High Resolution Remote Sensing Image Data In The Application Of The National Geographic Conditions Of Census

As national geographic conditions of census work in full swing for the first time, the amount of data and wide space of problems brought difficulties to the traditional field of surveying and mapping survey method, using the identification of high resolution remote sensing digital image data of geographic information census data generalization is a good way to solve this disadvantage. Because the high resolution remote sensing image has a macroscopic and timeliness, geographical conditions, using high resolution remote sensing image as a survey of the major data sources, greatly reducing the workload of outdoor.Based on the national geographic conditions for the first time this census project as the research background, through the analysis on the characteristics of the geographical conditions monitoring both at home and abroad, summarizes the research status of information extraction in the high resolution image, according to the requirements of the regulations of the state collection and classification, with the aid of e Cognition and ARCGIS software to geography from mentougou district of Beijing national census information extraction method of research, through setting weight, shape, and based on the spectral coefficients of multi-scale segmentation method, to calculate the optimal segmentation scale, and then use the adjacent classification method based on sample, carried out on the classification of the various geographic information. Image to extract the cultivated land, garden land, forest land, grass, houses, roads, structures and artificial pile digging for the classification standard of the eight class information classification module, the eight classes for primary classification standard of the general project. Using the method of artificial interpretation for the second and third level of fine classification, through the adoption of geographical conditions survey field final results to check the accuracy of automatic classification. The results showed that after the automatic classification of vegetation and the vegetation type discriminant accuracy is 93%. The cultivated land, garden land, forest land, meadow vegetation classification accuracy of 56.1%. Refinement of vegetation classification accuracy is not high, the result can’t meet the needs of the geographical conditions census classification, also need to refer to topographic map and other basic data for fine classification of artificial interpretation.Therefore, even though the overall classification accuracy of computer remote sensing interpretation is not high, the classification results can be interpreted as a reference, can quickly achieve accurate adjustment feature category, get more accurate classification result. Way for visual interpretation and automatic interpretation method of combining the advantage of high resolution remote sensing image interpretation is high flexibility, and because joined the artificial interpretation factors, information extraction accuracy is relatively high, can be used as the application of geographic conditions census information extraction method.