(1) ensures sufficient geometric accuracy without obvious geometric dislocation.
(2) The color of the whole drawing is uniform, and the colors of similar features are as close as possible, which is rich in information.
(3) The joint is not obvious.
Because the imaging time of different scene images is often different, the change of lighting conditions and ground landscape may cause great color difference between different scene images, so a series of technical measures such as geometric matching and brightness matching should be taken to ensure the mosaic quality.
(1) process flow of digital mosaic (fig. 1-6)
(2) Mosaic method
1. geometric registration
After geometric correction, the geometric distortion of a single scene image is limited to a certain range, but there is still a geometric mismatch for the image to be stitched, which affects the stitching quality. Therefore, it is necessary to register the overlapping areas of two images more accurately on the basis of geometric correction.
Figure 1-6 Digital Mosaic Technology Flowchart
In order to minimize the geometric loss of stitched images, the following two methods can be adopted:
(1) Select several control points in the overlapping area, and geometrically register the mosaic image with the reference map based on the standard image or map.
(2) Select points with the same name on two images for stitching, so that the other image can be geometrically registered with it based on one image (color map 1-7a, 7b). After this processing, the overlapping parts of the two images will be more geometrically consistent.
2. Brightness value matching between images
Because of the great difference in gray distribution between mosaic images of different scenes, even the same object has different brightness values, so it is necessary to match the brightness of each scene image.
Brightness value matching is to adjust the mean and variance of images, so that the gray distribution between images tends to be consistent, thus eliminating the difference of image brightness values. The specific method is: first, find out the overlapping areas of the images, and calculate their respective mean and variance, then select the reference image and transform another image to make the mean and variance consistent with the reference image (color map 1-8a). The calculation formula is as follows:
1∶ 250,000 remote sensing geological mapping method and technology
Where: I' B- the gray value of the transformed B image;
Ib, ia- gray values of images b and a before transformation;
And-the mean and variance of images b and a, respectively.
1∶ 250,000 remote sensing geological mapping method and technology
3. Selection of mosaic splicing points
Generally speaking, the stitching points of stitched images are selected by polyline stitching. In order to minimize the stitching effect of stitching, human-computer interaction is used to find the stitching points on the image, avoiding the parts with obvious brightness difference and eliminating the stitching phenomenon.
In addition, the stitching point can also be automatically identified by computer, mainly following the principle of minimum brightness difference, so as to minimize the brightness difference between the two images at the stitching place. Set in the overlapping area with width n, the brightness values of the scanning lines of the first image are G 1, G2, ..., Gn, and the corresponding brightness values of the second image are G' 1, G'2, ..., G' n ... Take a window with width w in the overlapping area of the images, in the neighborhood of the window with width w.
1∶ 250,000 remote sensing geological mapping method and technology
If Dj is the minimum value, choose J point as the splicing point.
Multi-band images need to be spliced one by one. If the stitching line is selected once for each band, it will inevitably lead to different bands and inconsistent stitching tone after synthesis. Each strip needs to be processed with the same stitching scheme to avoid seams in color composite images.
Step 4 Smooth brightness
After the stitching points of scanning lines in overlapping areas are selected, there may still be brightness differences between adjacent images, and further brightness smoothing is needed. In a neighborhood on both sides of the stitching point, the weighted average of the brightness of the two images is taken as the new brightness value, and the weight coefficient changes linearly in the opposite direction between the two images, and the formula is:
1∶ 250,000 remote sensing geological mapping method and technology
Where: I- the brightness value of the pixel obtained after brightness smoothing;
W—— the width of the smooth brightness area;
I65438+ 0 pixel brightness value of the first image;
I2- the brightness value of the pixel of the second image.
After brightness smoothing, the brightness difference near the stitching point is further eliminated, and the stitching quality of the image is improved (color map 1-8b).
(3) Large-area digital image splicing.
1. production method
Selection of (1) Map Projection Method
The digital image map is made by large-area free framing, and the map projection method generally chooses double-standard parallel isometric cone projection; You can also choose Gaussian-Kruger projection. The calculation of band transformation for different projection bands will bring the loss of accuracy. You can also choose other map projection methods according to different applications and accuracy requirements.
(2) Selection of satellite data
According to different application requirements, satellite data with close season and time and small cloud coverage are selected to avoid too big difference in hue and characteristics. At the same time, we should pay attention to the quality of satellite data and try to choose data without banding and noise.
(3) Pre-processing of satellite images
If there are many stripes and noises in satellite data, they need to be removed. The specific method is the same as above.
(4) Geometric correction of a single scene image
When the geometric correction of image is matched with topographic map, it is necessary to select at least one order of magnitude larger topographic data to ensure the accuracy of image map drawing. When the surveying and mapping scale is1∶ 500,000, the topographic data used for geometric correction are topographic maps or databases of more than1∶ 250,000.
Selection of image resolution: too high resolution will cause too much redundant workload; If it is too low, the drawing accuracy and quality of the image map cannot be guaranteed. Usually the output resolution is 100 ~ 200 dpi.
When terrain data is used for correction control, if the map projection mode of the correction result is different from the pre-selection mode, map projection conversion is needed.
(5) Digital mosaic of large images
When several corrected images of a single scene are digitally spliced and made into regional images, in order to ensure the accuracy to meet the application requirements, the splicing order should be that each column is spliced together first, and then spliced line by line.
(6) Drawing
Mathematical basis of map (scale, coordinate marking, etc.). ), the outline decoration of the map, geographical elements (roads, settlements, rivers, lakes, mountains, etc.). ) and humanistic elements (administrative divisions, etc. ) according to a certain drawing scheme and legend superimposed on the image.
2. Example scope
Dense vegetation area in Daxinganling: longitude:11407' 02 "~12614' 25".
Latitude: 5319 ′ 36 ″ ~ 47 09 ′ 56 ″.
The warp is about12; The weft difference is about 6 10'
Altun Mountain arid exposed area: longitude: 83 ~ 92.
Latitude: 35 30' ~ 40
The warp difference is about 9; The weft difference is about 4 30'
3. Projection mode
Double standard parallel equiangular cone projection is adopted in Daxinganling area, and the parameters are:
1∶ 250,000 remote sensing geological mapping method and technology
Wide-band Gaussian-Kruger projection is used in Altun Mountain area, with the central meridian of 87 (color map 1-9).