Wyniki wyszukiwania - BazTech

1

Land cover change detection in northwestern Vietnam using Landsat images and Google Earth Engine

Nguyen Luong B.

Journal of Water and Land Development

|

2020

|

no. 46

162--169

EN

Recently, Google Earth Engine (GEE) provides a new way to effectively classify land cover utilizing available in-built classifiers. However, there have a few studies on the applications of the GEE so far. Therefore, the goal of this study is to explore the capacity of the GEE platform in terms of land cover classification in Dien Bien Province of Vietnam. Land cover classification in the year of 2003 and 2010 were performed using multiple-temporal Landsat images. Two algorithms – GMO Max Entropy and Classification and Regression Tree (CART) integrated into the Google Earth Engine (GEE) platform – were applied for this classification. The results indicated that the CART algorithm performed better in terms of mapping land use. The overall accuracy of this algorithm in the year of 2003 and 2010 were 80.0% and 81.6%, respectively. Significant changes between 2003 and 2010 were found as an increase in barren land and a reduction in forest land. This is likely due to the slash-and-burn agricultural practice of ethnic minorities in the province. Barren land seems to occur more at locations near water sources, reflecting the local people’s unsuitable farming practice. This study may provide use-ful information in land cover change in Dien Bien Province, as well as analysis mechanisms of this change, supporting environmental and natural resource management for the local authorities.

2

Analysis of the suitability of selected image types in a texture analysis of satellite imagery

Kupidura P., Staniak K.

Teledetekcja Środowiska

|

2017

|

T. 57

27--34

EN

The article presents studies on the impact of the source image type on the efficacy of image texture analysis in the terms of distinguishing classes of land use or land cover (LULC). Single gray-scale images are usually the inputs for this type of operation, however their selection is not unambiguous, especially in the case of multispectral images. Two very high resolution satellite images were used in the study: Pleiades (GSD: 2 m) and QuickBird (2.4 m). Five different input images were tested: the original near-infrared and red bands, the images of the first two main components, and the image of the normalised difference vegetation index - NDVI. Five LULC classes were compared to each other: bare soil, low vegetation, deciduous forests, coniferous forests and built-up areas. Granulometric analysis, as the one of the high efficient methods of texture analysis, was used for the test. Research results have shown that the choice of source image for this kind of processing can be very important for the efficacy of distinguishing between different LULC classes. NDVI images, and also the near infrared band and the first principal component were found most useful.

PL

Artykuł przedstawia badania dotyczące wpływu typu obrazu źródłowego na skuteczność analizy teksturowej obrazu z punktu widzenia wyodrębniania klas użytkowania lub pokrycia terenu (LULC). Tego typu operacjom poddawane są zazwyczaj pojedyncze obrazy w skali szarości, jednak ich wybór nie jest jednoznaczny, zwłaszcza w przypadku obrazów wielospektralnych. W badaniach wykorzystano dwa obrazy satelitarne o bardzo wysokiej rozdzielczości: Pleiades (GSD: 2 m) oraz QuickBird (2,4 m). Testowano pięć różnych obrazów wejściowych: oryginalne kanały bliskiej podczerwieni oraz czerwieni, obrazy dwóch pierwszych składowych głównych oraz obraz wskaźnika NDVI. Porównano wzajemnie pięć klas użytkowania lub pokrycia terenu: odkrytą glebę, niską roślinność, lasy liściaste, lasy iglaste oraz tereny zabudowane. Jako narzędzie testów wybrano analizę granulometryczną, jedną z metod analizy teksturowej o wysokiej skuteczności. Wyniki badań pokazały, że wybór obrazu źródłowego do przetworzeń może mieć bardzo duże znaczenie przy rozróżnianiu różnych klas użytkowania lub pokrycia terenu. Największą przydatnością cechowały się obrazy NDVI oraz kanału bliskiej podczerwieni i pierwszej składowej głównej.

3

Testing Texture of VHR Panchromatic Data as a Feature of Land Cover Classification

Lewiński S., Aleksandrowicz S., Banaszkiewicz M.

Acta Geophysica

|

2015

|

Vol. 63, no. 2

547--567

EN

While it is well-known that texture can be used to classify very high resolution (VHR) data, the limits of its applicability have not been unequivocally specified. This study examines whether it is possible to divide satellite images into two classes associated with “low” and “high” texture values in the initial stage of processing VHR images. This approach can be effectively used in object-oriented classification. Based on the panchromatic channel of KOMPSAT-2 images from five areas of Europe, datasets with down-sampled pixel resolutions of 1, 2, 4, 8, and 16 m were prepared. These images were processed using different texture analysis techniques in order to discriminate between basic land cover classes. Results were assessed using the normalized feature space distance expressed by the Jeffries–Matusita distance. The best results were observed for images with the highest resolution processed by the Laplacian filter. Our research shows that a classification approach based on the idea of “low” and “high” textures can be effectively applied to panchromatic data with a resolution of 8 m or higher.

4

Remote Sensing Data Binary Classification Using Boosting with Simple Classifiers

Nowakowski A.

Acta Geophysica

|

2015

|

Vol. 63, no. 5

1447--1462

EN

Boosting is a classification method which has been proven useful in non-satellite image processing while it is still new to satellite remote sensing. It is a meta-algorithm, which builds a strong classifier from many weak ones in iterative way. We adapt the AdaBoost.M1 boosting algorithm in a new land cover classification scenario based on utilization of very simple threshold classifiers employing spectral and contextual information. Thresholds for the classifiers are automatically calculated adaptively to data statistics. The proposed method is employed for the exemplary problem of artificial area identification. Classification of IKONOS multispectral data results in short computational time and overall accuracy of 94.4% comparing to 94.0% obtained by using AdaBoost.M1 with trees and 93.8% achieved using Random Forest. The influence of a manipulation of the final threshold of the strong classifier on classification results is reported.