Due to the influence of sensor, remote sensing platform (satellite) and the rotation of the earth itself, geometric distortion will inevitably occur in the imaging process of remote sensing images, and it is necessary to ensure the working accuracy through geometric fine correction and registration with other information. According to the working accuracy requirements of this project, the undetermined coefficient method of bivariate quadratic polynomial is adopted, and its mathematical model is:
Study on Remote Sensing Geology of Coal Resources in Yunnan Province
Study on Remote Sensing Geology of Coal Resources in Yunnan Province
Where Ui and Vi are the image coordinates (line number and column number) of the ith point;
And yi are the ground coordinates corresponding to point I;
An, bn, n = 0, 1, 2, … are polynomial coefficients.
In order to reduce the loss of edge information and the reduction of local contrast caused by resampling in geometric correction, firstly, image processing software such as PCI(PCI GEOMATICA) or ENVI (Envi) is used to carry out contrast expansion or frequency filtering pretreatment on each synthetic band of the basic image. Then use 1∶65438+ million topographic map to select GCP (ground control point) evenly, and project the control point from geographical space to image space through coordinate transformation function, so as to carry out geometric fine correction for each band. The accuracy of correction depends on the accuracy, quantity and distribution of the selected ground control points, and the selection of ground control points follows the following principles: evenly distributed in the image; The target in the image is small, with obvious features, easy to identify, and has accurate positioning and identification marks, such as highway turning points, river forks, mountain elevation points, valley intersections and some micro-geomorphological feature points; On the topographic map of 1∶65438+, 5 ~ 10 control points are selected on average. After the ground control points are selected, the correction polynomial is selected for calculation. The more the number of correction polynomials, the higher the accuracy, but at the same time, the greater the deformation away from the control points. In this work, the correction polynomial is selected three times. At the same time, the root mean square error of ground control points is tested, and the results show that the correction error is less than 1 pixel, and the corrected image can fully meet the requirements of surveying and mapping accuracy.
Because the objects of this work are mainly coal-controlling structures, coal-bearing strata and surrounding geological bodies where coal-bearing strata are located, it is necessary to properly strengthen the discrimination of rocks, soil types and related information. TM image has seven bands, and each band has different abilities and uses to distinguish objects. TM3 band is the most favorable spectral region for identifying soil boundary and geological body boundary. TM4 band has a good recognition effect on geological structure boundaries and concealed structural geological bodies. TM7 band is suitable for geological mapping and geological survey. Considering the above factors comprehensively, combined with the characteristics of humid climate and lush vegetation in the study area, repeated experiments were carried out, and TM4, TM7 and TM3 were finally selected as the best synthesis scheme.
Use 1 ∶ 200,000 and1∶ 50,000 topographic maps to evenly select ground control points, and use PCI image processing software to perform geometric fine correction and image digital mosaic processing on ETM ++ images respectively, and the processing error is not greater than1pixel; Secondly, according to the application characteristics of each band of ETM ++ image, the working object and the development characteristics of geological structures in the region, the best band synthesis schemes 7, 4 and 3 which are easy to identify the target are selected for false color synthesis. Then, the ETM+743 composite image is combined with geographical notes and other elements, and finally the ETM+ satellite remote sensing image map (Figure 2-1:200,0001:50,000) that meets the requirements of survey accuracy and quality is produced.
Figure 2- 1 ETM4/7/3 band synthetic image of Zhaotong, Yunnan Province
On the basis of1∶ 50,000 geometric fine correction and image digital mosaic processing for ETM+ and SPOT2 images respectively, the best band combination which is easy for target recognition is selected for false color synthesis. ETM+ image adopts 7, 4 and 3 band synthesis scheme, and SPOT2 adopts panchromatic band. Secondly, two different image data, ETM+743 and SPOT2, are fused (the spatial resolution is 10m). Then, the ETM+743+SPOT 2 fusion image is combined with geographical notes and other elements, and finally the ETM+SPOT 2 satellite remote sensing image map that meets the accuracy and quality requirements of the key coal-bearing area1∶ 50,000 survey work is made.
When stitching images, the image in the center of the research area is selected as the standard image frame, and the subsequent stitching work is based on this image, which is conducive to controlling the accuracy, color, tone and other effects of the overall image. For example, a single image of Mangbang area 132/43 does not need to be spliced; The scene number of the standard image frame selected by Zhongdian District is 132/4 1, and then two images13141are embedded from the center to the periphery. The standard image scene number in Yongsheng District is 13 1/4 1, and then 13 1/42 images are spliced. The remote sensing image number in Xiangyun area is 13 1/42, and the remote sensing image number selected in Zhaotong area is 130/40. During mosaic, the original gray value of pixels outside the overlapping area remains unchanged, and the gray value of pixels with the same name in the overlapping area is weighted and averaged, and the calculation result is taken as the gray value of pixels in the overlapping area. In the process of stitching, even if the tone of the two images is adjusted, the tone at the joint of the two images cannot be exactly the same. Therefore, it is necessary to smooth the hue of the overlapping areas of the images so that no seams will appear in the mosaic image.
In order to make the basic image easy to interpret and compare with the conventional data, this work makes the geographic coordinates, text notes, work information and other contents into a vector format nested with the image map, and finally outputs the image map with the scale of1:200,000 and 1: 1 10,000.