnov test and parametric Student’s t
-test are selected as similarity measures in the classification process. The performance of both classifiers was assessed on a WorldView-2 image for four land cover classes (roads, buildings, grass and trees) and compared to two commonly used object-based classifiers—k
-Nearest Neighbor (k
-NN) and Support Vector Machine (SVM). Both proposed classifiers showed a slight improvement in the overall classification accuracies and produced more accurate classification maps when compared to the ground truth image.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited