Based On High Spatial Resolution Remote Sensing Imagesof Forest Resources Investigation Sub-compartment Division

The forest resources planning and design surveys was widely used in the survey system offorest resources in our Country, one of the most important is the sub-compartment division.The positioning accuracy of the spatial distribution of forest, the accuracy of area measurementand the reliability of the final resource, were directly affected by the sub-compartment division.The main method of current forest resources sub-compartment division is adopteddrawing division boundary according to the condition of opposite hillside in the field, whichexistence of human subjectivity and lacks of scientific objectivity in this approach. In thispaper, firstly, we analyzed the characteristics of Spot-5 remote sensing data and theshortcomings of the sub-compartment division arising from small field of vision due tomiscalculation in current forest. Secondly, this paper brings forward the strategy of objectorientedmulti-scale segmentation, which takes different segmentation scales according to thefeature information of different scales. Based on this, the multi-scale segmentation network isestablished and the layered extraction of the feature information is realized. The specificprocess, under the conditions of the control in sub-compartment border for the initialsegmentation of the SPOT-5 images, and output the results of vector, a "object" of thetopological relations (adjacency and connectivity relationships), for the establishment of multiscalesegmentation of data links between the foundation. At this point, identification the resultsof the segmentation block type, and format the initial results of sub-compartment division.Then, extract forest land from the vector of the block type diagram, in a forest patch within thecontrol of the image segmentation and vectorization. Finally, within the forest land bordercontrols of Vector-based results and the preliminary results of superposition of subcompartmentto form the second division results in sub-compartment, if not satisfied with thesecond division results in sub-compartment, using sub-compartment division modules to manually adjust the boundary to form the final sub-compartment border.In addition, the forest resource inventory data base platform production within theindustry system mainly achieved: the document management and view images of basicoperations, image preprocessing, edge extraction, image classification, sub-compartmentdivision, help models. The system provides a good platform for the implementation of subcompartmentdivision.Based this system carry on forest resources sub-compartment division for Wangqingcounty Jingoulin farm, experimental results show that the method achieved fairly good results.This is a new attempt of the automatic identification and sub-compartment division.