CN115457405A - High-resolution remote sensing target detection method supported by quantitative atmospheric correction - Google Patents
High-resolution remote sensing target detection method supported by quantitative atmospheric correction Download PDFInfo
- Publication number
- CN115457405A CN115457405A CN202211128954.1A CN202211128954A CN115457405A CN 115457405 A CN115457405 A CN 115457405A CN 202211128954 A CN202211128954 A CN 202211128954A CN 115457405 A CN115457405 A CN 115457405A
- Authority
- CN
- China
- Prior art keywords
- resolution
- remote sensing
- image
- atmospheric
- target detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 31
- 238000012937 correction Methods 0.000 title claims abstract description 27
- 238000013135 deep learning Methods 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 238000012795 verification Methods 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims abstract description 6
- 230000004927 fusion Effects 0.000 claims abstract description 4
- 238000007499 fusion processing Methods 0.000 claims abstract description 4
- 230000005855 radiation Effects 0.000 claims description 13
- 230000003287 optical effect Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000013136 deep learning model Methods 0.000 claims description 4
- 239000000443 aerosol Substances 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000002310 reflectometry Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 230000001360 synchronised effect Effects 0.000 abstract 1
- 238000001228 spectrum Methods 0.000 description 2
- 239000005427 atmospheric aerosol Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a high-resolution remote sensing target detection method supported by quantitative atmospheric correction, which comprises the following basic steps: acquiring a high-resolution satellite image containing a target object and free of clouds; sequentially carrying out orthorectification and geostationary satellite atmospheric product data support synchronous atmospheric rectification and multispectral and panchromatic image fusion on the high-resolution satellite image; carrying out sample marking on the high-resolution satellite image subjected to atmospheric correction and fusion processing, and dividing the marked sample into a training set, a verification set and a test set to obtain a target sample data set; and carrying out sample training, verification and testing by using a target detection network based on deep learning to obtain a target detection result. The method combines information of quantitative remote sensing and graphic images in the process of detecting the high-molecular remote sensing target by the deep learning network, thereby increasing the target information amount input into the network.
Description
Technical Field
The invention relates to the field of satellite remote sensing data processing technology and computer vision, in particular to the field of high-resolution remote sensing target detection application, and specifically relates to a high-resolution remote sensing target detection method supported by quantitative atmospheric correction.
Background
At present, the ground target is detected based on the high-spectrum optical remote sensing image, and the detection is realized by using a high-spectrum satellite image which is not corrected by atmosphere and a deep learning method based on computer vision. The deep learning method is widely applied because of the strong feature representation capability shown in the aspect of target detection, but a deep learning algorithm has a certain improvement space in detection precision and detection efficiency, and particularly aims at the target detection problem in small-scale high-resolution optical remote sensing images. On the other hand, at present, no people apply the quantified remote sensing information to the high-resolution satellite remote sensing target detection, only apply the deep learning network model to the remote sensing field, and the actual technical route is strongly related to deep learning, but the professional scientific knowledge in remote sensing is hardly mined and utilized for supporting detection. The detected target belongs to the earth surface signal of the lower boundary of the atmosphere in the remote sensing image, and the remote sensing image after atmosphere contribution is deducted really represents the ground target signal. The invention provides a high-resolution remote sensing target detection method supported by quantitative atmospheric correction, which is used for quantitative processing of remote sensing information by taking atmospheric radiation correction as a main means and fully combining information in two aspects of quantitative remote sensing and graphic images.
Disclosure of Invention
Technical problem to be solved
The invention provides a high-resolution remote sensing target detection method supported by quantitative atmospheric correction, which fully considers the application of specific quantitative information of remote sensing targets different from common ground targets, and combines target information in two aspects of quantitative remote sensing and graphic images to support high-resolution remote sensing target detection based on a deep learning method.
(II) technical scheme
In order to realize the technical task, the invention adopts the following technical scheme that the high-resolution remote sensing target detection method supported by quantitative atmospheric correction comprises the following steps:
(1) Acquiring and screening a high-resolution satellite image containing a target object and without clouds;
(2) Orthorectification is carried out on the high-resolution satellite image to obtain accurate longitude and latitude of each pixel in the image;
(3) Matching the high-resolution satellite images with the atmospheric parameters such as the optical thickness of the aerosol, the water vapor content of the column and the like in the same time and space by using the high-resolution geostationary satellite atmospheric product, and performing atmospheric radiation correction based on a radiation transmission model on the high-resolution satellite images to obtain the high-resolution satellite images after atmospheric correction, namely earth surface reflectivity images;
(4) Performing image fusion on the multispectral image and the panchromatic image after the atmospheric correction, and increasing the multispectral image of the satellite with high spatial resolution to the spatial resolution which is the same as that of the panchromatic image;
(5) Marking a target sample of the high-resolution satellite image subjected to atmospheric correction and fusion processing, and dividing marked sample data into a training set, a verification set and a test set;
(6) Inputting the marked sample data into a deep learning network to adjust parameters for training and verification, obtaining a deep learning model after training, and performing target detection on the test set by using the deep learning model to obtain target detection result precision;
further, the step (3) is quantitative processing of the remote sensing image by taking atmospheric radiation correction as a main means, and quantitative remote sensing information of radiation dimension is added for input of a deep learning network, so that information combination of quantitative remote sensing and graphic images is applied to high-resolution remote sensing target detection.
(III) advantageous effects
The advantages of the invention are embodied in that:
scientific knowledge in the field of remote sensing is fully utilized, specific quantitative information contained in the high-resolution remote sensing target is applied to a deep learning network, quantitative remote sensing and graphic image information are jointly used for high-resolution remote sensing target detection, and information of a ground target is enriched.
Drawings
FIG. 1 is a flow chart of the steps performed in the present invention.
Detailed Description
To further clarify the objects, contents and advantages of the present invention, a full and enabling description of the embodiments of the present invention will be given below by way of example only, with reference to the accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the invention, are within the scope of the invention.
The following detailed description of specific embodiments of the present invention is made with reference to the accompanying drawings and examples.
Referring to fig. 1, the present invention takes himwari-8 as a high orbit geostationary satellite, a high-grade second satellite as a low orbit high grade satellite, a deep learning network as a YOLOv5 network, and an airplane as a ground target as an embodiment, and provides a high grade remote sensing target detection method supported by quantitative atmospheric correction, which includes:
(1) Acquiring and screening a high-grade second satellite image which contains an airplane target and is free of clouds;
the high-resolution second satellite image is composed of a multispectral image with 4m spatial resolution and a panchromatic image with 1m spatial resolution.
(2) And performing orthorectification on the high-resolution second satellite image to obtain accurate longitude and latitude of each pixel in the image.
(3) Atmospheric radiation correction based on a 6S radiation transmission model is carried out on the second high-resolution satellite image by utilizing the matching of the atmospheric aerosol product observed by Himapari-8 and the spatial-temporal aerosol optical thickness of the second high-resolution satellite image after orthorectification to obtain the second high-resolution satellite image after atmospheric correction, namely a ground surface reflectivity image;
the remote sensing image quantitative processing with atmospheric radiation correction as a main means increases quantitative remote sensing information of radiation dimension for input of a deep learning network, and therefore information of quantitative remote sensing and graphic images is combined and applied to high-resolution second-grade remote sensing airplane detection.
(4) And carrying out image fusion on the multispectral image subjected to atmospheric correction and the panchromatic image, and increasing the multispectral image with the spatial resolution of 4m to the spatial resolution which is the same as that of the panchromatic image with the spatial resolution of 1m, so as to obtain the high-resolution second-order satellite image with the spatial resolution of 1m multiplied by 1 m.
(5) Carrying out airplane sample marking on the high-resolution second satellite image subjected to atmospheric correction and fusion processing, and dividing marked sample data into a training set, a verification set and a test set according to the proportion of 7.
(6) And inputting the marked sample data into a YOLOv5 network to adjust parameters for training and verification, obtaining a weight file with the best training result, and performing airplane detection on the test set in the network by using the weight file to finally obtain an airplane detection result and detection precision.
Claims (2)
1. A high-resolution remote sensing target detection method supported by quantitative atmospheric correction is characterized by comprising the following steps:
(1) Acquiring and screening out a high-resolution satellite image which contains a target object and is free of cloud;
(2) Orthorectification is carried out on the high-resolution satellite image to obtain accurate longitude and latitude of each pixel in the image;
(3) Matching the high-resolution satellite images with atmospheric parameters such as aerosol optical thickness, column water vapor content and the like of the same time and space by using a high-orbit geostationary satellite atmospheric product, and performing atmospheric radiation correction based on a radiation transmission model on the high-resolution satellite images to obtain high-resolution satellite images after atmospheric correction, namely earth surface reflectivity images;
(4) Performing image fusion on the multispectral image and the panchromatic image after the atmospheric correction, and increasing the multispectral image of the satellite with high spatial resolution to the spatial resolution which is the same as that of the panchromatic image;
(5) Marking a target sample of the high-resolution satellite image subjected to atmospheric correction and fusion processing, and dividing marked sample data into a training set, a verification set and a test set;
(6) And inputting the marked sample data into a deep learning network to adjust parameters for training and verification, obtaining a deep learning model after training, and performing target detection on the test set by using the deep learning model to obtain the target detection result precision.
2. The method for detecting the high-resolution remote sensing target supported by the quantified atmospheric correction according to claim 1, characterized in that: and (3) carrying out quantitative processing on the remote sensing image by taking atmospheric radiation correction as a main means, and adding quantitative remote sensing information of radiation dimension for the input of the deep learning network, so that the information of quantitative remote sensing and graphic images is combined and applied to high-resolution remote sensing target detection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211128954.1A CN115457405A (en) | 2022-09-16 | 2022-09-16 | High-resolution remote sensing target detection method supported by quantitative atmospheric correction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211128954.1A CN115457405A (en) | 2022-09-16 | 2022-09-16 | High-resolution remote sensing target detection method supported by quantitative atmospheric correction |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115457405A true CN115457405A (en) | 2022-12-09 |
Family
ID=84305296
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211128954.1A Pending CN115457405A (en) | 2022-09-16 | 2022-09-16 | High-resolution remote sensing target detection method supported by quantitative atmospheric correction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115457405A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116385903A (en) * | 2023-05-29 | 2023-07-04 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Anti-distortion on-orbit target detection method and model for 1-level remote sensing data |
-
2022
- 2022-09-16 CN CN202211128954.1A patent/CN115457405A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116385903A (en) * | 2023-05-29 | 2023-07-04 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Anti-distortion on-orbit target detection method and model for 1-level remote sensing data |
CN116385903B (en) * | 2023-05-29 | 2023-09-19 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Anti-distortion on-orbit target detection method and model for 1-level remote sensing data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Henriksen et al. | Extracting accurate and precise topography from LROC narrow angle camera stereo observations | |
Wolfe et al. | Achieving sub-pixel geolocation accuracy in support of MODIS land science | |
CN106767714B (en) | Improve the equivalent mismatch model multistage Calibration Method of satellite image positioning accuracy | |
Martínez-Vazquez et al. | Search for RR Lyrae stars in DES ultrafaint systems: Grus I, Kim 2, Phoenix II, and Grus II | |
Li et al. | ZY‐3 Block adjustment supported by glas laser altimetry data | |
Crawford et al. | The 50-year Landsat collection 2 archive | |
CN111473802A (en) | Optical sensor internal orientation element calibration method based on linear array push-scanning | |
CN104361563B (en) | GPS-based (global positioning system based) geometric precision correction method of hyperspectral remote sensing images | |
CN113284171A (en) | Vegetation height analysis method and system based on satellite remote sensing stereo imaging | |
Languille et al. | Sentinel-2 geometric image quality commissioning: First results | |
CN115457405A (en) | High-resolution remote sensing target detection method supported by quantitative atmospheric correction | |
CN104236553B (en) | Autonomous all-weather stellar refraction satellite location method | |
Ouyed et al. | Global three‐dimensional water vapor feature‐tracking for horizontal winds using hyperspectral infrared sounder data from overlapped tracks of two satellites | |
Li et al. | The influence of sub‐footprint cloudiness on three‐dimensional horizontal wind from geostationary hyperspectral infrared sounder observations | |
Muller et al. | EU-FP7-iMARS: analysis of Mars multi-resolution images using auto-coregistration, data mining and crowd source techniques: processed results–a first look | |
Kim et al. | Remote Sensing and Data Analyses on Planetary Topography | |
Wolfe | MODIS geolocation | |
Jeong et al. | Geometric accuracy analysis of the geostationary ocean color imager (GOCI) level 1B (L1B) product | |
Ihsan et al. | The Comparison of Stage and Result Processing of Photogrammetric Data Based on Online Cloud Processing | |
Jovanovic et al. | Multi-angle geometric processing for globally geo-located and co-registered MISR image data | |
Restrepo et al. | Building Lunar Maps for Terrain Relative Navigation and Hazard Detection Applications | |
Folger et al. | Lidar-based height correction for the assimilation of atmospheric motion vectors | |
Robbins et al. | Fully controlled 6 meters per pixel mosaic of Mars's south polar region | |
Languille et al. | Sentinel-2B Image Quality commissioning phase results and Sentinel2 constellation performances | |
Hare et al. | Cartography tools |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |