CN113920030A

CN113920030A - Large-area high-fidelity satellite remote sensing image uniform color mosaic processing method and device

Info

Publication number: CN113920030A
Application number: CN202111242549.8A
Authority: CN
Inventors: 李鸿洲; 刘书含; 薛白; 何昭宁; 王艺颖; 王霞; 郭莉
Original assignee: Ministry Of Natural Resources Land Satellite Remote Sensing Application Center
Current assignee: Ministry Of Natural Resources Land Satellite Remote Sensing Application Center
Priority date: 2021-10-25
Filing date: 2021-10-25
Publication date: 2022-01-11
Anticipated expiration: 2041-10-25
Also published as: CN113920030B

Abstract

The invention provides a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method and a device, which comprises the following steps of screening a fusion image, and checking, supplementing and replacing the fusion image; and carrying out mosaic preprocessing on the initial fusion image, carrying out mosaic line editing on the preprocessed image, carrying out balanced editing on the colors of the mosaic lines, carrying out mosaic, adjusting the definition of the image, eliminating atmospheric influence, carrying out refined adjustment on the local surface feature brightness and the colors of the image, carrying out position reduction processing on the image, and outputting to obtain an image product. The method solves the technical problems that in the prior art, due to the fact that imaging conditions, time and atmospheric environment of satellite images are different, color differences among images imaged in different time in the same area are large, and due to the difference among different sensors, the color homogenizing and embedding work of various satellite data sources becomes very complicated, and an effective fixed solution flow are lacked.

Description

Large-area high-fidelity satellite remote sensing image uniform color mosaic processing method and device

Technical Field

The invention relates to the field of surveying and mapping science and technology, in particular to a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method and device.

Background

Due to the difference of imaging conditions, time and atmospheric environment of satellite images, the difference of colors among images imaged in different time in the same area is large, and the difference among different sensors makes the uniform color and mosaic work of various satellite data sources very complicated, and an effective fixed solution and a flow are lacked.

From the aspect of satellite image imaging characteristics, the main problems influencing the satellite image homogenizing and mosaic processing are as follows: the method comprises the following steps of solving the problems of great difference and flaw of images caused by the difference of satellite sensors, the difference of satellite image radiation resolution, the difference of satellite images at different time phases, the problem of cloud and snow areas of satellite images, the problem of area edge connection, the problem of even color of an area with too single ground object type, the problem of satellite image color cast and distortion and the problem of information loss in the image reduction process, and the problems are all caused.

Disclosure of Invention

In view of the above, the present invention provides a large-area high-fidelity satellite remote sensing image color homogenizing and embedding processing method, which solves the technical problems in the prior art that due to the difference between the imaging conditions, time and atmospheric environment of the satellite images, the color difference between the images imaged in different times in the same area is large, and due to the difference between different sensors, the color homogenizing and embedding work of various satellite data sources becomes very complicated, and an effective fixed solution and a flow are lacked.

Specifically, the invention is realized by the following technical scheme:

in a first aspect, a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method is provided, which comprises the following steps:

s1: based on the screened fusion image, the fusion image in the region is checked, supplemented and replaced to obtain an initial fusion image;

s2: performing mosaic preprocessing on the initial fusion image to obtain a preprocessed image;

s3: performing mosaic line editing on the preprocessed image to obtain a mosaic line image;

s4: carrying out balanced editing on the colors of the mosaic lines of the mosaic image to obtain a color image;

s6: inlaying the color image to obtain an inlaying result image;

s7: adjusting the definition of the mosaic result image, eliminating atmospheric influence and obtaining a clear image;

s8: carrying out thinning adjustment on the local ground object brightness of the clear image to obtain an adjusted brightness image;

s9: refining and adjusting the local ground object color of the adjusted brightness image to obtain an adjusted color image;

s10: performing a reduction process on the adjusted color image to obtain a reduced image;

s11: and outputting the reduced image to obtain an image product.

In some embodiments, the method for obtaining the initial fused image by inspecting, supplementing and replacing the fused image in the region based on the screened fused image includes:

screening images by taking a single-scene full-color and multi-spectrum orthorectified true-color image as a starting point, checking the overall data coverage and image quality conditions of the operation area of each screened image and the type of regional ground objects, determining the area needing to be supplemented and replaced with the image, further completing the supplementation and replacement of the screened image, and obtaining a fused image which is a true-color fused image.

In some embodiments, the specific method for performing mosaic preprocessing on the initial fusion image to obtain a preprocessed image includes:

based on true color fusion images, the embedding pretreatment is completed under a GXL platform, specifically:

s21: setting parameters in a GXL platform, defining the effective range of the fusion image, and removing the influence of the image edge 0 value;

s22: and selecting a color homogenizing method and a binding method, and simultaneously selecting an automatic mosaic line setting method of 'minimum square difference' to finish mosaic line editing of each area.

In some embodiments, the method for performing mosaic line editing on the preprocessed image to obtain a mosaic line image specifically includes:

and checking the reasonability of the position of the mosaic line in the preprocessed image, and carrying out manual improvement on the unreasonable position, wherein the manual improvement method is to edit the position of the mosaic line according to the type of ground objects and the terrain conditions to obtain the mosaic line image.

In some embodiments, the color of the mosaic lines of the mosaic image is edited in a balanced manner to obtain a color image, and the specific method includes:

s41: based on a GXL-PCI software platform, the dodging point function of mosaictool software is used for adjusting colors between adjacent images to realize color equalization, a method of two-side change is used for editing the images at two sides, and when the image at one side has cloud or color and brightness deviation, a method of one-side change is adopted;

s42: and when inconsistency occurs during equalization and histogram equalization, adjusting different image colors by manually adding dodging points to achieve a color transition effect and obtain a color image.

In some embodiments, between the step S4 and the step S6, the step S5 is selectively added, specifically:

s5: performing image replacement on a local area of the color image;

s51: and modifying and/or replacing the local area of the color image after color equalization to achieve the integral mosaic effect.

In some embodiments, the color image is mosaiced, and the specific method for obtaining the mosaiced result image is:

s61: carrying out mosaic preprocessing on the color image; the preprocessing comprises the steps of setting the same coordinate system, resolution and spectrum sequence for the fused image, setting the same invalid value and background value in the fused image, and selecting a color homogenizing method and an automatic generation method for mosaic line editing;

s62: submitting the color image subjected to the embedding pretreatment to an image embedding treatment operation, and automatically embedding the orthophoto image to obtain an embedding result image;

in some embodiments, the method of adjusting the definition of the mosaic result image to eliminate atmospheric influence and obtain a clear image includes: using a color gradation tool and a curve tool to adjust the definition and contrast of the mosaic result image, and removing the influence of thin cloud, fog and atmosphere to obtain a clear image;

the method comprises the following steps of carrying out thinning adjustment on the local ground object brightness of the clear image to obtain an adjusted brightness image, and specifically comprises the following steps: adjusting the places with too bright urban areas and too dark mountain areas in the clear image by using a color level tool, a curve tool, a brightness/contrast tool and a shadow/highlight tool to obtain an adjusted brightness image;

refining and adjusting the local ground object color of the adjusted brightness image to obtain an adjusted color image, wherein the specific method comprises the following steps: and adjusting the color cast and distortion of vegetation and ground objects in the adjusted brightness image by using a color level tool, a color balance tool and a hue saturation tool to obtain an adjusted color image.

In some embodiments, the bit reduction processing is performed on the adjusted color image to obtain a bit reduced image, and the specific method includes: and (3) reducing the maximum value range value of the 16bit of the adjusted color image by using a color gradation tool until the maximum value range value is close to the histogram maximum value boundary, retaining all histogram information, and then converting into an 8bit image to obtain a reduced image.

In a second aspect, a large-area high-fidelity satellite remote sensing image uniform color mosaic processing device is provided, which comprises:

image screening unit: based on the screened fusion image, checking, supplementing and replacing the fusion image in the region to obtain an initial fusion image;

a pretreatment unit: performing mosaic preprocessing on the initial fusion image to obtain a preprocessed image;

a damascene line unit: performing mosaic line editing on the preprocessed image to obtain a mosaic line image;

a color image processing unit: carrying out balanced editing on the colors of the mosaic lines of the mosaic image to obtain a color image;

a mosaic processing unit: inlaying the color image to obtain an inlaying result image;

a definition adjustment unit: adjusting the definition of the mosaic result image, eliminating atmospheric influence and obtaining a clear image;

a brightness adjustment unit: refining and adjusting the local ground object brightness of the clear image to obtain an adjusted brightness image;

a color adjustment unit: refining and adjusting the local ground object color of the adjusted brightness image to obtain an adjusted color image;

a bit reduction processing unit: performing a reduction process on the adjusted color image to obtain a reduced image;

an image output unit: and outputting the reduced image to obtain an image product.

The application provides a processing method for uniform color mosaic of large-area high-fidelity satellite remote sensing images, which has the following beneficial effects:

according to the large-area high-fidelity satellite remote sensing image uniform color mosaic processing method, the problems in the satellite image uniform color mosaic processing process are solved through a set of complete data processing flow by means of 2 software platforms. One is GXL-PCI, and provides a method for homogenizing colors of various satellite images, wherein a bundling method calculates gray value information such as an image mean value by calculating a block beam method between images of each overlapped area to adjust the gray value of the images and adjacent images, so that a color balance is achieved, and the effect is good. In addition, the GXL software provides a function of manually increasing 'dodging points', namely dodging points are randomly increased on a mosaic line between adjacent images, local color information of the adjacent images can be adjusted, color equalization is realized, the dodging points can be set to be changed on two sides or one side, and the colors of the images in different seasons can be closer by manually increasing the 'dodging points' in a designated area, so that the optimal mosaic effect is achieved. GXL can well process the problems of color bordering and color homogenizing among different satellite data and images in different seasons, and simultaneously supports the functions of homogenizing and embedding large-area satellite images. The Smart Geofill function can complete the image replacement operation of the whole or local area in real time and quickly, and the adjustment of the mosaic effect is completed.

Drawings

FIG. 1 is a schematic flow chart of a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method;

FIG. 2 is an optimization flow chart of a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method;

FIG. 3 is a logic flow chart of a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method;

FIG. 4 is a diagram illustrating a mosaic line before and after editing according to an embodiment of the present invention;

FIG. 5 is a comparison graph of the results before and after editing the mosaic lines;

FIG. 6 is a comparison of before, during and after rule color balance editing;

FIG. 7 is a comparison of results before and after mosaic line color balance editing;

FIG. 8 is a comparison graph III of the results before, during and after mosaic line color equalization editing;

FIG. 9 is a first diagram illustrating a local area image replacement process;

FIG. 10 is a second schematic diagram illustrating a local area image replacement process;

FIG. 11 is a front-to-back contrast diagram of PS adjusted images;

FIG. 12 is a comparison chart of local feature brightness refinement adjustment;

FIG. 13 is a comparison graph before and after color cast distortion and color matching in an urban area;

FIG. 14 is a comparison graph before and after color cast distortion and color matching of a cultivated land area;

FIG. 15 is a comparison graph before and after color cast distortion and color matching in a river region;

FIG. 16 is a comparison graph of color cast distortion and color modulation in a reservoir region;

FIG. 17 is a comparison graph before and after color cast distortion and color matching of a tailing area;

FIG. 18 is a comparison graph of brightness adjustment before and after using a tone scale tool;

FIG. 19 is an image after using a tone scale tool to enhance brightness;

FIG. 20 is an image of the present invention using a curve tool to improve sharpness and contrast;

FIG. 21 is an image with improved sharpness and contrast using a curve tool;

FIG. 22 is a graph illustrating the reduction of blue haze effect in the atmosphere using a curved tool;

FIG. 23 is an original image;

FIG. 24 is an image after using a tone scale tool to enhance brightness;

FIG. 25 is an image with improved sharpness and contrast using a curve tool;

FIG. 26 is a graph illustrating the reduction of blue fog effect in the atmosphere using a curve tool;

FIG. 27 shows a comparison image before and after adjustment according to an embodiment of the present invention;

FIG. 28 is an image before and after brightness enhancement using a tone scale tool;

FIG. 29 is a schematic representation of an image with blue haze removed using a tone scale tool to improve contrast;

FIG. 30 is an image of a local cloud region (lower right corner haze region) adjusted using a tone scale tool;

FIG. 31 is a diagram of an original image according to an embodiment of the present invention;

FIG. 32 is an image with moderate brightness in town areas after eliminating atmospheric effects according to an embodiment of the present invention;

FIG. 33 is a diagram illustrating an image that continues to adjust to moderate brightness in mountainous areas based on FIG. 32 according to an embodiment of the present invention;

FIG. 34 is a final image of the exposed town area of FIG. 33 replaced with FIG. 32 in accordance with an embodiment of the present invention;

FIG. 35 is an image of a missing roof detail in accordance with an embodiment of the present invention;

FIG. 36 is an image of the roof details restored after adjustment using the shadow highlight tool according to an embodiment of the present invention;

FIG. 37 is a diagram of an image before and after color adjustment according to an embodiment of the present invention;

FIG. 38 is a diagram of a second image before and after color adjustment according to an embodiment of the present invention;

FIG. 39 is a diagram of a third image before and after color adjustment according to an embodiment of the present invention;

FIG. 40 is a diagram of an image before color adjustment according to an embodiment of the present invention;

FIG. 41 is a diagram illustrating an image after adjusting the "Red 2" channel based on FIG. 39 according to an embodiment of the present invention;

FIG. 42 is the final image after adjusting the "cyan" channel based on FIG. 41 according to the embodiment of the present invention.

Detailed Description

It should be noted that although the various steps of the methods of the present disclosure are depicted in the drawings in a particular order, this does not require or imply that these steps must be performed in this particular order, or that all of the depicted steps must be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken down into multiple step executions, etc.

Example 1:

FIG. 1 is a schematic flow chart of a large-area high-fidelity satellite remote sensing image uniform color mosaic processing method according to an embodiment of the present invention; as shown in fig. 1, a large-area high-fidelity satellite remote sensing image color homogenizing mosaic processing method includes:

s6: inlaying the color image to obtain an inlaying result image;

s11: and outputting the reduced image to obtain an image product.

Wherein, in particular,

s1: based on the screened fusion image, the fusion image in the region is checked, supplemented and replaced to obtain an initial fusion image, and the specific method comprises the following steps:

And screening the image by taking the single-scene full-color and multi-spectral ortho-correction true-color fusion image as a starting point. And checking the overall data coverage and image quality condition of the operation area, knowing the type of the ground object in the area, and determining the area needing to supplement and replace the image so as to complete image replacement. This step needs to consider 4 aspects: coverage proportion of effective data in the operation area and area size of the loophole area; coverage proportion of cloud and snow areas; the difference of adjacent images in seasons; the size of the overlapping area of the adjacent images. The coverage proportion of effective data in the operation area and the area size of the loophole area need to meet project operation requirements; in the aspect of coverage proportion of the cloud and snow areas, the cloud amount of a single-scene image is controlled to be less than 20% as much as possible, the cloud amount is too large, and the drawing significance is lost when too much ground shielding objects are covered; in the aspect of seasonal difference of adjacent images, the same season is selected according to priority, adjacent seasons are selected, and finally, the principle of non-adjacent seasons is selected; the size of the overlapping area of the adjacent images is not less than 20% of that of each scene, so that the images in the covered area of the snow clouds can be replaced in the subsequent mosaic line editing step, and the effective covered area is increased.

S2: the specific method for performing mosaic preprocessing on the initial fusion image to obtain a preprocessed image comprises the following steps:

s21: setting parameters in a GXL platform, defining an effective range of an image, and removing the influence of the 0 value of the edge of the image;

Step S2 is to complete the mosaic preprocessing based on the true color fusion image on the GXL platform, and this step generates a mosaic preview image, a mosaic line, an engineering document, and the like. In parameter setting, firstly defining an effective range of an image, and removing the influence of an image edge 0 value; secondly, selecting a color homogenizing method and selecting a binding method, wherein the method can well keep the image color effect of the whole area; the method for automatically embedding the lines selects the minimum square difference, and the minimum square difference method can well bypass the ground objects such as water bodies, buildings and the like because the embedding lines among the images need to be selected in the image overlapping area.

S3: the specific method for editing the mosaic lines of the preprocessed image to obtain the mosaic line image comprises the following steps:

Aiming at unreasonable positions, the integrity of ground object types is guaranteed as much as possible, and manual improvement can be performed along roads, rivers, terrain conditions and the like.

The step S3 is that the mosaic line is an important link in the satellite image stitching process, and the automatic mosaic line setting method of "least square difference" is selected in the previous step, so that the mosaic line editing in most regions can be realized, and the method can bypass the same ground feature, but the automatic mosaic effect in urban regions with relatively complex water areas, large and dispersed river water systems, large types of ground features, and small areas of ground features does not always meet the requirement, the rationality of the mosaic line position needs to be carefully checked, the position of the mosaic line is further refined manually, and the position of the mosaic line is edited according to the original line form of the ground feature.

S4: the specific method for carrying out balanced editing on the colors of the mosaic lines of the mosaic image to obtain the color image comprises the following steps:

one-sided: taking one side as a reference, and adjusting the image of the other side; suitable for newly added images or one side of images with color or brightness deviation

And (3) two sides: the method mainly adopts a double-side editing method, and achieves consistent color and brightness through mutual adjustment.

And step S4, based on the GXL-PCI software platform, adjusting the color between adjacent images by using the evasion point function of the Mosaic tool software to realize color equalization. And judging and setting the image colors at two sides of the dodging point to be bilateral change and unilateral change according to the actual situation, and the step solves the problem of uniform color in a single region of the ground feature type. For example, in desert, water body and other areas, the gray value range in the image is concentrated, inconsistency is easy to occur during homogenization and histogram equalization, and the colors of different images can be adjusted to be closer by manually increasing 'evasion points', so that the optimal color transition effect is achieved.

S6: the specific method for inlaying the color image to obtain an inlaying result image comprises the following steps:

s61: carrying out mosaic preprocessing on the color image; the preprocessing comprises the steps of setting the same coordinate system, resolution and spectrum sequence for the fused image, setting the same invalid value and background value in the fused image, and selecting a bundle color homogenizing method and a mosaic line editing automatic generation method; (ii) a

S62: and submitting the color image subjected to the mosaic preprocessing to an image mosaic processing operation, and automatically mosaicing the orthophoto image to obtain a mosaic result image.

And the step S6 is based on the GXL-PCI software platform and outputs the mosaic result. After the image pre-embedding is finished, the image embedding processing operation can be submitted, and the orthophoto image is automatically embedded; the output XML file of the mosaic preprocessing process is the input of the mosaic generating module; the XML file provides all information (i.e., path of input images, sequencing order, color balance coefficient of each image, mosaic line, etc.) required for generating the final mosaic for the mosaic generation process; the workflow predefines the configuration and sets the path of the input image file and the path of the output folder. The file option output file type setting is convenient for a subsequent PS software platform to open, edit and select the Geotiff (tif); projection line selection is set to output map projection, a resampling type is set to set a bilinear difference value, and because the error caused by rounding of a near interpolation algorithm is overcome, four real pixel values around a virtual point in a source map are fully utilized to jointly determine one pixel value in a target map; the tile option tile specification is set as a unique block, the unique block is selected to be output, the problem of nonuniform and inconsistent color matching of subsequent PS software platform blocks is avoided, the mosaic splicing is avoided again, and the efficiency is improved; the feather width is set to be 10 pixels, which is necessary, and the feather width is worth setting to ensure that the ground objects and the colors of the adjacent images are uniform and excessive in the processes of mosaic line editing and color balance editing, so that the existence of hard connecting edges is avoided.

S7: the specific method for adjusting the definition of the mosaic result image to eliminate atmospheric influence and obtain a clear image comprises the following steps:

and (3) adjusting the definition and contrast of the mosaic result image by using a color level tool and a curve tool, and removing the influence of thin cloud, fog and atmosphere to obtain a clear image.

Step S7 is based on a PS software platform, and the definition and the contrast are adjusted by using a 'color gradation' and 'curve' tool to remove the influence of thin cloud, fog and atmosphere; opening the mosaic result image on the PS software platform, and copying the backup initial image layer, so as to reserve the original record of image color for subsequent searching and comparison.

In some embodiments, the first and second electrodes are, in particular,

s8: the specific method for carrying out thinning adjustment on the local ground object brightness of the clear image to obtain the adjusted brightness image comprises the following steps:

and adjusting the places with too bright urban areas and too dark mountain areas in the clear image by using a color level tool, a curve tool, a brightness/contrast tool and a shadow/highlight tool to obtain an adjusted brightness image.

Step S8 is based on the PS software platform, and uses the "color level", "curve", "brightness/contrast", "shadow/highlight" tools to adjust the situation of too bright urban area and too dark mountain area. In the last step, when the atmospheric influence is removed and the image definition is adjusted, the image brightness is improved under the condition that no exposure is ensured in the urban area, but the mountain area still has a dark condition. That is, if it is ensured that the urban area is not exposed, the brightness of the mountain area is not ideal; if the brightness of the mountainous area is improved to achieve the ideal effect, the urban area can be exposed. Two approaches can be taken to address such situations in general.

S9: the specific method for obtaining the adjusted color image by refining and adjusting the local ground object color of the adjusted brightness image comprises the following steps:

and adjusting the color cast and distortion of vegetation and ground objects in the adjusted brightness image by using a tool of color level, color balance and hue saturation to obtain an adjusted color image.

The step S9 is based on a PS software platform, and the color cast and distortion problems of ground features such as vegetation and the like are solved by using tools of 'color gradation', 'color balance', 'hue saturation'; the color cast and distortion problems of the ground objects are solved by using the adjustment of a color level tool: for example, when the city area of the following graph has a red bias condition, the color gradation channel selects red, the middle slide block position is moved, the output value of the red channel is reduced, the color change is checked, the change of the brightness caused by the change of the output value of the red channel usually, the RGB channel is reselected, the middle slide block position is moved to adjust the output value of the brightness, the situation that the change of the output value is too large and the connection between the periphery is abrupt every time is avoided, and the adjustment is performed between the color bias channel and the RGB channel to achieve the ideal effect.

S10: the specific method for performing the reduction processing on the adjusted color image to obtain the reduced image comprises the following steps:

and (3) reducing the maximum value range value of the 16bit of the adjusted color image by using a color gradation tool until the maximum value range value is close to the maximum value boundary of the histogram, retaining all histogram information, and then converting into an 8bit image to obtain a bit reduction image. Step S10 is based on PS software platform, carries on the bit reduction to the mosaic result, uses the tone scale tool to reduce the maximum value range value to the 16bit mosaic result until it is close to the histogram maximum value boundary, so as to keep all histogram information, finally turns into 8 bit; by this method, the texture and characteristic information of the original image are preserved.

S11: and outputting the reduced image to obtain an image product. And outputting the reduced image to obtain an image product.

Example 2:

according to the method in embodiment 1, in some embodiments, as shown in fig. 2, on the basis of the method in embodiment 1, between the step S4 and the step S6, a step S5 is selectively added, specifically:

s5: performing image replacement on a local area of the color image;

The step S5 is practical for selecting and replacing data of images with multiple overlapping in the area. For the image of the two-degree overlapping area, only the mosaic line editing is needed to be carried out in the overlapping area, but for the image of the multi-degree overlapping area, the selection and editing of the mosaic line are very complicated. Therefore, the method is based on the single scene result and the mosaic result after color equalization, and on the basis, the local area is modified and edited, so that the whole mosaic effect is achieved.

Example 3:

according to a specific embodiment and actual operation, the embodiment provides a large-area high-fidelity satellite remote sensing image color homogenizing mosaic processing method, as shown in fig. 3, which specifically comprises the following steps:

s1: based on the screened fusion image, the fusion image in the region is checked, supplemented and replaced to obtain an initial fusion image:

image screening: and (4) screening the image by taking the single-scene full-color and multi-spectral ortho-correction true-color fusion image as a starting point. And checking the overall data coverage and image quality condition of the operation area, knowing the type of the ground object in the area, and determining the area needing to supplement and replace the image so as to complete the image replacement. This step needs to consider 4 aspects: coverage proportion of effective data in the operation area and area size of the loophole area; coverage proportion of cloud and snow areas; the seasonal difference of adjacent images; the size of the overlapping area of the adjacent images. The coverage proportion of effective data in the operation area and the area size of the loophole area need to meet project operation requirements; in the aspect of coverage proportion of the cloud and snow areas, the cloud amount of a single-scene image is controlled to be less than 20% as much as possible, the cloud amount is too large, and the drawing significance is lost when too much ground shielding objects are covered; in the aspect of seasonal difference of adjacent images, the same season is selected according to priority, then adjacent seasons are selected, and finally the principle of non-adjacent seasons is selected; the size of the overlapping area of the adjacent images is not less than 20% of that of each scene image, so that the images in the cloud and snow coverage area can be replaced in the subsequent mosaic line editing step, and the effective coverage area is increased.

S2: carrying out mosaic preprocessing on the initial fusion image to obtain a preprocessed image:

mosaic pretreatment: and (3) finishing mosaic preprocessing under a GXL platform based on the true color fusion image, wherein a mosaic preview image, a mosaic line, an engineering file and the like can be generated in the step. In parameter setting, firstly defining an effective range of an image, and removing the influence of an image edge 0 value; secondly, selecting a color homogenizing method and selecting a binding method, wherein the method can well keep the image color effect of the whole area; the method for automatically embedding the lines selects the minimum square difference, and the minimum square difference method can well bypass the ground objects such as water bodies, buildings and the like because the embedding lines among the images need to be selected in the image overlapping area.

The method supports images of different satellite data sources and different radiation resolutions to carry out color homogenizing and mosaic processing. The problem of image edge connection when the quantity of satellite images is too large and the data quantity is too large is solved synchronously.

editing a mosaic line: the mosaic line is an important link in the satellite image splicing process, and the mosaic line editing in most regions can be realized by selecting the automatic mosaic line setting method of the minimum square difference in the last step, and the method can bypass the same ground feature, but the automatic mosaic effect of urban regions with relatively complex water areas, large water areas, dispersed rivers, multiple ground feature types and small ground area can not always meet the requirement, the rationality of the mosaic line position needs to be carefully checked, the manual work is further refined and perfected, and the position of the mosaic line is edited according to the original line form of the ground feature. Several cases requiring a mosaic line editing process are listed below:

the mosaic lines of the adjacent water areas in the large area are edited, because the gray values of the water bodies in the images are single and the range of the gray values is not large, but the gray average values in different images have large difference, the color homogenization treatment cannot be performed in a targeted manner, and therefore the mosaic lines need to be edited first. As shown in fig. 4, fig. 4 is a schematic diagram of a mosaic line before and after editing according to an embodiment of the present invention, where the left side is before editing and the right side is after editing.

Fig. 5 is a comparison graph of the results before and after editing the inlaid strand according to the embodiment of the present invention, where the left side is before editing, and the right side is after editing, where the automatic inlaid strand can bypass part of the water body, but for more complicated rivers and water systems, it is necessary to assist manual editing, and the upper graph is a dense cloud reservoir area in north of beijing, where the water area is large and dispersed, and the integrity of the ground features is ensured as much as possible mainly by manual editing.

For the editing of inlaid lines under different terrain conditions, for areas which are adjacent and have obvious boundary lines when being driven to flat ground and mountain land, the difference of ground object types cannot be accurately reflected by the automatic inlaying result, and the areas need to be distinguished through the editing of the inlaid lines.

FIG. 6 is a comparison graph I of the front, middle and rear sides of an inserted line color balance editing according to an embodiment of the present invention, where the left side is before editing, the middle is during editing, and the right side is after editing; the method comprises the steps of editing inlaying lines inside urban areas, wherein buildings or ground objects inside and around the urban areas are regular, the integrity of the same building and ground object type needs to be guaranteed in the inlaying process, generally, linear ground objects such as roads or rivers can be divided, and the inlaying lines need to be edited in the urban areas and the surrounding areas. The lower graph reflects the mosaic line effect of the urban area, and because the urban land has many types and small land area, the automatic mosaic effect can not always meet the requirements and needs to be carefully checked.

Mosaic line editing is a basic key link for ensuring color homogenizing and mosaic processing, and is a precondition for subsequently utilizing various color homogenizing processing methods, namely, the preparation work of mosaic color homogenizing processing images.

color balance editing: the color balance editing of the Mosaic lines is based on a GXL-PCI software platform, and the dodging point function of the Mosaic tool software is used for adjusting the colors between adjacent images to realize color balance. And judging and setting the image colors at two sides of the dodging point to be bilateral change and unilateral change according to the actual situation, and the step solves the color homogenizing problem of the single region of the ground feature type. For example, in desert, water body and other areas, the gray value range in the image is concentrated, inconsistency is easy to occur during homogenization and histogram equalization, and different image colors can be adjusted to be closer by manually increasing 'evasion points', so that the optimal color transition effect is achieved.

The step can well solve the color transition between adjacent images in different time phases.

The two-scene images shown in fig. 7 are GF1B star 2021 year 3 month 25 (left) and GF1C star 2020 year 11 month 3 (right) images. FIG. 9 is a comparison chart of the results before and after mosaic line color balance editing according to the embodiment of the present invention, where the left side is before editing and the right side is after editing;

fig. 8 is a comparison graph three of the results before, during, and after mosaic line color equalization editing in the embodiment of the present invention, where the left side is before editing, the middle is during editing, and the right side is after editing, and the results of the original fused image, the mosaic preprocessing, and the color equalization processing are respectively displayed. In the satellite image, the gray values in the sea area or the water body are concentrated, no obvious characteristic information exists, the color homogenizing method can be rarely and effectively applied to color processing of the sea area and the water body, and due to lack of regularity, even if the difficulty of manual color homogenizing is high, the effect is not ideal. The method is actually arranged in the overlapping area of the sea area and the land area, and avoids points are respectively added to complete the color balance transition.

S5: performing image replacement on a local area of the color image;

local replacement of snow clouds: the function of image replacement for local area of color image has practical operation for data selection and replacement of image with multiple overlapping in area. For the image of the two-degree overlapping area, only the mosaic line editing is needed in the overlapping area, but for the image of the multi-degree overlapping, the selection and editing of the mosaic line are very complicated. Therefore, the method is based on the single scene result and the mosaic result after the color balance, and on the basis, the local area is modified and edited, so that the whole mosaic effect is achieved.

Fig. 9 shows two images of day 5/month 9 in year 2021 of ZY3-03 star and day 3/month 6 in year 2021 of ZY3-03 star, and fig. 9 is a schematic diagram of a local area image replacement process according to an embodiment of the present invention, where the left side is before editing and the right side is after editing; wherein, the image with cloud on the left side of the upper image is the image imaged in 5 months and 9 days. The right image shows the image effect after cloud-free replacement, the imaging time of the image in the replacement area is 3 months and 6 days, the image on the upper image shows the effect of directly replacing the original satellite image by directly using the Smart Geofill function of the PCI software, and the method can quickly realize the image replacement of the local area, but the image color difference after the replacement is very obvious and the visual effect is poor.

Fig. 10 is a result of performing local area replacement based on an image after mosaic color equalization processing, and from the final result, several cloud-replaced areas do not have the problem of color hard edge, so as to achieve the effect of color uniformity requirement, fig. 10 is a schematic diagram of a local area image replacement process according to an embodiment of the present invention, where the left side is before editing, the middle is in editing, and the right side is after editing.

S6: inlaying the color image to obtain an inlaying result image;

embedding the result image: and outputting the mosaic result based on the GXL-PCI software platform. After the image pre-embedding is finished, the image embedding processing operation can be submitted to automatically embed the orthophoto image. The output XML file of the tessellation pre-processing process is the input to the tessellation generation module. The XML file provides all the information needed to generate the final mosaic for the mosaic generation process (i.e., path of input images, ordering order, color balance coefficients for each image, mosaic lines, etc.). The workflow predefines the configuration and sets the path of the input image file and the path of the output folder. The file option output file type setting facilitates the subsequent PS software platform to open edit selection geotiff (tif). The projection line selection is used for setting projection of an output map, the resampling type is used for setting a bilinear difference value, errors caused by rounding of a near interpolation algorithm are overcome, and four real pixel values around a virtual point in a source map are fully utilized to jointly determine one pixel value in a target map. The tile option tile specification is set to be a unique block, the unique block is selected to be output, the problem of nonuniform color mixing of subsequent PS software platform blocks is avoided, the problem of mosaic splicing again is avoided, and the efficiency is improved. The feather width is set to be 10 pixels, which is necessary, and the feather width is worth setting to ensure that the adjacent image ground objects and colors are uniform and excessive in the processes of mosaic line editing and color balance editing, so as to avoid the existence of hard edges.

adjusting the image definition: and adjusting the definition of the mosaic result image to eliminate atmospheric influence, wherein the definition and the contrast are adjusted by using a 'color gradation' and 'curve' tool based on a PS (packet switched) software platform to remove the influence of clouds, fog and atmosphere. Opening the mosaic result image on the PS software platform, and copying the backup initial image layer, so as to reserve the original record of image color for subsequent searching and comparison. Editing the backup layer, adjusting the whole image by using a 'color level' tool to brighten the image, seeing the color level histogram of the image in the input color level frame, adjusting the positions of the histogram to have three positions, respectively distributing three sliders at the front end point, the rear end point and the middle position of the histogram, and changing the output value of the color level histogram by moving the position of the slider. The two end point sliders are typically used as a pair of joint adjustments to adjust image contrast. The sliding block at the middle position is only needed to be adjusted for adjusting the overall brightness of the image, and the brightness is improved by moving the position of the sliding block towards the highest point of the histogram. During adjustment, the region (usually an urban region) with high brightness of the original image is observed to avoid the exposure condition after the brightness is adjusted, the process is observed to be carried out for 2-3 times, namely, after a small amount of adjustment, the color gradation function is determined to be completed by clicking, and after the image is observed to change, the color gradation tool is started again to continuously adjust the position of the middle slide block, so that the process is repeated. Usually, when the middle slider corresponds to the peak position of the histogram, the histogram curve is relatively smooth and symmetrical, and then the color gradation function is completed.

And then, adjusting the whole image by using a curve tool to improve the image contrast. The curve tool is opened and the diagonal line has two intersections with the histogram curve, a1, a 2. Click position a1 generates point a1 ', and click position a2 generates point a 2'. The position of point a 2' in the graph is adjusted to increase its output value. During adjustment, attention is paid to the high-brightness region (usually the urban region) of the original image to avoid exposure after brightness adjustment, and attention is still paid to a method of taking a small amount of times. After a certain effect is achieved, the adjustment can be continued according to the judgment of different image conditions, the output value of A1' is reduced, and the adjustment is carried out for a few times. The image affected by the atmosphere is blue, a blue channel is selected when the whole image is adjusted, the position of the histogram peak value corresponding to the diagonal line is clicked, the output value is reduced, and the image is adjusted for a few times until the ideal effect is achieved. Input and output values may be recorded for data of the same track, and the same input and output values may be used. May be substantially uniform at the joint edges.

After the whole image is adjusted, the image is enlarged, the color mixing area is reduced, whether an area with an unsatisfactory effect and influenced by the thin cloud exists in the local area is checked, and a local adjusting stage is started. For thin cloud regions, the atmospheric influence around the cloud is usually relatively small, and the atmospheric influence in the middle of the cloud is relatively large. When the local adjustment area is drawn, a larger area including the cloud and the surrounding area is selected, and the eclosion value is set to be 10-20 usually. For local adjustment, a color level tool is used, a channel is selected to be blue, an intermediate sliding block is adjusted to reduce an output value, the situation that the image is yellow due to the fact that the blue value is reduced is avoided, the brightness of the local image is affected after the blue value of the color level is changed, the channel is selected to be RGB again, the position of the intermediate sliding block of the histogram is moved to adjust the brightness of the image, and the situation that fog sense is emphasized due to the fact that the brightness is too high is avoided. And after the RGB is adjusted, the local color blue value of the image is influenced, the blue channel is reselected to continue to be adjusted, the adjustment is repeated for a few times, then the sketched peripheral area is observed, the sketched local area is cancelled after the ideal effect is achieved, and the peripheral atmosphere elimination adjustment is completed. The local adjusting area is continuously drawn, the range is reduced compared with the previous range, the adjusted peripheral area is avoided from approaching to the center, the adjustment is continuously carried out due to the fact that the middle is greatly influenced by the atmosphere, the peripheral area around and the middle area are overlapped and gradually adjusted, and the adjusting effect cannot be simultaneously met. And finally, removing the influence of atmospheric thin clouds in local regions according to the concept of from outside to inside.

As shown in fig. 11, which is a comparison graph of the front and the back of the PS adjusted image according to the embodiment of the present invention, the left side is before editing, and the right side is after editing; and adjusting the image definition, and eliminating the atmospheric influence on the front and rear contrast images.

And adjusting the image definition and eliminating atmospheric influence. Based on the PS software platform, the whole image range is firstly toned. Firstly, using a 'color gradation' tool to adjust the brightness of the whole image, secondly using a 'curve' tool to adjust the definition and the contrast of the whole image, and finally using the 'curve' tool to adjust the blue fog effect under the action of atmosphere. And further adjusting the local cloud and fog area after the whole image range is adjusted, removing the blue and fog effect by using a color gradation tool in the local adjustment process, and finally removing the influences of the thin cloud, the fog and the atmosphere from the whole to the local.

S71: the full picture brightness is adjusted using a "tone scale" tool.

Opening the mosaic result image on the PS software platform, backing up the original image layer, generating a backup layer, and editing on the backup layer. The purpose is to reserve the original record of image color for the subsequent search and comparison. And editing the backup layer, and adjusting the whole image by using a color level tool to improve the brightness of the whole image. The image color level histogram can be seen in the input color level frame, three positions of the histogram can be adjusted, three sliders are respectively distributed at the front end point, the rear end point and the middle position of the histogram, and the output numerical value of the color level histogram is changed by moving the positions of the sliders. The two end point sliders are typically used as a pair of joint adjustments to adjust image contrast. The sliding block at the middle position is only needed to be adjusted for adjusting the overall brightness of the image, and the effect of improving the brightness of the whole image is achieved by moving the position of the sliding block in the direction of the highest point of the histogram. During adjustment, attention needs to be paid to control the brightness improvement range in the first brightness improvement process, and an area (usually a town area) with high brightness of an original image is checked to avoid exposure after brightness adjustment. PS software tone scale tool brightness adjustment contrast, and fig. 18 is a contrast chart before and after using the tone scale tool brightness adjustment.

Using a color gradation tool to adjust the whole image, firstly finding an area (usually a town area) with the highest brightness of the original image, and amplifying the area image until the house, the building and the ground object are in the scale size of the image: the method can clearly distinguish the high reflective points on the roof, the texture details of the roof of the building, and the minimum details of the urban and rural roads, the field roads, the forest roads and the like. The images before color mixing can basically see the ground objects, but the whole brightness is still insufficient. In the color gradation adjustment process, the ground object is prevented from being over exposed, and the texture detail features are reserved, so that the ground object can reach the maximum brightness under the condition of no exposure. After the brightness area of the original image is gradually adjusted to the highest brightness in small steps, the texture details can still be preserved in other areas. The fog feeling of the adjusted image vision is aggravated, and the image vision is a normal phenomenon and then is further adjusted in a curve tool.

The use of a tone scale tool to adjust the brightness avoids overexposure. The excessive exposure can cause the image brightness to be too large, namely the texture information is lost, and the too strong light can not clearly see the details of the ground object when being irradiated on the ground object. The toned image with the lost texture information cannot be recovered through other operations, and the detailed information of the ground features can only be recovered through the backed-up original image, so that the process of adjusting the tone scale needs to be carried out for 2-3 times, namely, the tone scale function is determined by clicking after a small amount of adjustment, the tone scale tool is started again after the image is checked and the position of the middle sliding block is continuously adjusted, when the middle sliding block corresponds to the peak position of the histogram, the curve of the histogram conforms to the normal distribution rule and is relatively smooth and symmetrical, so that 2-3 cyclic adjustments are completed, and then the function of improving the brightness of the whole image by using the tone scale is completed. The fog feeling of the adjusted image vision is aggravated, and the image vision is further adjusted in a curve tool for a normal phenomenon.

S72: the definition and contrast of the whole image are adjusted by using a curve tool.

After the brightness is increased by using the color gradation tool to finish the whole image, the fog feeling in the whole image is usually found to exist, and then the curve tool is used. Fig. 19 shows an image with brightness enhanced by using a color gradation tool, and fig. 20 shows an image with sharpness and contrast enhanced by using a curve tool.

The diagonal lines have two intersections with the histogram curve, a1, a 2. Click position a1 generates point a1 ', click position a2 generates point a 2'. The position of point a 2' in the graph is adjusted to increase its output value. During adjustment, the region with high brightness of the original image (usually the urban region) is observed to avoid exposure after brightness adjustment, and when the output value of the curve is adjusted, the phenomenon of one-time adjustment is forbidden, and a small amount of multiple times of adjustment is still adopted to avoid ground object distortion. After a certain effect is achieved, the adjustment can be continued according to the judgment of different image conditions, the output value of A1' is reduced, and the adjustment is performed for a few times. The layering of the visible ground objects is prominent after adjustment, and the contrast is enhanced. The two steps of adjusting the definition and the contrast of the whole image by using a curve tool and adjusting the brightness of the whole image by using a color level tool are usually matched and are circularly performed for a plurality of times. Namely, the color level is used to improve the brightness of the whole image and the curve is used to adjust the contrast of the whole image, so that the image is prevented from being overexposed and losing texture information.

S73: the blue fog effect under atmospheric action was adjusted using a "curve" tool.

The image affected by the atmosphere is blue in whole, channel blue in a curve tool channel is selected when the whole image is adjusted, the intersection position of the peak value of the histogram and the diagonal line is clicked, the output value is reduced, and the image is adjusted for a few times until the ideal effect is achieved, wherein the image is the image with the definition and the contrast improved by using the curve tool; FIG. 22 is an image of the reduction of blue haze effect under atmospheric conditions using a curvilinear tool.

For data of the same track, a color gradation, a curve tool adjustment output value can be recorded, and the same adjustment value, such as a value adjusted by RGB and blue channels respectively, can be used. The uniform color consistency of the same-track image is improved, and the color tone consistency of the joint edge is kept to the maximum extent.

Fig. 23-26 use a "curve" tool to improve the sharpness, contrast map. Fig. 23 is an image with brightness enhanced by using a curve tool, fig. 24 is an image 25 with brightness and contrast enhanced by using a curve tool, fig. 23 is an original image, fig. 24 is an image with brightness enhanced by using a curve tool, fig. 25 is an image with brightness and contrast enhanced by using a curve tool, and fig. 26 is an image with brightness and contrast enhanced by using a curve tool.

S74: local cloud area adjustment

After the whole image is adjusted, the image is enlarged, the color mixing area is reduced, whether an area with an unsatisfactory effect and influenced by the thin cloud exists in the local area is checked, and a local adjusting stage is started. For thin cloud regions, the atmospheric influence around the cloud is usually relatively small, and the atmospheric influence in the middle of the cloud is relatively large. When a local adjustment area is drawn, local connection with the whole area needs to be considered, firstly, all local areas to be adjusted are selected to comprise clouds and areas affected by atmosphere on the periphery, a feather value is set to be 10-20 generally, a color level tool is used for adjusting a middle sliding block to reduce an output value, the situation that the image is yellow due to the fact that the blue value is reduced too much is avoided, blue and yellow are in a pair in color balance, and the reduction of blue in color adjustment can correspondingly increase yellow. FIG. 27 is a comparison image before and after adjustment of the local cloud region.

Local image brightness can be influenced after the color level blue value is changed, the channel is reselected to be RGB, the image brightness is adjusted by moving the position of the middle slide block in the histogram, and the fog sense aggravation condition caused by overhigh brightness is avoided. And after the RGB is adjusted, the local color blue value of the image is influenced, the blue channel is reselected to continue to be adjusted, the adjustment is repeated for a few times, then the drawn peripheral area is observed, the drawn local area is cancelled after the ideal effect is achieved, and the peripheral atmosphere elimination adjustment is completed. The local adjusting area is continuously drawn, the range is reduced compared with the previous range, the adjusted peripheral area is avoided from approaching to the center, the adjustment is continuously needed due to the fact that the middle is greatly influenced by the atmosphere, the adjustment of the peripheral area and the middle area in four weeks is overlapped and is gradual, and the adjusting effect cannot be simultaneously met. And finally, removing the influence of atmospheric thin clouds in local regions according to the concept of from outside to inside.

Fig. 28 shows contrast in which the image sharpness is adjusted and atmospheric influences are eliminated. Fig. 28 shows images before and after brightness enhancement using a tone scale tool, fig. 29 shows an image after blue fog removal with contrast enhancement using a curve tool, and fig. 30 shows an image after adjustment of a local cloud area (bottom right haze area) using a tone scale tool.

and (3) refining and adjusting local feature brightness: and the local ground object brightness of the clear image is refined and adjusted by using a color level, a curve, brightness/contrast and shadow/highlight tool to adjust the conditions of over-bright urban areas and over-dark mountain areas based on a PS (packet switch) software platform. In the last step, when the atmospheric influence is removed and the image definition is adjusted, the image brightness is improved under the condition that no exposure is ensured in the urban area, but the mountain area is still dark normally. That is, if the urban area is ensured not to be exposed, the brightness of the mountain area is not ideal; if the brightness of the mountainous area is improved to achieve the ideal effect, the urban area can be exposed. Two approaches can be taken to address such situations in general.

First, when the brightness of the urban area reaches the ideal effect, the copy is backed up. And continuing curve function adjustment until the brightness of the mountainous area reaches an ideal effect, and then deducting the exposed urban area to replace the urban area with a copy. Deducting the urban area, using a mode of drawing the boundary, adding a feather value, replacing the urban area and the mountain area which are connected with each other by copies, and adjusting the edge-connected brightness of the two images by using a color level function until the images are excessively uniform. This method is applicable in all cases.

FIG. 31-FIG. 34 are partial feature brightness refining contrast adjustment, and FIG. 31 is an original image; FIG. 32 is an image of a town area with moderate brightness after atmospheric effects are eliminated; FIG. 33 is an image based on FIG. 32 with continued adjustment to moderate brightness in mountainous areas, but with exposure (circled) in town areas; FIG. 34 is a final image of the replacement of the exposed town area of FIG. 33 with FIG. 32.

Secondly, for images with small brightness difference between the urban area and the mountain area, the mountain brightness can be directly adjusted to an ideal state by using a curve function, and then the brightness contrast and shadow/highlight function callback are locally applied to the urban area, so that the texture information of the images is enriched. Similarly, the city boundary is sketched and the feather value is added, the brightness contrast function is used, the brightness is reduced, the contrast is improved, and the situation that the one-time output value is changed to be too large and abrupt is avoided. Texture information can be restored for large-area urban roof areas using a "shadow/highlight" function. FIG. 12 is a comparison chart before and after local feature brightness refinement adjustment according to the embodiment of the present invention; the roof texture is restored after the shadow/highlight "function is used.

Fig. 35-36 show the comparison of local feature brightness refinement adjustment, fig. 35 shows the image with missing roof detail, and fig. 36 shows the image with restored roof detail after the adjustment using the shadow highlight tool.

and (3) local ground object color refinement and adjustment: and the local ground object colors of the brightness image after adjustment are finely adjusted, and the color cast and distortion problems of ground objects such as vegetation and the like are solved by adjusting and solving tools of 'color gradation', 'color balance' and 'hue saturation' based on a PS software platform. The color cast and distortion problems of the ground objects are solved by using the adjustment of a color level tool: for example, when the city area of the following graph has a red bias condition, the color gradation channel selects red, the middle slide block position is moved, the output value of the red channel is reduced, the color change is checked, the change of the brightness caused by the change of the output value of the red channel usually, the RGB channel is reselected, the middle slide block position is moved to adjust the output value of the brightness, the situation that the output value is changed too much and the peripheral connection is abrupt every time is avoided, and the back and forth adjustment between the color bias channel and the RGB channel is carried out until the ideal effect is achieved. FIG. 7 is a diagram of the image contrast before and after color adjustment of an embodiment of the present invention

FIG. 13 is a comparison diagram before and after color cast distortion and color matching of an urban area, and the color cast and distortion problems of the ground objects are solved by using a color balance tool for adjustment: the "color balance" command alters the overall color mix of the image. "shading", "midtones" or "highlights" are selected to select the range of tones that are to be heavily modified. "keep luminance" may be selected to prevent the luminance value of the image from changing with the change of color. This option can maintain the tone balance of the image. Dragging the slider to a color to be added in the image; or drag the slider away from the color to be reduced in the image. The values above the color bars show the color change for the red, green and blue channels. Typically the adjustment value will not range from-20 to + 20. FIG. 38 is a second image comparison diagram before and after color adjustment according to an embodiment of the present invention; FIG. 39 is a diagram of a third image comparison before and after color adjustment according to the embodiment of the present invention. FIG. 14 is a comparison graph before and after color cast distortion and color matching of a cultivated land area according to an embodiment of the invention; FIG. 15 is a comparison graph before and after color matching of color cast distortion in a river region according to an embodiment of the present invention; FIG. 16 is a comparison graph of color cast distortion and color modulation in reservoir areas according to an embodiment of the present invention.

The color cast and distortion problems of the ground features are solved by using the adjustment of a 'hue saturation' tool: this function is used for images with intense local color distortion. The color range is modified using a pipette tool or an adjustment slide. Click or slide through the image to select a color range. One of the white triangular sliders is dragged to adjust the color attenuation (feathering adjustment) without affecting the range. The area between the triangle and vertical bar is dragged to adjust the range without affecting the amount of attenuation. The center region is dragged to move the entire adjustment slider (including the triangle and the vertical bar) to select another color region. The range of the color components is adjusted by dragging one of the white vertical bars. The vertical bar is moved outward from the center of the adjustment slider and brought closer to the triangle, thereby increasing the color gamut and reducing attenuation. The vertical bar is moved closer to the center of the adjustment slider and away from the triangle, thereby narrowing the color gamut and increasing attenuation. The color bar is dragged so that the different colors are located in the center of the color bar. FIG. 17 is a comparison graph before and after color matching of color cast distortion in a tailing area according to an embodiment of the invention.

FIGS. 39-42 use hue saturation tools to locally refine feature color to adjust contrast, FIG. 39 is a pre-color adjusted image; FIG. 41 is the image of FIG. 39 after adjusting the "Red 2" channel, excluding the red color cast in FIG. 40; fig. 42 is the final image obtained by adjusting the "cyan" channel based on the image in fig. 41, excluding the cyan color cast condition S10 in fig. 41: performing a reduction process on the adjusted color image to obtain a reduced image;

and (3) bit reduction treatment: and (3) performing bit reduction processing on the adjusted color image, namely performing bit reduction processing on the mosaic result based on a PS (packet switched) software platform, firstly reducing the maximum value range value of the 16-bit mosaic result by using a color gradation tool until the maximum value range value is close to the maximum value boundary of the histogram, so as to retain all histogram information, and finally converting the histogram information into 8 bits. By the method, the texture and characteristic information of the original image is kept.

S11: outputting the reduced image to obtain an image product;

image production:

and (4) storing the result after the steps are finished, and outputting an image mosaic color-homogenizing product, namely outputting the reduced image to obtain an image product.

Example 4:

this embodiment provides a processing apparatus is inlayed to even look of large area high fidelity satellite remote sensing image, includes:

a bit reduction processing unit: performing a reduction process on the adjusted color image to obtain a reduced image; an image output unit: and outputting the reduced image to obtain an image product.

The processing apparatus is inlayed to high-fidelity satellite remote sensing image uniform color of big region that this embodiment provided has following beneficial effect:

(1) the method can effectively solve the influence of the difference of the satellite sensors on the image, and can reduce or eliminate the difference caused by the spectral band ranges received by different satellite sensors, including the difference of the received spectral band ranges of full-color and multi-spectral images.

(2) The influence on the image caused by the difference of the radiation resolution of the satellite image can be effectively solved. The difference of the radiation resolution specifically reflects the difference of the gray value range of the image, and the satellite images with different radiation resolutions are subjected to gray value equalization processing, so that the influence can be effectively eliminated, and the image quality is improved.

(3) The difference of satellite imaging of different time phases can be well eliminated. For example, the imaging difference of satellite images in the same region under different seasons and different weather conditions is large, the four seasons in the north are clear, the difference between vegetation and ground objects is large, the southern fog is more, and the definition and contrast of the images are very obvious. By the method, the problem of color homogenization among images in different time phases can be solved, and the technical difficulty is solved.

(4) The problem of satellite image snow cloud region can be solved. The optical satellite remote sensing image is greatly influenced by weather conditions, and local snow and cloud areas often appear in the image, so that multiple images need to be superposed in the same area to finish the embedding work. Meanwhile, due to the difference of the positions of the cloud and the snow, the multi-scene images in the same area bring much uncertainty and complexity to the subsequent mosaic line editing and color homogenizing work. Therefore, by effectively processing the single-scene image before the mosaic, a lot of post-work can be reduced, and a better color-homogenizing effect can be obtained. Or after the mosaic processing is finished, the local area is subjected to image replacement to achieve the optimal color homogenizing effect.

(5) The problem of regional bordering can be solved. When the quantity of the satellite images is too large and the data volume is too large, partition processing is needed, the partitions are influenced by image ground feature characteristics and operating personnel, the problem of inconsistent color and brightness exists in the edge connecting overlapping area of the partitions, and the problem of area edge connecting can be well solved by mosaic processing and the influence of mosaic results.

(6) The uniform color problem of an excessively single ground object type area can be solved, the area with the single ground object type mainly is an area such as a desert, a water body and the like, the gray value range in the image is concentrated, and the inconsistent situation is easy to occur during uniform color and histogram equalization processing. The brightness of local ground objects of the clear image is finely adjusted to obtain an adjusted brightness image, so that the problems can be effectively solved.

(7) The satellite image color cast and distortion problems can be solved, three spectral bands of blue, green and red are generally used in the process of homogenizing and inlaying the satellite image, and after the data of the three spectral bands are respectively subjected to radiation correction, fusion and color homogenizing treatment, the color cast problem can occur in each spectral band, and the deviation exists between the spectral band and the actual color condition of a ground object. The local ground object color of the adjusted brightness image is refined and adjusted to obtain an adjusted color image, so that the problems can be effectively solved.

(8) The problem of information loss in the image bit reduction process can be solved, and the adjusted color image is subjected to bit reduction processing, so that the bit reduction can be realized and the integrity of data can be maintained.

Claims

1. A large-area high-fidelity satellite remote sensing image uniform color mosaic processing method is characterized by comprising the following steps:

s6: inlaying the color image to obtain an inlaying result image;

s8: refining and adjusting the local ground object brightness of the clear image to obtain an adjusted brightness image;

s11: and outputting the reduced image to obtain an image product.

2. The homogeneous mosaic processing method for the large-area high-fidelity satellite remote sensing image according to claim 1, wherein the specific method for obtaining the initial fusion image by checking, supplementing and replacing the fusion image in the area based on the screened fusion image is as follows:

screening images by taking a single-scene full-color and multi-spectrum orthorectified true-color image as a starting point, checking the overall data coverage and image quality conditions of the operation area of each screened image and the type of regional ground objects, determining the area needing image supplement and replacement, and further completing the supplement and replacement of the screened images to obtain a fused image, wherein the fused image is a true-color fused image.

3. The large-area high-fidelity satellite remote sensing image uniform mosaic processing method according to claim 2, wherein the specific method for carrying out mosaic preprocessing on the initial fusion image to obtain a preprocessed image is as follows:

4. The large-area high-fidelity satellite remote sensing image uniform mosaic processing method according to claim 3, wherein the specific method for mosaic line editing of the preprocessed image to obtain a mosaic line image is as follows:

and checking the reasonability of the position of the mosaic line in the preprocessed image, and carrying out artificial improvement on the unreasonable position, wherein the artificial improvement method is to edit the position of the mosaic line according to the type of ground objects and the terrain conditions to obtain the mosaic line image.

5. The large-area high-fidelity satellite remote sensing image uniform color mosaic processing method according to claim 4, characterized in that the colors of mosaic lines of the mosaic image are edited in a balanced manner to obtain a color image, and the specific method comprises the following steps:

s41: based on a GXL-PCI software platform, the dodging point function of mosaictool software is used for adjusting the color between adjacent images to realize color equalization, a method of changing at two sides is used for editing the images at two sides, and when the image at one side has cloud or color and brightness deviation, a method of changing at one side is adopted;

s42: and when inconsistency occurs during equalization and histogram equalization, adjusting the colors of different images by manually adding evasion points to achieve a color transition effect and obtain a color image.

6. The homomosaic processing method for the large-area high-fidelity satellite remote sensing image according to claim 5, wherein between the step S4 and the step S6, the step S5 is selectively added, specifically:

s5: performing image replacement on a local area of the color image;

s51: and modifying and/or replacing the local area of the color image after color equalization to achieve the whole mosaic effect.

7. The large-area high-fidelity satellite remote sensing image color homogenizing mosaic processing method according to claim 6, wherein the specific method for mosaicing the color image to obtain a mosaic result image is as follows:

8. The homochromy mosaic processing method of the large-area high-fidelity satellite remote sensing image according to claim 7,

adjusting the definition of the mosaic result image, eliminating atmospheric influence and obtaining a clear image, wherein the specific method comprises the following steps: using a color gradation tool and a curve tool to adjust the definition and contrast of the mosaic result image, and removing the influence of thin cloud, fog and atmosphere to obtain a clear image;

9. The large-area high-fidelity satellite remote sensing image color homogenizing mosaic processing method according to claim 8, characterized in that the adjusted color image is subjected to bit reduction processing to obtain a bit reduced image, and the specific method comprises the following steps:

and (3) reducing the maximum value range value of the 16bit of the adjusted color image by using a color gradation tool until the maximum value range value is close to the maximum value boundary of the histogram, retaining all histogram information, and then converting into an 8bit image to obtain a reduced image.

10. The utility model provides a processing apparatus is inlayed to even look of big regional high fidelity satellite remote sensing image which characterized in that, the device includes:

image screening unit: based on the screened fusion image, the fusion image in the region is checked, supplemented and replaced to obtain an initial fusion image;