4.2. 1 geographic data processing
Including1:250000,1:100000 topographic maps, digital elevation model (DEM) and raster map (DRG) data processing in paper media. The purpose is to provide geographical elements and control data for the production of remote sensing image map and remote sensing orthophoto map, and to provide digital map for remote sensing geological interpretation and field geological survey.
4.2. 1. 1 making of digital elevation model
DEM data can be purchased directly from the National Basic Geographic Information Center or collected from topographic maps. The method of obtain topographic map is as follows: firstly, scan that topographic map of 1∶ 100000, and refine, vectorize, edit, assign and orient the contour lines in the space coordinates by using the man-machine interactive contour vectorization method; Then, according to the distribution range of interpolation points, interpolation is divided into three categories: global interpolation, block interpolation and point-by-point interpolation. According to certain interpolation methods (such as kriging, etc. ), interpolate contour lines and extract elevation information. Then DEM mosaic and data format conversion are carried out according to the correction unit to generate the mosaic DEM of the whole area; Finally, check whether the stitching accuracy meets the requirements. The method is to check whether there is any error in DEM by generating the DEM light and dark map.
4.2. 1.2 raster map (DRG) making
DRG is a raster data file formed after scanning, geometric correction and color correction of topographic map with scale of 1 ∶ 10000, and its content, geometric accuracy and color are consistent with the original map. The manufacturing method and steps are as follows.
(1) topographic map scanning
Paper topographic maps are scanned according to a certain scan resolution (generally 150 ~ 300 dpi) and stored in TIF image format.
(2) Control points of map sheet generation
Using the standard map information set by the user, the intersection point of the kilometer grid is automatically calculated as the control point. Before generating map control points, you need to set map information and specify internal map contour points. The steps are as follows:
1) Set worksheet information.
A. paper number. The standard sheet number of the map.
B. grid spacing. The grid spacing of the standard sheet should be consistent with the grid spacing of the calibration chart.
C. coordinate system. The coordinate systems used in the map are mainly 54 coordinate system and 80 coordinate system. If geodetic coordinates are selected, geodetic coordinates (unit: m) will be used for the generated standard sheet, otherwise, sheet coordinates will be used.
2) Set control point information for generating map sheets.
A. map coordinates. By selecting the coordinate points of the map frame on the image, the internal map frame points can be located.
B. Minimum interval. When generating control points, the minimum spacing of control points is discarded.
3) Locate the contour points in the internal map.
Determine the positions of four inner contour points on the image. Complete the parameter setting and the input of the point information of the inner contour, automatically calculate the theoretical coordinates of the control points, and calculate the image coordinates of the control points according to the theoretical coordinates.
(3) modify the control points in sequence
The image coordinates of the control points generated by the map sheet are calculated according to the theoretical coordinates of the intersection point of the corresponding kilometer grid, but there is some distortion in the original image. So this value is not necessarily the same as the coordinate value of the intersection point of the kilometer grid on the original picture, and it needs to be corrected.
(4) Grid-by-grid correction
It is necessary to input the image range (that is, the logical coordinate range of the corrected image), the image output resolution and the image contour (that is, the distance extending outward relative to the inner contour, and the unit is consistent with the frame coordinates). By setting the contour distance, the image within a certain distance outside the contour boundary in the map sheet will not be deformed in the image correction process.
(5) Grid map control accuracy requirements
The corrected control point residual is less than1m; Resampling interval1m; The coordinate deviation of contour points, kilometer grids and their intersections should not be greater than1m.
1:100000 DEM grid spacing and elevation error requirements are: the grid spacing of the flat DEM is 50m, and the elevation error is 6m; The DEM grid spacing in hilly area is 50m and the median elevation error is10m; DEM grid spacing in middle and low mountains is 50m, and the median elevation error is10m; The median elevation error of high mountains and extremely high mountains can be relaxed to 1.5 times accordingly.
(6) Accuracy evaluation
The accuracy evaluation of raster map includes the sheet quality evaluation of original map quality evaluation, the quality evaluation of corrected DRG and the check of standard frame fitting.
1) Quality evaluation of original drawings. This item is to evaluate the quality of the original data generated by raster map, which mainly reflects whether the original map has folds and whether it is equal when scanned. If the quality of the original image is not good, the corrected raster map will definitely be affected.
In order to evaluate the quality of the original image, it is necessary to modify the control points in turn. When all the control points are modified, the values in the map quality file reflect the quality of the original map image, and the file parameters are the maximum residual and medium error before image correction. The median error value reflects the overall quality of the original image, and the larger the value, the worse the quality; The maximum residual value reflects the point number and deviation value of the control points with the largest deviation in the original image.
2) Quality evaluation of calibration chart. This item is used to evaluate the quality of DRG data generated by calibration. After the frame-by-frame correction is completed, four inner contour points of the image are located according to the map information and the method of adding inner contour points to the control points generated according to the map, and a quality evaluation file reflecting the image correction is generated. The file parameter is the median error of the corrected image, and the median error value reflects the overall quality of the corrected image. Inspection of outline side length and diagonal dimension (unit: m): check up, down, left, right, diagonal 1, diagonal 2, outline side length and diagonal dimension. By comparing the measured value of the profile side length with the theoretical value, it is checked whether the profile side length and diagonal side length meet the accuracy requirements.
3) Inspection of drawing frame parts. When evaluating the quality of DRG data generated by calibration, we can also use the method of nested comparison between the generated theoretical grid and the kilometer grid on the calibration map to check whether the accuracy of the kilometer grid is within the specified tolerance range. The quality of the corrected DRG data can be judged by checking its nesting.
(7) storage format
The DRG storage format made by ENVI software is *. Tif and *. The DRG storage format made by imgMapGIS system is *.MSI.
(8) Use
Grid map is the basic geographic reference image for making remote sensing image map, generating digital elevation model data and geometric correction.
Image preprocessing
On the premise of maintaining sufficient information and clarity, for images with more noise and fringes, it is necessary to denoise and filter the fringes by methods such as gray value substitution method of adjacent pixels, low-pass filtering method, full-line substitution method, Fourier transform method, etc., and correct the radiation for images with large radiation distortion.
4.2.3 Calibration and Registration
Calibration and registration model selection
Physical and fitting polynomials are often used to correct the model. Calibration and registration shall be carried out in all frequency bands.
Physical model is suitable for image data that can provide strict satellite ephemeris parameters, and requires DEM data at the same time, and controls the distribution of the whole scene; Rational polynomial model is suitable for satellite data with complex attitude and difficult to obtain the external geometric parameters of linear sensors, which requires both DEM data and the overall distribution of control points. Geometric polynomial model is suitable for flat areas, and is usually used to deal with areas where it is difficult to provide satellite ephemeris parameters and DEM data for acquired images. Generally speaking, according to the data source, the physical model is the first choice in areas with large terrain elevation difference, followed by the rational polynomial model using DEM data for orthorectification, and the plain area using1:100000 DRG data and geometric polynomial model for geometric correction.
4.2.3.2 control point selection
Control points should control the periphery of the image and be evenly distributed. The number of control points should be determined according to the calibration model and terrain conditions. The physical model establishes a rigorous model according to the satellite ephemeris parameters, and only 9 control points can be selected, generally more than 20. The model requires the distribution of control points in the whole scene data; The fitting polynomial model is related to its correction order (n). When n= 1, each scene needs at least 7 control points, generally more than 9; When n=2, it is appropriate to select 13 ~ 16 control points for each scene. The model requires that the whole scene data has the distribution of control points.
4.2.3.3 Corrects and Registers Control Point Error Requirements
The errors in leveling terrain correction control points are 1 ~ 1.5 pixels, in hilly terrain correction control points are 1 ~ 1.5 pixels, and in mountainous terrain correction control points are 1.5 ~ 2 pixels, and the maximum residual error of correction control points is not more than twice the average error.
The error in the registration control points of flat terrain is 0.5 ~ 1 pixel, the error in the registration control points of hilly terrain is 0.5 ~ 1 pixel, and the error in the registration control points of mountainous terrain is 1 ~ 1.5 pixel. The maximum residual error of registration control points is not more than 2 times of the median error.
Resampling methods include adjacent element method, bilinear interpolation method and cubic convolution method.
For the resampling of digital orthophoto map (DOM), the resampling interval should be determined according to the drawing scale, and the resampling interval of 1∶250000 scale should be 30m 1: 100000 scale resampling interval15m; DOM boundary difference requires that the boundary difference of flat terrain is 0.8mm, that of hilly terrain is 0.8mm, and that of mountainous terrain is1.2 mm. For linear elements such as roads and rivers, even if the boundary difference meets the above requirements, when the mosaic image appears ghost and blur, boundary processing should be carried out. DOM images should be clear and rich in texture information, and the images between scenes should be as uniform as possible with moderate contrast.
4.2.4 Image Fusion
Image fusion refers to a composite model structure, which integrates remote sensing data of different sensors or information provided by different types of data sources to obtain high-quality image information, eliminates information redundancy between sensors, reduces uncertainty, improves interpretation accuracy and reliability, and forms a relatively complete and consistent information display for the target. Fusion of panchromatic data and multispectral data, correction results of SPOT and TM data, for example, ETM+ (panchromatic) and TM7,4, 1, TM5,4,3, TM5,3,2; SPOT is fused with TM5,3,2 to form a fused image with high-resolution spatial information and multi-spectral color information. There are many fusion methods, such as principal component analysis, weighted multiplication, IHS transform and so on.
The matching accuracy of image fusion can be checked by image fusion method or image superposition method, and the matching accuracy in plain and hilly areas is required to be 0.5 pixels, with the maximum not exceeding 1 pixel; The mountainous area can be appropriately relaxed to 1.5 pixels. Before fusion, the hue of the image must be adjusted to improve the brightness of high-resolution data, enhance local contrast, highlight texture details and reduce noise; Enhance the color of multi-spectral data, broaden the color contrast between different land types and highlight its multi-spectral color information.
Check whether there is ghosting and blurring after fusion. Check the texture details and colors of the image, judge whether the processing before fusion is correct, and return to re-processing if there is any problem. If the brightness of the fused image is low and the gray scale range is narrow, linear stretching and brightness contrast adjustment can be used for processing, and the spectral information and spatial information of the fused data should be preserved as much as possible during the processing.
Image mosaic
The standard map involves multiple scene data or multiple correction partitions, so the indirect edge of the image must be considered. The margin of the edge is 0.8mm mountain area 1.2 mm in both flat and hilly areas.
Digital stitching method is to select the same name point as the stitching control point in the overlapping area of adjacent images, which requires two scenes with the same name to be strictly aligned, and the fitting error is about 1 pixel; Brightness matching is needed between the two images to reduce the brightness difference; The selection of mosaic stitching lines, whether interactive or automatic, must be polylines or curves; In order to further improve the image stitching quality, the weighted average method should be used to smooth the brightness on both sides of the stitching point.
Edge detection can use image superposition method or checkpoint selection method. The edge image is superimposed by image superposition method, and the error is extracted by combining visual interpretation and point measurement. The checkpoint selection method calculates the poor average error by selecting the same name points in the common area of DOM images.
When the trimming error exceeds the specification requirements, the reasons should be analyzed, and the control points of the previous process should be checked and modified; If the edge stitching error meets the requirements, but some features (such as roads and rivers) are misaligned, resulting in ghost and blur in the stitched image, edge stitching correction should be carried out.
Mosaic images should be uniform in tone and moderate in contrast, and obvious blurring or ghosting is not allowed in overlapping areas. In order to ensure natural mosaic, the mosaic image should have an overlap of 10 ~ 50 pixels.
4.2.6 Paper Finishing and Information Management
4.2.6. 1 profile arrangement
Profile decoration includes inner profile, outer profile and coordinate notes, and the requirements are as follows:
1) The outline of the inner map should be a curve, the east-west map can be drawn as a straight line, and the north-south map is an arc, which can be represented as a straight line in sections. The width of the map outline is 1 pixel.
2) The contour line of the map is parallel to the contour line of the inner map, with a spacing of 10mm, the width of the main map 1mm and the width of the auxiliary map 1 pixel, which are parallel to each other with a spacing of 2 mm. ..
3) The content of map contour coordinate annotation is latitude and longitude and kilometer network. Write down the latitude and longitude coordinates at the interval of longitude difference 15' and latitude difference 10' on the external profile, and write down the short line with a length of 2mm and a width of 1 pixel between the main profile and the sub-profile, and the interval of the kilometer network running through the drawing is 10km.
The latitude and longitude notes of the four corners of the contour are marked on both sides of the extension line of the four corners of the inner contour, and the prefix is upward. Longitude notes cross the meridian, with "degree" on the left and "minute" and "second" on the right; Latitude is marked up and down across latitude, with "degree" on it and "minute" and "second" on it.
The annotation of the kilometer network requires that each square line should be marked with two digits (km) of its coordinate value between drawings, and the annotation of the first and last square lines and the 100-kilometer square line should be marked with a complete integer (km). Two kilometers between the north and south drawings should be marked on the right side of the square line, those with 100 or more should be marked on the left side of the square line, and two kilometers between the east and west drawings should be marked above the square line.
The coordinate note is Song Ti. Note that the whole 10km word is 3mm high, and the number with the whole Qian Qian meter word is 2mm high.
Arrangement and Annotation of 4.2.6.2 Drawings
1) Drawing arrangement requires drawing name, drawing joint list, digital scale and line scale, classification, etc.
A. the name of the picture. Use horizontal marking at the outer center of the outline of North Map, with bold font, word height 10mm and word spacing10mm; Pay attention to the page number below the drawing name, the font is bold and the word height is 5 mm
B. scale. Mark the center position outside the south contour. Digital scale and linear scale should be drawn simultaneously.
C. Legend content. Include geographical elements and thematic elements. Generally, it is arranged on the outside of the east drawing, arranged from top to bottom along the outline of the outside drawing, and the upper part is flat with the outline of the north inner drawing.
D. map sheet connection table configuration. It's to the west of the north map.
E. map classification. It is divided into three types: confidential, secret and internal use. The classification is marked in the east outside the north profile, and the last word is aligned with the east profile. The font is in Song style and the height is 5 mm. ..
F. Notes on the outside and west side of the southern map section. Including the type of remote sensing data, time phase and band combination, control data, etc. The font is in Song style and the height is 5 mm. ..
G. oriental notes outside the outline of the southern map. Work unit, font is Song Ti, and word height is 8 mm ..
2) Mark information such as place names, vector elements and thematic elements according to application requirements. The name is noted in Song Dynasty, and the word height is twice that of the line-drawn topographic map.
4.2.6.3 Information Management
DOM images are generated by framing with the standard topographic map of 1∶ 100000 as the unit. On this basis, the contents of map decoration are layered and superimposed to form a DOM information management file. The content and order of each layer are map decoration, annotation, administrative field and DOM respectively.
4.2.7 Inspection and acceptance
1) The image map shall strictly meet the requirements of technical design and task book, and meet the needs of application.
2) The image map requires clear images, moderate contrast, colorless bias, rich information and prominent levels.
3) The outline size, kilometer network, latitude and longitude, interior and exterior decoration and annotation of the map sheet shall meet the requirements.
4) Check the mathematical accuracy: randomly select more than 25 evenly distributed points in each map, and read the coordinates of feature points with the same name as true values on the line-drawn topographic map, digital map or image map with the scale of 1∶ 100000 or above, and calculate the mean error of randomly sampled points.
1∶ 250,000 Remote Sensing Geological Interpretation Technical Guide
Where: m is the point median error, mm; δ x and δ y are the coordinate difference of random sampling points, mm; N is the number of random sampling points.
The maximum residual error of random sampling points is less than 2 times, and the average error is qualified.
Application of 4.2.81∶ 250,000 remote sensing image map
4.2.8. 1 Apply the combined image map with different bands.
The band combination of remote sensing image should be selected according to the application purpose of the image, the ground features in the mapping area and the information content of the image. For TM/ETM+ and ASTER multispectral data, the band combination is required to cover all bands of visible light (B 1, B2, B3), near infrared (B4) and mid infrared (B5, B7). The correlation coefficient between bands is the smallest and the geological information is the richest, which is beneficial to the interpretation of lithology and large structural information. The commonly used band combination is B5. In arid and bare areas, the band combinations of B7, B4 and B 1 are selected; In the vegetation-covered area, the images of low vegetation season in winter are preferred to minimize the influence of vegetation, and the combination of B5, B3 and B2 bands is selected, which is less affected by vegetation, and the image interpretation has better recognition and the geological interpretation effect is the best; Images fused with TM7, TM4, TM 1, TM5, TM4, TM3 and TM5, TM3 and TM2 respectively have good geological interpretation effect.
CEBRS data usually choose the combination of B2, B3 and B4.
Application of Image Maps with Different Data Sources and Scales in 4.2.8.2
1) In order to meet the accuracy requirements of1:250,000 scale remote sensing geological survey, the image map scale should be1:100000.
2) 1∶50000 fusion image is the important remote sensing data of 1∶250000 remote sensing geological survey.
3)TM/ETM+ and ASTER images have multi-level and rich colors and information, which well reflect different geological phenomena. Therefore, TM/ETM+ and ASTER data should be the best data sources for1∶ 250,000 remote sensing geological survey.
4) 4) The fusion image formed by SPOT and TM plays an outstanding role in explaining the ancient volcanic mechanism because of its high resolution and strong stereoscopic impression, but its color tone is not as rich as TM itself and its shadow is too large, which can only play an auxiliary role in lithologic division.
5) The fusion image of radar and TM is not as rich as TM in tone level, and has no greater advantage than radar image in stereo effect and shading. This kind of image is not the first choice for 1∶250000 remote sensing mapping.
6) In terms of data availability, comprehensive application effect and ability to solve geological problems, remote sensing geological interpretation in 1∶250000 remote sensing geological survey should be based on 1∶250000 scale image, supplemented by 1 ∶ 10000 for interactive interpretation to ensure the repeatability of interpretation results.
7) Indoor interpretation should make full use of the advantages of integration of remote sensing orthophoto map and GIS system, and carry out composite processing and interpretation of multi-source data.
8) Orthographic remote sensing image and three-dimensional visual remote sensing image can better highlight the landscape features of topography and geomorphology, and can more intuitively extract structural, lithologic zoning and ecological geological factors, and divide geomorphologic units, so the geological interpretation effect is more prominent.