The Rapid Extraction Method For Disturbances Caused By Production And Construction Projects Based On GF-1

The distribution of disturbances caused by production and construction projects is important. It provides basic data for administrations to find illegal projects,and it provides evidence for administrations to enforce the law.Traditional methods have the disadvantages of long cycle, poor timeliness and low efficiency, and remote sensing technology can make up for them.After analyzing characterizations of disturbances caused by production and construction projects, disturbances were divided into two types:new land and new impervious surface. Based on the part of Yulin City as the research area, using GF-1 images acquired in 2013 and 2014, through change detection, disturbances were acquired.Then the decision tree classification was used to remove the pseudo disturbance area,and human-computer interaction was used to remove disturbances which were not caused by production and construction projects.Results showed:(1) Although principal component analysis could be used to extract disturbances, to determine the most appropriate method, principal component and threshold were very complex. In addition, only the difference principal component method (the first principal component) could extract two types of disturbances simultaneously, and there were many wrong polygons.(2) Compared with principal component analysis, the pixel-based brightness difference method was simple and could reduce the misclassification phenomenon of new impervious surface, but it couldn’t reduce pseudo disturbances in new bare land.(3) The object-oriented brightness difference method could accurately extract disturbances, effectively remove pseudo disturbances. Its overall classification accuracy was 94.03%, and kappa coefficient was 0.7726.(4) About 2 hours were spent extracting disturbances caused by production and construction projects automatically using computer,and another 2 hours were spent removing wrong polygons and adding correct polygons.However, about 8 hours were spent extracting disturbances through human-computer interaction. Therefore, it could improve the working efficiency by using the computer to automatically extract disturbances caused by production and construction projects.(5) In the research area, the number of disturbances was 158, and their area was 675.50ha. Among them, the number of new bare land was 124, and its area was 594.49ha; the number of new impervious surface was 34, and its area was 81.01ha.