Research On The Field Information Extraction By RS On Various Scales

According to the demands of field graphics management in actualizing precisionagriculture, this research aimed at the frequent changes of field parcel information dueto constructing new countryside and land arrangement, put forward different methodsusing remote sensing to adjust and modify different scale acquirement of field parcelinformation. On accuracy demand, it proposed two concepts of village-town-level andfield-parcel-level. To adjust and modify the village-town-level field parcelinformation, the SPOT 5 images were adopted. And UAV remote sensing technologywere used to adjust and modify precise field-parcel-level information.The main results of this research were summarized as follows:(1) A method of processing SPOT 5 images which gave prominence to theboundary of field parcel was proposed. Panchromatic band SPOT5 images whichwere after high pass filters, and fusion images based on multiplicative method werecombined to do visual interpretation. The spatial matching precision of the points areas follows: 65.68% error on X axes, 85.69% error on Y axes, and 48.52% error ofpoint were lower than 8 m; the median error on X, Y and point was respectively7.91m,6.08 m, and 9.93m; 85% of the angular error of line sample was within 2°, and60% within 1°; 40% of the error distance at middle point was lower than 6m; 65% ofthe error of line length was within 5 m, 30% within 3 m, and 15% within 1 m; and76.67% of field batches obtained acceptable area precision (over 85%); and acomparative area precision of 93.33% field batches was more than 85%. The aboveanalysis demonstrated that the geometric correction methods used in this researchcould achieve reasonable geometric precision in extracting village-town-level fieldparcel information.(2) A method of geometric correction of UAV remote sensing images wasproposed. Before rectifying the UAV remote sensing images, the images existedobvious aberration. In an image, the farther a point away from the central point, the notable the aberration occurred. After a cubic convolution correction, the precision ofpoint, line and polygon came to 92.45%, 94.13%, 96.41%. They could well satisfy theprecision demand of acquiring village-town-level field parcel information.