The Impact of the Type and Spatial Resolution of a Source Image on the Effectiveness of Texture Analysis

This paper presents a study on the effectiveness of texture analysis remote sensing imagery depending type and spatial resolution source image. The used following image types: near-infrared band, red first principal component, second component normalized difference vegetation index images pixel size from 2 m to 30 m, generated multispectral WorldView-2 evaluated separability selected pairs land cover classes: bare soil, low vegetation, coniferous forest, deciduous water reservoirs, built-up areas. tool for was granulometric based morphological operations—one less popular methods which, however, as demonstrated by previous studies, shows high in separating classes different texture. conducted enabled researchers evaluate significance visibility possibility using differentiate between classes. obtained results showed that there is no single, universal combination conditions analysis, which would be best point view all For most analyzed classes, were highest (2–3 m), but class areas stands out this comparison—the distinction with average (10–15 m). Research has also shown single universally basis analysis. While majority best, it channel.