Satellite image generation system and method based on STARFM algorithm

By using the fusion processing of pixel decomposition and STARFM algorithm in the satellite image generation system, the contradiction between satellite platforms in obtaining high-spatial-temporal resolution data is solved, and the generation of high-spatial-temporal resolution images is realized, which improves monitoring capabilities and application prospects.

CN120182092APending Publication Date: 2025-06-20NANCHANG HANGKONG UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510221389.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-02-27
Publication Date
2025-06-20

AI Technical Summary

Technical Problem

Existing satellite platforms face the contradiction between spatial resolution and temporal resolution when acquiring high-spatial-spatial-resolution data. The temporal resolution of high-spatial-spatial-resolution data is insufficient, and the details of low-spatial-resolution data are lacking, making it difficult to meet the needs of high-frequency monitoring and refined management.

Method used

A satellite image generation system based on STARFM algorithm is used to process low-resolution images through pixel decomposition technology, significantly increasing the number of pixels in the image, and combining the STARFM algorithm to fuse the decomposed image with high-resolution images to generate high-temporal and spatial resolution images. The system includes data preprocessing, pixel decomposition, STARFM algorithm fusion and Gaussian smoothing processing modules.

Benefits of technology

The generation of high-temporal and spatial resolution images is realized, the spatial details and time transferability of the image are improved, the problem of spatial and spatial resolution imbalance in the prior art is solved, the monitoring ability of dynamically changing land objects is enhanced, weather interference is reduced, and it has wide application prospects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120182092A_ABST
    Figure CN120182092A_ABST
Patent Text Reader

Abstract

The invention mainly discloses a high-temporal-spatial-resolution satellite image generation system based on an STARFM algorithm. The high-temporal-spatial-resolution satellite image generation system comprises a data preprocessing module, a pixel decomposition module, an STARFM algorithm fusion module and a Gaussian smoothing module. The data preprocessing module is connected with the pixel decomposition module, the pixel decomposition module is connected with the STARFM algorithm fusion module, and the STARFM algorithm fusion module is connected with the Gaussian smoothing module. Compared with the prior art, the method has the advantages that the low-resolution image is processed through the pixel decomposition technology, the number of pixels of the image is remarkably increased, and therefore the spatial resolution is improved. And fusing the decomposed image and the high-resolution image by using an STARFM algorithm. A data preprocessing module and a Gaussian smoothing module are introduced, the accuracy and consistency of input data are ensured, the quality of a fused image is further improved, and noise interference is reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of satellite remote sensing data processing, and particularly to a satellite image generation system and method based on the STARFM algorithm. Background Art

[0002] With the continuous progress and innovation of technology, satellite remote sensing technology, as an important means of modern Earth observation, has been widely applied in multiple fields such as agriculture, environmental monitoring, urban planning, and disaster warning. Through high-resolution optical and radar sensors, it provides real-time and dynamic Earth surface information for global users. In recent years, with the rapid development of low-Earth orbit satellite constellations and the continuous maturity of related technologies, the demand for satellite remote sensing in obtaining high spatio-temporal resolution data has been increasing day by day, becoming one of the key directions to promote the development of the industry.

[0003] However, existing satellite platforms still face many technical bottlenecks when obtaining high spatio-temporal resolution data. On the one hand, high-spatial-resolution satellite images usually have a long revisit period, making it difficult to meet the requirements of high-frequency monitoring; on the other hand, although low-spatial-resolution satellite images can provide a high temporal resolution, they lack sufficient detailed information and cannot meet the requirements of refined monitoring. This contradiction in spatio-temporal resolution limits the application of satellite remote sensing technology in fields such as dynamic change monitoring, precision agriculture, and urban refined management.

[0004] To overcome this contradiction, researchers have developed a variety of spatio-temporal fusion methods. For example, the traditional STARFM algorithm generates high spatio-temporal resolution images by combining the temporal and spatial information of low-resolution and high-resolution images. However, these methods still have deficiencies in processing high-spatial-resolution images, especially when the land cover type changes or the covered area has strong heterogeneity, the accuracy and detail performance of the fusion results are not good. In addition, most existing spatio-temporal fusion technologies rely on specific application scenarios and are difficult to be widely promoted.

[0005] In recent years, some research teams have further improved the spatio-temporal fusion effect through technological innovation. For example, the StarFusion model proposed by the Chen Jin team at Beijing Normal University combines deep learning and traditional spatio-temporal fusion methods, can effectively overcome the dependence of existing algorithms on high-spatial-resolution time series images, and at the same time maintain high fusion accuracy and good spatial details. By integrating the SRGAN (Super-Resolution Generative Adversarial Network) and PLSR (Partial Least Squares Regression) models, this model significantly improves the spatial resolution and time transferability of images.

[0006] Meanwhile, commercial companies are also exploring ways to improve the spatio-temporal resolution of remote sensing images through multi-source data fusion technology. For example, Maxar provides sub-meter spatial resolution through high-resolution optical satellite images, but its temporal resolution is still limited by the satellite's revisit cycle. Planet has achieved global coverage with a daily 5-meter spatial resolution through a constellation of CubeSats, but its high cost limits the widespread application of the data.

[0007] However, there are still some limitations in the existing technologies. For example, optical satellite images are vulnerable to cloud cover and atmospheric conditions, resulting in reduced data availability. In addition, multi-source data fusion technology still faces problems such as inconsistent data formats and low processing efficiency in practical applications. Therefore, developing a satellite image processing system that can efficiently fuse multi-source data and simultaneously improve spatio-temporal resolution has important practical significance and broad application prospects.

[0008] In summary, with the continuous development of satellite remote sensing technology, the demand for high spatio-temporal resolution data is increasing. The existing technologies still have deficiencies in aspects such as the balance of spatio-temporal resolution, data fusion accuracy, and anti-weather interference ability. Therefore, proposing a high spatio-temporal resolution satellite image generation system and method that combines pixel decomposition and the STARFM algorithm is expected to break through the limitations of the existing technologies and provide a new solution for the widespread application of satellite remote sensing technology. Summary of the Invention

[0009] The object of the present invention is to provide a satellite image generation system and method based on the STARFM algorithm, a system and method for generating high spatio-temporal resolution images through pixel decomposition and STARFM algorithm fusion. The core goal is to solve the contradiction faced by existing satellite platforms in obtaining high spatio-temporal resolution data, that is, the temporal resolution of high spatial resolution data is insufficient, and the detailed information of low spatial resolution data is lacking.

[0010] The present invention is realized through the following technical solutions. A high spatio-temporal resolution satellite image generation system based on the STARFM algorithm is divided into a data preprocessing module, a pixel decomposition module, a STARFM algorithm fusion module, and a Gaussian smoothing processing module; the data preprocessing module is connected to the pixel decomposition module, the pixel decomposition module is connected to the STARFM algorithm fusion module, and the STARFM algorithm fusion module is connected to the Gaussian smoothing processing module.

[0011] Further, the data preprocessing module includes geometric correction, radiometric calibration, noise removal, and band selection. By performing geometric correction on the original satellite data, the spatial position of the image is ensured to be accurate. Radiometric calibration is used to calibrate the radiometric information of the image and eliminate the influence of sensors and atmospheric conditions. The noise removal module can effectively reduce the random noise in the image and improve the image quality. Band selection selects the most suitable band combination according to specific application requirements, providing a high-quality data basis for subsequent fusion processing. Data preprocessing is one of the key steps of the present invention.

[0012] Further, the pixel decomposition module decomposes each pixel of the low-resolution image into multiple sub-pixels, significantly increasing the number of pixels in the image. This process not only improves the spatial resolution of the image but also provides richer detail information for subsequent fusion processing. The image after pixel decomposition is more abundant in spatial details and can better reflect the microscopic features of ground objects. Pixel decomposition technology is the core means to improve the spatial resolution of the image.

[0013] Further, the STARFM algorithm fusion module dynamically selects the best fusion strategy by analyzing image information in different bands and time scales. During the fusion process, the STARFM algorithm can make full use of the time information of the low-resolution image and the spatial information of the high-resolution image to generate a fusion image with both high temporal and high spatial resolutions. The STARFM algorithm is the key technology for the present invention to achieve the generation of high spatio-temporal resolution images. This algorithm is particularly suitable for processing dynamic ground object information such as vegetation growth and land use changes, and can effectively improve the temporal transferability and spatial detail performance of the image.

[0014] Further, the Gaussian smoothing processing module performs smoothing processing on the fused image through a Gaussian filter, which can effectively reduce the noise and artifacts in the image and improve the visual effect and analysis accuracy of the image. Gaussian smoothing processing is the post-processing step of the present invention, aiming to further improve the quality of the fused image. It is particularly important for subsequent application analysis and can ensure that the image has higher reliability and usability in practical applications.

[0015] Further, a satellite image generation method based on the STARFM algorithm includes the following steps:

[0016] Step 1: Decompose the pixels of the low-resolution image to increase the number of pixels;

[0017] Step 2: Use the STARFM algorithm to fuse the decomposed image with the high-resolution image;

[0018] Step 3: Perform Gaussian smoothing processing on the fused image to improve the image quality.

[0019] Advantages of the present invention compared with the prior art: The present invention processes low-resolution images through pixel decomposition technology, significantly increasing the number of pixels in the images, thereby improving the spatial resolution. On this basis, the STARFM algorithm is used to fuse the decomposed images with high-resolution images. The STARFM algorithm analyzes image information in different bands and time scales, selects the optimal fusion strategy, and generates high-quality fused images. In addition, the present invention also introduces data preprocessing and Gaussian smoothing processing modules to ensure the accuracy and consistency of the input data, further improve the quality of the fused images, and reduce noise interference.

[0020] High spatio-temporal resolution: Through pixel decomposition and STARFM algorithm fusion, the present invention can generate images with both high temporal and high spatial resolution, effectively solving the limitations of the prior art. Anti-weather interference: Compared with traditional satellite data processing methods, the present invention reduces the impact of weather on data quality through data fusion technology. Wide applicability: The system can be applied to multiple fields such as agriculture, environmental monitoring, and urban planning, and has broad application prospects. Brief Description of the Drawings

[0021] Figure 1 It is a flowchart of the satellite image generation method based on the STARFM algorithm of the present invention.

[0022] Figure 2 It is a flowchart of the present invention. Detailed Embodiments

[0023] The present invention will be further described below in conjunction with the drawings and specific embodiments, but it is not intended to limit the present invention.

[0024] As Figure 1-2 shown, the present invention works and is implemented as follows. A high spatio-temporal resolution satellite image generation system based on the STARFM algorithm is divided into a data preprocessing module, a pixel decomposition module, a STARFM algorithm fusion module, and a Gaussian smoothing processing module; the data preprocessing module is connected to the pixel decomposition module, the pixel decomposition module is connected to the STARFM algorithm fusion module, and the STARFM algorithm fusion module is connected to the Gaussian smoothing processing module.

[0025] Furthermore, the data preprocessing module includes geometric correction, radiometric calibration, noise removal, and band selection. By performing geometric correction on the original satellite data, the spatial position of the image is ensured to be accurate; radiometric calibration is used to calibrate the radiometric information of the image and eliminate the influence of sensors and atmospheric conditions; the noise removal module can effectively reduce the random noise in the image and improve the image quality; band selection selects the most suitable band combination according to specific application requirements, providing a high-quality data basis for subsequent fusion processing. Data preprocessing is one of the key steps of the present invention.

[0026] Furthermore, the pixel decomposition module decomposes each pixel of the low-resolution image into multiple sub-pixels, significantly increasing the number of pixels in the image. This process not only improves the spatial resolution of the image but also provides richer detailed information for subsequent fusion processing. The image after pixel decomposition is richer in spatial details and can better reflect the microscopic features of ground objects. Pixel decomposition technology is the core means to improve the spatial resolution of the image.

[0027] Furthermore, the STARFM algorithm fusion module dynamically selects the best fusion strategy by analyzing image information in different bands and time scales. During the fusion process, the STARFM algorithm can make full use of the time information of the low-resolution image and the spatial information of the high-resolution image to generate a fusion image with both high temporal and high spatial resolutions. The STARFM algorithm is the key technology for the present invention to achieve the generation of high spatio-temporal resolution images. This algorithm is particularly suitable for processing dynamic ground object information such as vegetation growth and land use changes, and can effectively improve the temporal transferability and spatial detail performance of the image.

[0028] Furthermore, the Gaussian smoothing processing module smooths the fused image through a Gaussian filter, which can effectively reduce noise and artifacts in the image, improving the visual effect and analysis accuracy of the image. Gaussian smoothing processing is a post-processing step of the present invention, aiming to further improve the quality of the fused image. It is particularly important for subsequent application analysis and can ensure higher reliability and usability of the image in practical applications.

[0029] Specifically, for a satellite image generation method based on the STARFM algorithm, the following steps are required:

[0030] The first step is data acquisition. First, low-resolution and high-resolution image data are obtained from different satellite platforms. Low-resolution images usually come from satellites with higher temporal resolution (such as MODIS), while high-resolution images come from satellites with higher spatial resolution (such as Landsat). These data need to cover the same geographical area for subsequent fusion processing.

[0031] The second step is data preprocessing. Before fusion, the collected image data is preprocessed to ensure the accuracy and consistency of the data. The preprocessing steps include:

[0032] Geometric correction: Geographically match images from different sources to ensure they are in the same geographical coordinate system.

[0033] Radiometric calibration: Convert the DN value (digital value) of the remote sensing image into a radiance value for subsequent spectral analysis.

[0034] Noise removal: Remove noise and outliers in the image through techniques such as filtering.

[0035] Band selection: Select important bands for fusion according to the actual application requirements.

[0036] The third step, pixel decomposition; Perform pixel decomposition on the low-resolution image to significantly improve the spatial resolution of the image by increasing the number of pixels. This process uses advanced pixel decomposition algorithms to decompose each low-resolution pixel into multiple sub-pixels, thus providing richer detail information for subsequent fusion.

[0037] The fourth step, STARFM algorithm fusion;

[0038] Use the STARFM algorithm to fuse the decomposed low-resolution image and the high-resolution image.

[0039] The STARFM algorithm is implemented through the following steps:

[0040] Construct a regression tree: According to the characteristics of the input image, construct a regression tree to determine the weights at each band and time scale.

[0041] Pixel-level fusion: According to the weights determined by the regression tree, perform pixel-level fusion on the low-resolution and high-resolution images to generate a preliminary fusion result.

[0042] Feature-level fusion: Extract and optimize the features in the preliminary fusion result to improve the quality of the fused image.

[0043] Spatial adaptive filtering: Perform spatial adaptive filtering on the fusion result to remove noise and outliers and improve the visual effect of the image.

[0044] The fifth step, Gaussian smoothing processing; To further improve the quality of the fused image, perform Gaussian smoothing processing on the fused image. Gaussian smoothing is a commonly used image denoising technique that reduces noise and details by applying a Gaussian filter to the image, thus improving the visual effect of the image and the accuracy of subsequent analysis.

[0045] The sixth step, post-processing and result output; Perform final post-processing on the image after Gaussian smoothing processing, including cropping, format conversion, etc., to meet the requirements of different application scenarios. The finally generated high spatio-temporal resolution image can be used in multiple fields such as agriculture, environmental monitoring, and urban planning.

Claims

1. A high temporal and spatial resolution satellite image generation system based on STARFM algorithm, characterized by: It is divided into data preprocessing module, pixel decomposition module, STARFM algorithm fusion module and Gaussian smoothing processing module; The data preprocessing module is connected to the pixel decomposition module, the pixel decomposition module is connected to the STARFM algorithm fusion module, and the STARFM algorithm fusion module is connected to the Gaussian smoothing processing module.

2. The high spatiotemporal resolution satellite image generation system based on the STARFM algorithm according to claim 1, characterized in that: The data preprocessing module includes geometric correction, radiometric calibration, noise removal and band selection. It performs geometric correction on the original satellite data to ensure the accuracy of the spatial position of the image. Radiometric calibration is used to calibrate the radiation information of the image to eliminate the influence of sensors and atmospheric conditions. The noise removal module can effectively reduce the random noise in the image and improve the image quality; the band selection module selects the most suitable band combination according to the specific application requirements to provide a high-quality data foundation for subsequent fusion processing.

3. A high spatiotemporal resolution satellite image generation system based on STARFM algorithm according to claim 2, characterized in that: The pixel decomposition module decomposes each pixel of the low-resolution image into multiple sub-pixels, significantly increasing the number of pixels in the image.

4. The high temporal and spatial resolution satellite image generation system based on the STARFM algorithm according to claim 3, characterized in that: The STARFM algorithm fusion module dynamically selects the best fusion strategy by analyzing image information of different bands and time scales. During the fusion process, the STARFM algorithm can make full use of the temporal information of low-resolution images and the spatial information of high-resolution images to generate a fused image with both high temporal and high spatial resolution.

5. The high spatiotemporal resolution satellite image generation system based on the STARFM algorithm according to claim 4, characterized in that: The Gaussian smoothing processing module smoothes the fused image through a Gaussian filter, which can effectively reduce the noise and artifacts in the image and improve the visual effect and analysis accuracy of the image.

6. The high temporal and spatial resolution satellite image generation system based on the STARFM algorithm according to claim 5, characterized in that: A satellite image generation method based on the STARFM algorithm comprises the following steps: Step 1: Decompose the low-resolution image into pixels and increase the number of pixels; Step 2: Use the STARFM algorithm to fuse the decomposed image with the high-resolution image; Step 3: Perform Gaussian smoothing on the fused image to improve image quality.