The Research Of Road Information Extraction Method Based On Knowledge

With the development of scientific technology, the resolution of theremote sensing image keeps improving, high-resolution remote sense has become thefocus of research and been widely used in some departments like agricultural, forestry,water conservancy, environmental protection and urban planning, playing an importantrole in land resource investigation, land-use update, information extraction and theecological environment quality evaluation. Road is the most obvious feature in animage. Road has a close relationship with the development of a region or country, asone of the basic geographic information, to some extent, the development of roadreflects the level of country,which also has a far-reaching influence on the country’seconomic, military strategic. Nowadays the global digital level is getting higher andhigher, China’s digital process has also increasingly accelerated, which makes road`sdigital construction and management become the urgent need,especially the acquiringof basic information of road. High-resolution remote sensing images bring people aclearer perception, however, this also has increased the difficulty of the roadextraction.This article will first do road-extraction research on WorldView-2and IKONOSimages using road knowledge and object-oriented technology, then complete theroad-extraction module redevelope through ArcGIS Engine and IDL, finally, articlewill summarize the method and look forward to the future development.The road-extraction test contains some necessary steps including preprocessing,knowledge discovery, multi-segmentation based on object-oriented technology andbuilding rules.Preprocessing: correction,subset,image fusion, the research of the bestband combination.Knowledge discovery: spectral, texture, other knowledge.Mult-segmentation: combining the texture to explore the best scale for theroad-extraction. Road-extraction: building rules according to various kinds of knowledge, distinguishing the roads and non-road feature.Re-development of the functional modules: using construct the operator interface,using IDL to achieve the feature extraction module.