LAUSR

ABSTRACT Extraction of rural residential areas from high spatial resolution remote sensing images is important for monitoring urban development and assessing disasters. In practice, plant rows in dry lands, field ridges and vegetable greenhouses around rural residential areas are easily confused in the images. To eliminate or reduce these confusions, an edge-suppressed points voting (ESPV) method is proposed. The rationale is that these nonrelated line-segment features have striking differences in perpendicular directional filtering responses. Comparison experiments indicate that ESPV can eliminate a lot of confusions and perform well in rural residential area extraction in accuracy aspect.