作者: P. M. Atkinson
关键词: Pixel 、 Artificial intelligence 、 Subpixel rendering 、 Grid 、 Scale (map) 、 Pattern recognition 、 Realization (probability) 、 Histogram 、 Computer science 、 Point (geometry) 、 Land cover
摘要: A geostatistical optimization algorithm is proposed for super-resolution land cover classification from remotely sensed imagery. The requires as input, a soft of obtained image. (sub-pixel scale) grid defined. proportions (pixel are then transformed into hard (subpixel by allocating classes randomly to the sub-pixels. number allocated per pixel determined in proportion original pixel. optimizes match between target and current realization two-point histogram swapping sub-pixel within pixels such that class defined maintained. demonstrated two simple simulated images. advantages approach its ability recreate any spatial distribution work with features both large small relative size, combination.