Space-based infrared detection scene cloud Mask image simulation method

Abstract

The invention discloses a space-based infrared detection scene cloud Mask image simulation method. The method comprises the following steps: firstly, preprocessing a space-based infrared on-orbit image to improve the image quality; then, utilizing a threshold segmentation technology to accurately extract cloud layer contours and distribution; a segmentation result is optimized through morphological operation, a cloud cluster region is extracted and stored by using a connected region marking algorithm, and a cloud cluster database is constructed; and finally, randomly generating a simulation cloud Mask image based on the database, iteratively covering the cloud cluster to a target area by setting initial parameters and coverage rate requirements, and outputting a cloud Mask image meeting conditions. According to the method, the simulation cloud Mask graph conforming to the actual distribution rule can be generated, and the task requirement of space-based infrared detection cloud scene digital simulation can be effectively met.

Description

Technical Field

[0001] The present invention belongs to the field of infrared image processing technology and image simulation. The main research content is an on-orbit cloud mask image simulation algorithm for complex space-based infrared ground detection scenes, and particularly a cloud mask image simulation method for space-based infrared detection scenes. Background Art

[0002] With the rapid development of technologies such as artificial intelligence, the Internet of Things, and big data, the concept and application of digital models have gradually become popular and become a research hotspot. At present, the demand for digital modeling of space infrared payloads is increasing. In the digital modeling of infrared payloads, high-precision detection background is an important input of the digital model, which directly affects the performance of model simulation. Cloud background is one of the most common backgrounds in space-based infrared detection, which is due to the wide distribution of clouds in the earth's atmosphere, especially in medium and low orbit detection, when clouds appear in the field of view at high frequency. There are various types of clouds, and there are significant differences in their temperature, thickness, and optical properties, which makes the cloud background present complex changes. In addition, the temperature difference between the cloud background and the aerial target is significant, which becomes the main interference factor in infrared detection. Therefore, in order to solve the problem of lack of high-precision detection background faced by the current space-based infrared detection mission, the present invention aims to face the digitization of infrared payloads, simulate and generate complex cloud images in high-fidelity space-based wide-area detection scenes, and provide scene input for the digital development of space-based infrared payloads.

[0003] There are three main types of cloud simulation methods: the first is the method based on measured data, which simulates the cloud morphology through spectrum transformation, resolution adjustment and feature extraction. This method is easy to operate and the results are verifiable, but it is limited by the data source; the second is the method based on physical models, which constructs a fluid model by solving partial differential equations, and generates a more realistic cloud simulation effect according to the set parameters and initial conditions, but this method has large computational complexity and low real-time performance; the last is the method based on deep learning, which uses technologies such as neural networks to generate cloud scene images. This method is highly flexible but relies on a large amount of data sets. Summary of the invention

[0004] The purpose of the present invention is to provide a method for simulating cloud mask images of space-based infrared detection scenes, which mainly solves the above-mentioned technical problems of low simulation efficiency and limitation to databases. The present invention can not only generate a simulated cloud mask image that conforms to the actual distribution law, but also effectively meet the task requirements of digital simulation of space-based infrared detection cloud scenes.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution of the present invention is:

[0006] A space-based infrared detection scene cloud Mask image simulation method, characterized in that the method steps are:

[0007] First, preprocess the space-based infrared on-orbit images (denoising and contrast enhancement) to improve image quality;

[0008] Then, the threshold segmentation technique is used to accurately extract the cloud contours and distribution;

[0009] The segmentation results are optimized through morphological operations (hole filling, closing operation, opening operation, and erosion), and the cloud area is extracted and stored using the connected region labeling algorithm to build a cloud database.

[0010] Finally, a simulated cloud Mask image is randomly generated based on the database. By setting the initial parameters and coverage requirements, the cloud cluster is iteratively covered to the target area, and a cloud Mask image that meets the conditions is output.

[0011] Through the above method steps, it is possible to dynamically generate simulated cloud images that conform to the actual distribution characteristics to meet the simulation and prediction needs. The overall process is highly automated, covering the complete application chain from preprocessing, segmentation optimization to simulation generation. It has strong versatility and innovation, and provides a practical and efficient solution for cloud mask image simulation. This method is particularly suitable for the simulation and radiation characteristic calculation of space-based infrared cloud images, and can be used as an input data source for payload digitization scenes, providing important support for related research and applications.

[0012] The space-based infrared detection scene cloud Mask image simulation method is characterized in that the method steps specifically include:

[0013] A. Obtain high-resolution space-based on-orbit image data. These images cover visible light and infrared bands and can clearly present the distribution of clouds. Before cloud segmentation, the image needs to be preprocessed, including denoising, contrast enhancement and geometric correction, to improve image quality and analysis accuracy. The space-based detection scene required by the present invention refers to the scene of long-term observation of the earth in space-based target detection.

[0014] B. Use threshold segmentation technology to select appropriate thresholds based on the difference in grayscale characteristics between clouds and background, accurately separate clouds from the background, and obtain clear cloud contours and distribution maps. This process not only considers static thresholds, but also uses adaptive threshold methods, which can better cope with cloud segmentation tasks under different lighting and atmospheric conditions.

[0015] C. Establish a cloud database based on the cloud map segmented by on-orbit images. First, assume that the initial image is img, and fill the inner holes in the binary cloud map through hole filling of morphological reconstruction to ensure the integrity of the cloud area. Let the hole filling structure be SE, and use morphological closing operation to fill the small gaps or breaks at the edge of the cloud to make the target area more coherent:

[0016]

[0017] Then, the isolated noise in the image is removed by morphological opening operation;

[0018] img open =img close oS

[0019] Then, the artifacts or interference at the edge of the cloud are removed through corrosion operation.

[0020]

[0021] Finally, the separated clouds in the image are marked by the connected region labeling algorithm, and the number of regions is counted, so as to achieve cloud segmentation and analysis, and establish a cloud database.

[0022] D. Set the initial parameters of the simulated cloud mask image, such as cloud coverage and cloud mask image size.

[0023] E. Simulate cloud mask images based on cloud database. First, randomly select a cloud from the cloud database to determine whether the target area in the current mask image is empty (i.e., whether it is a cloud-free area); if the area is cloud-free, cover the area with the selected cloud; during the covering process, ensure that the shape and position of the cloud match the target area while avoiding interference with other areas. Repeating this operation can effectively generate randomized simulated cloud mask images, providing support for further research or algorithm testing.

[0024] F. Determine whether the coverage meets the set requirements. If not, repeat step E. If it does, output the cloud mask image.

[0025] The cloud mask image simulation method in a space-based detection scenario is characterized in that: in step C, a cloud database is established based on the cloud map segmented by on-orbit images.

[0026] The cloud mask image simulation method in a space-based detection scenario is characterized by: the cloud mask image is randomly simulated based on the cloud database in step E.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. The method of the present invention effectively ensures the integrity of cloud shape and the accuracy of segmentation through morphological operations and region marking.

[0029] 2. The method of the present invention is based on space-based infrared on-orbit images. We have constructed a real cloud database that can provide data support for cloud simulation.

[0030] 3. The method of the present invention can effectively solve the drawback of cloud image simulation that relies heavily on data through the random simulated cloud Mask image algorithm of the on-orbit cloud database, and can realize random cloud images with multiple coverage and high confidence. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a flowchart of the method of the present invention.

[0032] Figure 2 It is the original on-track image of the method embodiment of the present invention.

[0033] Figure 3 This is the image after threshold segmentation in the method embodiment of the present invention.

[0034] Figure 4 This is the image after morphology and connected domain labeling in the embodiment of the method of the present invention.

[0035] Figure 5 These are Mask simulation images of different cloud coverage ratios according to the method embodiment of the present invention. DETAILED DESCRIPTION

[0036] The present invention discloses a method for simulating cloud Mask images of space-based infrared detection scenes. Figure 1 As shown: The method steps are:

[0037] First, preprocess the space-based infrared on-orbit images (denoising and contrast enhancement) to improve image quality;

[0038] Then, the threshold segmentation technique is used to accurately extract the cloud contours and distribution;

[0039] The segmentation results are optimized through morphological operations (hole filling, closing operation, opening operation, and erosion), and the cloud area is extracted and stored using the connected region labeling algorithm to build a cloud database.

[0040] Finally, a simulated cloud Mask image is randomly generated based on the database. By setting the initial parameters and coverage requirements, the cloud cluster is iteratively covered to the target area, and a cloud Mask image that meets the conditions is output.

[0041] Through the above method steps, it is possible to dynamically generate simulated cloud images that conform to the actual distribution characteristics to meet the simulation and prediction needs. The overall process is highly automated, covering the complete application chain from preprocessing, segmentation optimization to simulation generation. It has strong versatility and innovation, and provides a practical and efficient solution for cloud mask image simulation. This method is particularly suitable for the simulation and radiation characteristic calculation of space-based infrared cloud images, and can be used as an input data source for payload digitization scenes, providing important support for related research and applications.

[0042] Please read further Figure 1 The method steps specifically include:

[0043] A. Obtain high-resolution space-based on-orbit image data. These images cover visible light and infrared bands and can clearly present the distribution of clouds. Before cloud segmentation, the image needs to be preprocessed, including denoising, contrast enhancement and geometric correction, to improve image quality and analysis accuracy. The space-based detection scene required by the present invention refers to the scene of long-term observation of the earth in space-based target detection.

[0044] B. Use threshold segmentation technology to select appropriate thresholds based on the difference in grayscale characteristics between clouds and background, accurately separate clouds from the background, and obtain clear cloud contours and distribution maps. This process not only considers static thresholds, but also uses adaptive threshold methods, which can better cope with cloud segmentation tasks under different lighting and atmospheric conditions.

[0045] C. Establish a cloud database based on the cloud map segmented by on-orbit images. First, assume that the initial image is img, and fill the inner holes in the binary cloud map through hole filling of morphological reconstruction to ensure the integrity of the cloud area. Let the hole filling structure be SE, and use morphological closing operation to fill the small gaps or breaks at the edge of the cloud to make the target area more coherent:

[0046]

[0047] Then, the isolated noise in the image is removed by morphological opening operation;

[0048] img open =img close oS

[0049] Then, the artifacts or interference at the edge of the cloud are removed through corrosion operation.

[0050]

[0051] Finally, the separated clouds in the image are marked by the connected region labeling algorithm, and the number of regions is counted, so as to achieve cloud segmentation and analysis, and establish a cloud database.

[0052] D. Set the initial parameters of the simulated cloud mask image, such as cloud coverage and cloud mask image size.

[0053] E. Simulate cloud mask images based on cloud database. First, randomly select a cloud from the cloud database to determine whether the target area in the current mask image is empty (i.e., whether it is a cloud-free area); if the area is cloud-free, cover the area with the selected cloud; during the covering process, ensure that the shape and position of the cloud match the target area while avoiding interference with other areas. Repeating this operation can effectively generate randomized simulated cloud mask images, providing support for further research or algorithm testing.

[0054] F. Determine whether the coverage meets the set requirements. If not, repeat step E. If it does, output the cloud mask image.

[0055] The method of the present invention will be described in detail below in conjunction with the embodiments and drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0056] First, obtain the on-orbit images in the space-based detection scene, such as Figure 2 As shown. Threshold segmentation is performed on the on-track image, where Figure 2 The selected segmentation threshold is 0.9, such as Figure 3 To extract clouds from on-orbit images, firstly, holes are filled, and the filling parameter is a disk with a radius of 1 pixel. Then, morphological operations, namely closing, opening and erosion operations, are performed, and connected domain detection is used to extract clouds, as shown in Fig. Figure 4 As shown. Set the initial cloud mask image size to 256*320, set the cloud coverage, and generate a random cloud mask image based on the on-orbit cloud database. Figure 5 As shown (coverages are 0.1, 0.2, 0.3, 0.4 and 0.5 respectively).

Claims

1. A method for simulating cloud Mask images in space-based infrared detection scenes, characterized by: The steps of this method are: First, pre-process the space-based infrared on-orbit images to improve image quality; Then, the threshold segmentation technique is used to accurately extract the cloud contours and distribution; The segmentation results are optimized through morphological operations, and the cloud regions are extracted and stored using the connected region labeling algorithm to build a cloud database. Finally, a simulated cloud Mask image is randomly generated based on the database. By setting the initial parameters and coverage requirements, the cloud cluster is iteratively covered to the target area, and a cloud Mask image that meets the conditions is output.

2. The method for simulating cloud Mask images of space-based infrared detection scenes according to claim 1, characterized in that: The method steps specifically include: A. Obtain remote sensing image data collected by on-orbit space-based infrared detectors that can clearly present cloud distribution; before cloud segmentation, the image needs to be preprocessed; this preprocessing includes denoising and contrast enhancement to improve image quality and analysis accuracy; B. Using threshold segmentation technology, based on the difference in grayscale characteristics between clouds and background, select a suitable threshold to accurately separate clouds from the background and obtain a clear cloud outline and distribution map; C. Establish a cloud database based on the cloud map segmented by on-orbit images. First, assume that the initial image is img, and fill the inner holes in the binary cloud map through hole filling of morphological reconstruction to ensure the integrity of the cloud area. Let the hole filling structure be SE, and use morphological closing operation to fill the small gaps or breaks at the edge of the cloud to make the target area more coherent: Then, the isolated noise in the image is removed by morphological opening operation; img open =img close oSE Then, the artifacts or interference at the edge of the cloud are removed through the erosion operation; Finally, the separated clouds in the image are marked by the connected region labeling algorithm, and the number of regions is counted, so as to achieve the segmentation and analysis of clouds and establish a cloud database; D. Set the initial parameters of the cloud mask simulation image, including: cloud coverage and cloud mask image size; E. Simulate cloud mask images based on cloud database. First, randomly select a cloud from the cloud database to determine whether the target area in the current mask image is empty, that is, whether it is a cloud-free area. If the area is cloud-free, cover the selected cloud with the area. During the covering process, ensure that the shape and position of the cloud match the target area while avoiding interference with other areas. Repeat the operation to effectively generate randomized simulated cloud mask images, which provides support for further research or algorithm testing. F. Determine whether the coverage meets the set requirements. If not, repeat step E. If it does, output the cloud mask image.

3. The cloud mask image simulation method in a space-based detection scenario according to claim 2, characterized in that: In step C, a cloud database is established based on the cloud map segmented from the on-orbit image.

4. The cloud mask image simulation method in a space-based detection scenario according to claim 2, characterized in that: In step E, the simulated cloud Mask image is randomized based on the cloud database.

Images