Road And Water Information Extraction And Analysis Oriented To Geographical Conditions Monitoring

Geographical conditions as China’s important basic national conditions hasplayed an important role in socio-economic development and urbanization, now itoccurs great changes. Strengthening geographical conditions monitoring has becomenew goals of our surveying career. Through dynamic monitoring and statisticalanalysis for geographical situation elements, it obtains statistical analysis of theresults which reflects geographical situation elements’ spatial distribution andevolution characteristics. Road and Water as the important two geographical situationelements. It is Meaningful to dynamic monitoring the spatial distribution andfrequency of changes of road and water. With the development of remote sensingtechnology, high-resolution remote sensing image provides technical support forgetting data. This paper adopting data source of wanzhou district’s high-resolutionremote sensing image of WordViewⅡin2012, adopting five experimental zone toextract road、city road、rural road、railroad，and carrying out the road and waterextraction based on Object-Oriented，and the road and water extraction supervisedclassification，and the road extraction based on ArcGIS spatial analysis. Finally，Onthe basis of road and water extraction；Carrying out statistical analysis work for roadand water information of wanzhou district based on geographical conditionsmonitoring. Through experimental studying, we get scientific significance of theconclusions as follows：①This paper adopts multiscale segmentation, quadtree segmentation andChessboard segmentation to cut up of water information in experimental zone onebased on eCognition software and uses accuracy omission error and redundancy errorto assess segmentation quality of water information. Studies show that the waterinformation’s accuracy, omission error and redundancy is highest by multi-scalesegmentation in experimental zone one, They are99.66%,0.34%,3.16%.②Previous experiment shows that the Multi-scale segmentation method is aoptimum algorithm to product image object；This paper adopts the multi-scalesegmentation based on minimum heterogeneity to cut up the experimental zone one,two three, four, five for repeat tests and obtains the optimal segmentation for water,road、city road、rural road、railroad is90,80,70,100,100in each experimental zone.③In the optimal segmentation based on quality, this paper adopts the membership functions based on eCognition to extract water information inexperimental zone one and road in experimental zone two and to assess classificationaccuracy for experimental zone one and n experimental zone two based on confusionmatrix. In experimental zone one: Experiments show that the overall classificationaccuracy and the overall kappa coefficient is87.21%and0.8198; The water’s useraccuracy, producer accuracy, kappa coefficient is81.25%,81.25%,0.7899. Inexperimental zone two: Experiments show that the overall classification accuracy andthe overall kappa coefficient is84.10%and0.7824; The water’s user accuracy,producer accuracy, kappa coefficient is92.11%,85.37%,0.8410.④Simultaneously，This paper adopting the road and water extraction based onObject-Oriented、 the road and water extraction supervised classification fromExperimental One、Two、Three、Four and Five； and extracting the road based onArcGIS spatial analysis. using Completeness、Correctness、Quality evaluation indexwhich Wiedamann proposed to evaluate extraction Quality and proposing a road andwater extraction method which suitable for road and water information statisticalanalysis.Through the experimental study shows：1) The processing unit of road and water extraction based on Object-Oriented isimage object；Comparing the road and water extraction based on pixel，it uses roadand water’s space、context、semantic effectivelyand so on from sensing image；and itovercome “synonyms spectrum”and “foreign body with spectrum”effectively fromsensing image. and it has the feature of extracting speed、high precision、high degreeof automation and so on.2) When the road space relationship of remote sensing images is complex andthere has other influencing factor（for example：Building、Shadow、Vehicle and soon）； This paper proposes the road extraction ArcGIS spatial analysis； theexperimental shows that it has the feature of extracting speed、high precision、highdegree of automation and so on.⑤On the basis of extracting road and water information， This paper carrys outbasic statistics and comprehensive statistics for road and water information by elevenStreet zoning of WanZhou based on geographical conditions monitoring. Experimentscalculate basic statistics and comprehensive statistics index for road and waterinformation of wanzhou districts in2012and obtain some charts,thematic maps toreflect spatial distribution of road and water information.⑥Finally，On the foundation of basic statistics and comprehensive statistics for road and water information of wanzhou districts in2012; this paper calculates thespeed and intensity of road land expansion, the rate of the Yangtze River areaexpansion, regional economic development index combining related thematic data of2013and obtains the related analysis of the results of wanzhou district from2003to2012. Through analysis and evaluation of road and the Yangtze River, it well revealsthe temporal and spatial variations for road land and he Yangtze River area ofwanzhou districts from2003to2012.