CN-122024084-A - Remote sensing image space-time fusion method for complex mountain area

CN122024084ACN 122024084 ACN122024084 ACN 122024084ACN-122024084-A

Abstract

The invention relates to the technical field of remote sensing, in particular to a space-time fusion method of remote sensing images facing complex mountain areas, which comprises the steps of S1, collecting Landsat and MODIS time sequence data, topographic data and ground cover product data and carrying out data preprocessing, S2, extracting a slope unit, forming a coupling diagram based on the data collected in the step S1 and dividing the coupling diagram to generate a slope unit, S3, simulating reflectivity change caused by topographic effects by combining a topographic correction model on the basis of introducing the slope unit, constructing different spatial resolution image reflectivity conversion coefficients under complex topography, and finally, calculating according to a weight matrix and the conversion coefficients to generate a fusion image. The invention not only has excellent spectrum precision and space information retention, but also has higher flexibility and robustness when processing complex terrain areas, and overcomes the limitation of ESTARFM in the complex terrain areas.

Inventors

FAN HUI
PAN YINGYING
YANG RONG
ZUO TIANXIANG
WANG HUI

Assignees

云南大学
云南省测绘资料档案馆(云南省基础地理信息中心)

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. A remote sensing image space-time fusion method facing complex mountain areas is characterized by comprising the following steps: S1, collecting Landsat and MODIS time sequence data, topographic data and land covering product data and performing data preprocessing; S2, extracting a slope unit, namely forming a coupling diagram based on the data acquired in the step S1, and dividing the coupling diagram to generate the slope unit; S3, on the basis of introducing the slope unit, simulating reflectivity change caused by a terrain effect by combining a terrain correction model, constructing different spatial resolution image reflectivity conversion coefficients under complex terrain, and finally, calculating and generating a fusion image according to a weight matrix and the conversion coefficients.
2. The method for space-time fusion of remote sensing images in a complicated mountain area according to claim 1, wherein the specific operation of the step S2 comprises the steps of extracting the topographic parameters according to the DEM, generating a topographic parameter coupling map, and then performing image segmentation on the topographic parameter coupling map by adopting simple non-iterative clustering provided by GEE to generate various homogeneous and heterogeneous fragments, namely a slope unit.
3. The space-time fusion method of remote sensing images facing complex mountain areas according to claim 2, wherein the homogeneity inside the slope units is evaluated by a local variance V, and the heterogeneity between the slope units is evaluated by an autocorrelation index I, wherein the calculation formulas of the local variance V and the autocorrelation index I are shown in formula (1) and formula (2): Formula (1) Formula (2) In the formula, For all of the given partitions The first ramp unit The number of the two-dimensional space-saving type, Is the first The surface area of the individual ramp units, Is the first The circular variance of the slope directions of the ramp units, Is a spatial adjacency matrix, Is the slope direction of the ramp unit.
4. The method for spatial-temporal fusion of remote sensing images in a complex mountain area according to claim 2, wherein the optimal cutting parameters for the extraction of the slope units are determined based on the scale of a moving window, and the method comprises randomly extracting sample points in a typical research area according to a minimum sample size formula, generating an inscribed rectangle as an circumscribed circle of the moving window, and calculating the slope units extracted from different superpixel seed position pitches in the circumscribed circle To determine the optimal cutting parameters and extract the ramp unit accordingly, wherein, The calculation formula of (2) is shown as formula (3): Formula (3) In the formula, Is that The upper 5% quantile of (c) is, Is that The lower 5% quantile of (c) is, Is that The upper 5% quantile of (c) is, Is that The lower 5% quantile of (c).
5. The method for space-time fusion of remote sensing images for complex mountain areas according to claim 1, wherein the specific operation step of step S3 comprises: S301, screening similar pixels based on a slope unit, and calculating a weight matrix according to the screened similar pixels; S302, constructing different spatial resolution image reflectivity conversion coefficients under complex terrain by combining with a terrain correction model; S303, calculating and generating a fusion image according to the weight matrix and the conversion coefficient.
6. The method for space-time fusion of remote sensing images in a complex mountain area according to claim 5, wherein the calculation formula of the similar pixel screening is shown in formula (4): Formula (4) In which the window is moved Position pixel and center pixel Is used for the reflection rate difference of (a), As a result of the spectral threshold value, Is a slope unit where the center pixel is located.
7. The method of claim 5, wherein the step S302 further comprises the steps of Time high spatial resolution remote sensing image The calculation formulas of the flat earth surface reflectivity of the high-spatial-resolution remote sensing image at the moment are shown as a formula (5) and a formula (6): formula (5) Formula (6) In the formula, And Is that And Time of day prediction The flat earth surface reflectivity of the remote sensing image with high spatial resolution at any time; Is that The reflectivity of the earth's surface is flattened at all times, Is that The reflectivity of the earth surface is flattened at any moment; when the surface reflectivity changes due to the topographic effect, then The earth surface reflectivity of the time high-spatial resolution remote sensing image is obtained by calculation of a formula (7) and a formula (8): Formula (7) Formula (8) In the formula, Is that Time of day prediction The earth surface reflectivity of the remote sensing image with high spatial resolution at any time, Is that Time of day prediction The earth surface reflectivity of the remote sensing image with high spatial resolution at any time, The method is an empirical parameter obtained by calculation according to the linear relation between the reflectivity in the actual image and the solar altitude; Is that The zenith angle of the sun at any moment, Is a gradient.
8. The method of claim 7, wherein the step S302 further comprises the steps of Time of day and forecast Time high spatial resolution remote sensing image The method comprises the steps of building a regression model by using images with low spatial resolution at moment, and fully utilizing information of adjacent similar pixels to calculate a conversion coefficient, wherein a calculation formula of the conversion coefficient is shown as a formula (9) and a formula (10): Formula (9) Formula (10) In the formula, And Is that The reflectivity of the similar pixels with high spatial resolution at the moment, And Is based on Time of day simulation The reflectivity of the similar pixels with high spatial resolution at the moment, 、、 Is that The temporal low spatial resolution resembles the pixel reflectivity, For the high spatial resolution image similar pixels, the coefficients of the regression model established for the low spatial resolution image similar pixels, 、 Coefficients are converted for the center pixel.
9. The method of claim 7, wherein the step S303 further comprises converting coefficients of reflectivity of images with different spatial resolutions according to the step S302 based on Time of day prediction The calculation formula of the Landsat at the moment is shown as formula (11) and formula (12): formula (11) Formula (12) In the formula, Is that At the moment Landsat the pixel reflectivity, A weight matrix calculated for the screened similar pixels, For a matrix of conversion coefficients that is modeled under complex terrain, Is that To the point of The difference in reflectivity of the MODIS picture elements at the moment in time, Is that To the point of Errors brought by the scs+c model at the moment.
10. The method for space-time fusion of remote sensing images for complex mountain areas according to claim 1, wherein the data preprocessing comprises atmospheric correction, geometric correction, cloud removal and cloud shadow removal.

Description

Remote sensing image space-time fusion method for complex mountain area Technical Field The invention relates to the technical field of remote sensing, in particular to a remote sensing image space-time fusion method for complex mountain areas. Background Long-time-sequence remote sensing imaging is important for monitoring heterogeneous landscape dynamics, especially in the fields of ecosystem monitoring, land utilization/land cover change detection, vegetation/crop monitoring and the like. However, there is always a trade-off between the spatial resolution and the time resolution of the remote sensing image, the spatial resolution of the Landsat series image is 30m, which is often used for long-time-sequence land coverage and vegetation monitoring, but the revisit period of 16 days makes capturing of the dynamic change of the earth surface in a short time difficult, the revisit period of satellite sensors such as MODIS, AVHRR and the like is shorter, the earth can be covered for multiple times in a short time, and the quick change information of the earth surface is captured, but the insufficient spatial resolution of the image is difficult to satisfy the earth surface monitoring of a highly heterogeneous region. Meanwhile, due to the tradeoff between the spatial resolution of the satellite sensor and the revisit period, and the problems of cloud pollution, sensor faults and the like, acquiring the high-space-time resolution remote sensing image becomes very challenging. Space-time fusion based on multi-source remote sensing data is an effective way for solving the problem of difficult acquisition of high space-time resolution remote sensing images. The remote sensing image processing method can combine the remote sensing data acquired by a plurality of sensors on different dates and with different spatial resolutions, thereby generating the remote sensing image of a new time point. In the remote sensing space-time fusion method, a space-time adaptive reflectivity fusion model is particularly classical. In various space-time fusion methods, ESTARFM has higher fusion precision in heterogeneous areas and can realize on-line rapid fusion in a large area range by being applied to a cloud platform. However, ESTARFM has a limitation on the performance of a terrain complex region, the similar pixel screening strategy adopted by ESTARFM is a spectrum thresholding method, the strategy depends on the similarity degree among the spectra of ground objects, and terrain effects easily cause similar pixel mismatching and generate uncertainty. Particularly in mountainous areas with complex topography, the backlight surface and the light-directing surface may cause the phenomenon of 'same object and different spectrum' or 'foreign object and same spectrum', which affects the subsequent operation. Meanwhile, the mountain area accounts for 26.4% of the global land area, and has rich ecological systems, species and genetic diversity, and the mountain area has difficulty in acquiring a high space-time resolution remote sensing image due to complex vertical climate and strong space heterogeneity, so that remote sensing space-time fusion is more needed. Therefore, the improvement ESTARFM of the fusion accuracy in the complex terrain area has important significance, but the prior art rarely involves, and the prime is required to be improved. Disclosure of Invention Features and advantages of the invention will be set forth in part in the description which follows, or may be obvious from the description, or may be learned by practice of the invention. In order to overcome the problems in the prior art, the invention provides a remote sensing image space-time fusion method for a complex mountain area, which specifically comprises the following steps: S1, collecting Landsat and MODIS time sequence data, topographic data and land covering product data and performing data preprocessing; S2, extracting a slope unit, namely forming a coupling diagram based on the data acquired in the step S1, and dividing the coupling diagram to generate the slope unit; S3, on the basis of introducing the slope unit, simulating reflectivity change caused by a terrain effect by combining a terrain correction model, constructing different spatial resolution image reflectivity conversion coefficients under complex terrain, and finally, calculating and generating a fusion image according to a weight matrix and the conversion coefficients. Preferably, the specific operation of the step S2 includes extracting the topographic parameters according to the DEM, generating a topographic parameter coupling map, and performing image segmentation on the topographic parameter coupling map by using the simple non-iterative clustering provided by the GEE, so as to generate various homogeneous and heterogeneous fragments, i.e. ramp units. Preferably, the homogeneity inside the ramp units is evaluated by a local variance V, and the heterogeneity between the ramp units is