CN118691536B - Automatic screening method and system based on remote sensing image content - Google Patents
Automatic screening method and system based on remote sensing image content Download PDFInfo
- Publication number
- CN118691536B CN118691536B CN202410233861.8A CN202410233861A CN118691536B CN 118691536 B CN118691536 B CN 118691536B CN 202410233861 A CN202410233861 A CN 202410233861A CN 118691536 B CN118691536 B CN 118691536B
- Authority
- CN
- China
- Prior art keywords
- image
- satellite
- regional
- screening
- task
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/54—Browsing; Visualisation therefor
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/56—Information retrieval; Database structures therefor; File system structures therefor of still image data having vectorial format
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06V10/766—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using regression, e.g. by projecting features on hyperplanes
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Remote Sensing (AREA)
- Probability & Statistics with Applications (AREA)
- Astronomy & Astrophysics (AREA)
- Quality & Reliability (AREA)
- Image Processing (AREA)
Abstract
本发明公开了基于遥感影像内容的自动筛选方法及系统，涉及卫星遥感影像筛选技术领域，通过使用聚类算法对区域参数进行聚类分析，获得K个区域聚类簇，对于每个区域聚类簇，训练出判断卫星图像是否被选中的遥感图像筛选模型和评估每张卫星图像的排序序号的图像序号预测模型，计算获得任务区域参数对应的任务区域聚类簇，使用任务区域聚类簇对应的遥感图像筛选模型，筛选出任务选中图像集合，使用任务区域聚类簇对应的图像序号预测模型，将卫星图像进行编号；实现了对卫星图像的自动筛选和排序，减少了筛选人员的劳动成本，提高卫星遥感图像的筛选效率。
The invention discloses an automatic screening method and system based on remote sensing image content, and relates to the technical field of satellite remote sensing image screening. By using a clustering algorithm to perform cluster analysis on regional parameters, K regional cluster clusters are obtained. For each regional cluster cluster, a remote sensing image screening model for judging whether a satellite image is selected and an image sequence number prediction model for evaluating the sorting sequence number of each satellite image are trained. A task area cluster cluster corresponding to a task area parameter is obtained by calculation. The remote sensing image screening model corresponding to the task area cluster cluster is used to screen out a task selected image set. The image sequence number prediction model corresponding to the task area cluster cluster is used to number the satellite images. The automatic screening and sorting of satellite images is realized, the labor cost of screening personnel is reduced, and the screening efficiency of satellite remote sensing images is improved.
Description
Technical Field
The invention relates to the technical field of satellite remote sensing image screening, in particular to an automatic screening method and system based on remote sensing image content.
Background
With the development of remote sensing technology, satellite images have become an important means for acquiring earth information. However, since the number of satellite images is enormous and there is a difference in image quality, how to effectively screen and process these images becomes an important study subject;
the number of satellite images is very large, and images acquired by the satellites contain abundant earth information such as topography, land feature, vegetation, water body and the like.
And because of the large number of satellites, there are differences in the quality of satellite images. Due to different factors such as the sensor type, orbit height and shooting time of the satellite, obvious quality difference can exist in satellite images in the same region, for example, the resolution of some images can be lower and details on the ground can not be clearly seen;
Therefore, when the satellite images of a region are referred, the satellite images need to be screened, so that satellite images with higher quality or shorter time are screened;
at present, the screening work of satellite images is mainly performed in a manual screening mode, a great deal of manpower is required to be consumed, or a plurality of screening rules are customized, for example, the screening rules are preferably selected to be smaller in cloud quantity, the selection time is relatively short under the condition that the cloud quantity is the same, and the screening rules are relatively dead, so that satellite images with relatively poor quality can be frequently screened out;
Therefore, the invention provides an automatic screening method and system based on remote sensing image content.
Disclosure of Invention
The present invention aims to solve at least one of the technical problems existing in the prior art. Therefore, the invention provides the automatic screening method and the system based on the remote sensing image content, which realize the automatic screening and sequencing of the satellite images, reduce the labor cost of screening personnel and improve the screening efficiency of the satellite remote sensing images.
In order to achieve the above object, embodiment 1 of the present invention provides an automatic screening method based on remote sensing image content, comprising the following steps:
Collecting a plurality of regional image data sets, and dividing each regional image data set into a selected data set and an unselected data set;
collecting the regional parameters corresponding to each regional image data set, and collecting the image parameters of each satellite image in each selected data set and each unselected data set;
Performing cluster analysis on the regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters;
training a remote sensing image screening model for judging whether the satellite images are selected or not by using the image parameters of each satellite image in the selected data set and the unselected data set for each regional cluster;
collecting the display sequence number of each satellite image in each selected data set in each regional cluster;
Training an image sequence number prediction model for evaluating the sequencing sequence number of each satellite image based on the image parameters and the display sequence number of each satellite image in each selected data set for each regional cluster;
Step seven, after the remote sensing image screening model and the image sequence number prediction model are trained, collecting task area parameters and task image sets corresponding to the latest image screening task;
Step eight, calculating and obtaining a region cluster corresponding to the task region parameters as a task region cluster;
step nine, screening a task selected image set from the task image set by using a remote sensing image screening model corresponding to the task regional cluster;
numbering satellite images in the task selected image set by using an image sequence number prediction model corresponding to the task region cluster, and displaying the satellite images in the task selected image set according to the numbering sequence;
the method for collecting a plurality of regional image data sets and dividing each regional image data set into a selected data set and an unselected data set is as follows:
Collecting all satellite images corresponding to the screening when the satellite images are screened manually each time; all satellite images when manually screening satellite images are used as a group of regional image data sets each time;
when each time of manual screening is carried out, partial satellite images screened by screening personnel form a selected data set corresponding to the regional image data set, and other satellite images except the selected data set in the regional image data set form an unselected data set;
The method for collecting the regional parameters corresponding to each regional image dataset comprises the following steps:
for each regional image dataset, collecting the longitude median value, the latitude median value, the longitude span, the latitude span, the regional type and the regional area of the ground region corresponding to the regional image dataset together form regional parameters during manual screening;
The method for collecting the image parameters of each satellite image in each selected data set and each unselected data set is as follows:
for each satellite image in each selected data set and each unselected data set, collecting the effective percentage, resolution, cloud cover, side view angle, shooting time difference and ground object resolution of a coverage selection area corresponding to the satellite image as image parameters;
The effective percentage of the coverage selection area is the proportion of the satellite image covering the target area;
the effective percentage of the coverage selection area is calculated by the following steps:
collecting a vector range (selection range) of a target area;
Converting the boundary of the satellite image into corresponding ground space coordinates by using an RPC model, wherein the converted ground space coordinates form an image coverage area;
Calculating the range of the image coverage area in the vector range of the target area as an image effective coverage selection area, calculating the proportion of the area of the image effective coverage selection area to the area of the target area, and taking the proportion as the effective percentage of the image coverage selection area;
the clustering algorithm is used for carrying out clustering analysis on the regional parameters, and the mode for obtaining K regional clustering clusters is as follows:
forming a one-dimensional array by corresponding region parameters of each region image dataset, and taking the one-dimensional array as coordinates of a data point;
dividing K clustering clusters for all data points by using a clustering algorithm;
the mode of training the remote sensing image screening model for judging whether the satellite image is selected is as follows:
For each regional cluster:
Marking the selection label of each satellite image in the selected data set of all the regional image data sets in the regional cluster as 1, and marking the selection label of each satellite image in the unselected data set as 0;
The method comprises the steps of forming a feature vector by image parameters of each satellite image, taking the feature vector as input of a remote sensing image screening model, taking a selection tag predicted value of each satellite image as output, taking the selection tag predicted value as 0 or 1, taking a selection tag of each satellite image as a predicted target, taking the square of the difference between the selection tag predicted value and the selection tag as a first predicted error corresponding to each satellite image, and minimizing the sum of the first predicted errors of all satellite images as a training target;
In the collecting of each regional cluster, the mode of showing sequence numbers of each satellite image in each selected data set is as follows:
when screening personnel manually screen the satellite images each time, collecting serial numbers which are sequenced by the screening personnel for the selected satellite images as display serial numbers for each satellite image in the selected data set;
the mode of training the image sequence number prediction model for evaluating the sequence number of each satellite image is as follows:
The image parameter of each satellite image is formed into a feature vector and is used as an input of an image sequence number prediction model, the image sequence number prediction model takes a display sequence number prediction value of each satellite image as an output, the display sequence number prediction value is a positive integer, the display sequence number of each satellite image is used as a prediction target, the square of the difference between the display sequence number prediction value and the display sequence number is used as a second prediction error corresponding to each satellite image, and the image sequence number prediction model takes the sum of the second prediction errors of all the minimized satellite images as a training target;
the method for screening the task selected image set from the task image set comprises the following steps:
inputting each satellite image in the task image set into a remote sensing image screening model, obtaining a selection tag predicted value output by the remote sensing image screening model, and storing the satellite image with the selection tag predicted value of 1 into the task selected image set;
the method for numbering the satellite images in the task selected image set is as follows:
And inputting each satellite image in the task selected image set into the image sequence number prediction model to obtain a display sequence number predicted value output by the image sequence number prediction model, and numbering each satellite image according to the display sequence number predicted value from small to large.
The embodiment 2 of the invention provides an automatic screening system based on remote sensing image content, which comprises a sample data collection module, a model training module and an image screening module, wherein the modules are electrically connected;
The sample data collection module is mainly used for collecting a plurality of regional image data sets, dividing each regional image data set into a selected data set and an unselected data set, collecting regional parameters corresponding to each regional image data set, collecting image parameters of each satellite image in each selected data set and unselected data set, and collecting a display sequence number of each satellite image in each regional cluster;
The model training module is mainly used for carrying out cluster analysis on regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters, for each regional cluster, the remote sensing image screening model for judging whether the satellite image is selected or not is trained by using the image parameters of each satellite image in a selected data set and an unselected data set, and for each regional cluster, the image sequence number prediction model for evaluating the sequencing sequence number of each satellite image is trained based on the image parameters and the display sequence number of each satellite image in each selected data set;
The image screening module is mainly used for collecting task area parameters and task image sets corresponding to the latest image screening task after the remote sensing image screening model and the image sequence number prediction model are trained, calculating to obtain area clusters corresponding to the task area parameters as task area clusters, screening task selected image sets from the task image sets by using the remote sensing image screening model corresponding to the task area clusters, numbering satellite images in the task selected image sets by using the image sequence number prediction model corresponding to the task area clusters, and displaying satellite images in the task selected image sets according to the numbering sequence.
An electronic device according to embodiment 3 of the present invention includes a processor and a memory, wherein the memory stores a computer program that is called by the processor;
The processor executes the automatic screening method based on the remote sensing image content by calling the computer program stored in the memory.
A computer-readable storage medium according to embodiment 4 of the present invention has stored thereon a computer program that is erasable;
when the computer program runs on the computer equipment, the computer equipment is caused to execute the automatic screening method based on the remote sensing image content.
Compared with the prior art, the invention has the beneficial effects that:
The invention collects the regional parameters corresponding to each regional image dataset and the regional parameters of each satellite image in each selected dataset and the unselected dataset by collecting a plurality of regional image datasets in advance, and collecting the regional parameters corresponding to each regional image dataset and the image parameters of each satellite image in each selected dataset and the unselected dataset, using a clustering algorithm to conduct clustering analysis on the regional parameters to obtain K regional cluster, K is the preset cluster number, for each regional cluster, using the regional parameters of each satellite image in the selected dataset and the unselected dataset, training out a remote sensing image screening model for judging whether the satellite image is selected, collecting the image parameters and the display sequence number of each satellite image in each selected dataset, training out an image sequence number prediction model for evaluating the sequencing sequence number of each satellite image based on the image parameters and the display sequence number of each satellite image in each selected dataset, after the remote sensing image screening model and the image sequence number prediction model are trained, collecting the task regional parameters and the task image set corresponding to the latest image screening task image, calculating to obtain the region cluster corresponding to the task regional parameters as the region cluster, using the cluster of the selected remote sensing image in the cluster to apply the cluster image sequence number of the task image screening task image, filtering model to select the remote sensing image by the screening image sequence number of the cluster image screening model according to the screening strategy, filtering the cluster sequence number of the cluster image of the cluster of the task image, filtering sequence number of the cluster is applied to the cluster image filtering sequence number of the cluster is selected by the filtering image filtering sequence of the cluster, the cluster is used in the filtering sequence of the cluster of the task image filtering sequence of the task image filtering model, therefore, automatic screening and arrangement of satellite images are realized, the labor cost of screening personnel is reduced, and the screening efficiency of satellite remote sensing images is improved.
Drawings
FIG. 1 is a flowchart of an automatic screening method based on remote sensing image content in embodiment 1 of the present invention;
fig. 2 is a block connection diagram of an automatic screening system based on remote sensing image content in embodiment 2 of the present invention;
fig. 3 is a schematic structural diagram of an electronic device in embodiment 3 of the present invention;
fig. 4 is a schematic diagram of a computer-readable storage medium according to embodiment 4 of the present invention.
Detailed Description
The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
As shown in fig. 1, the automatic screening method based on remote sensing image content comprises the following steps:
Collecting a plurality of regional image data sets, and dividing each regional image data set into a selected data set and an unselected data set;
collecting the regional parameters corresponding to each regional image data set, and collecting the image parameters of each satellite image in each selected data set and each unselected data set;
Performing cluster analysis on the regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters;
training a remote sensing image screening model for judging whether the satellite images are selected or not by using the image parameters of each satellite image in the selected data set and the unselected data set for each regional cluster;
collecting the display sequence number of each satellite image in each selected data set in each regional cluster;
Training an image sequence number prediction model for evaluating the sequencing sequence number of each satellite image based on the image parameters and the display sequence number of each satellite image in each selected data set for each regional cluster;
Step seven, after the remote sensing image screening model and the image sequence number prediction model are trained, collecting task area parameters and task image sets corresponding to the latest image screening task;
Step eight, calculating and obtaining a region cluster corresponding to the task region parameters as a task region cluster;
step nine, screening a task selected image set from the task image set by using a remote sensing image screening model corresponding to the task regional cluster;
numbering satellite images in the task selected image set by using an image sequence number prediction model corresponding to the task region cluster, and displaying the satellite images in the task selected image set according to the numbering sequence;
The method for collecting a plurality of regional image data sets and dividing each regional image data set into a selected data set and an unselected data set is as follows:
Collecting all satellite images corresponding to the screening when the satellite images are screened manually each time; all satellite images when manually screening satellite images are used as a group of regional image data sets each time;
It should be noted that, each time a person performs manual screening, a ground area, for example, XX city area is selected by the screening person, then a part of satellite images meeting expected rules are screened from all satellite images in the area, and the selected satellite images are ordered, because the expected rules of satellites shot in the ground area include, but are not limited to, whether the satellite images meet cloud cover, side view angle cover, resolution cover, ground object discrimination and the like, the selected satellite images can be ordered manually according to the expected rules, the cloud cover is set because cloud cover can cover the ground, thereby the ground in the satellite images is covered, the look is affected, the side view angle cover is set because the side view can affect the view covered by the satellite images, for example, when the side view angle is smaller, namely, when the ground is directly seen, the covered view is the most comprehensive, the resolution cover is set because the resolution can affect the definition of the satellite images, the ground object discrimination is set because various ground objects in the satellite images are needed, for example, the color difference between the ground objects can be difficult to cause on the ground object is small;
when each time of manual screening is carried out, partial satellite images screened by screening personnel form a selected data set corresponding to the regional image data set, and other satellite images except the selected data set in the regional image data set form an unselected data set;
further, the method for collecting the regional parameters corresponding to each regional image dataset is as follows:
for each regional image dataset, collecting the longitude median value, the latitude median value, the longitude span, the latitude span, the regional type and the regional area of the ground region corresponding to the regional image dataset together form regional parameters during manual screening;
The longitude median value and the latitude median value are respectively the central values of the longitude ranges covered by the ground areas, the latitude median value is the central value of the latitude ranges covered by the ground areas, and the area type can be the landform type of the ground areas or the area construction type;
Further, the method for collecting the image parameters of each satellite image in each selected data set and each unselected data set is as follows:
for each satellite image in each selected data set and each unselected data set, collecting the effective percentage, resolution, cloud cover, side view angle, shooting time difference and ground object resolution of a coverage selection area corresponding to the satellite image as image parameters;
the effective percentage calculation mode of the coverage selection area is as follows:
collecting a vector range (selection range) of a target area;
Converting the boundary of the satellite image into corresponding ground space coordinates by using an RPC model, wherein the converted ground space coordinates form an image coverage area;
Calculating the range of the image coverage area in the vector range of the target area as an image effective coverage selection area, calculating the proportion of the area of the image effective coverage selection area to the area of the target area, and taking the proportion as the effective percentage of the image coverage selection area;
the resolution and the side view angle are parameters directly calculated according to the position of the satellite in the orbit when the satellite is shot, and belong to the inherent attribute of the satellite image;
The shooting time difference is a difference value of the shooting time distance screening time;
the cloud amount calculating method is a conventional technical means in the field, for example, chinese patent with an authorized bulletin number of CN107529647B discloses a cloud amount calculating method based on a multilayer unsupervised sparse learning network, firstly, unsupervised layer-by-layer feature encoding is carried out on pictures by using a forward layer-by-layer sparse automatic encoder to obtain high-order semantic information, then, the cloud images are divided into thick cloud, thin cloud and clear sky by using the high-order semantic information, and finally, the total cloud amount in the cloud images is calculated by using a space correlation method;
and the ground object resolution is represented by using the variance of all pixel values in the satellite image;
further, the clustering algorithm is used for carrying out clustering analysis on the regional parameters, and the mode for obtaining K regional clustering clusters is as follows:
forming a one-dimensional array by corresponding region parameters of each region image dataset, and taking the one-dimensional array as coordinates of a data point;
Dividing K clustering clusters for all data points by using a clustering algorithm, wherein the clustering algorithm can be any one of K-means or FCM algorithm;
Further, for each regional cluster, the method for training the remote sensing image screening model for judging whether the satellite image is selected by using the image parameters of each satellite image in the selected data set and the unselected data set is as follows:
For each regional cluster:
Marking the selection label of each satellite image in the selected data set of all the regional image data sets in the regional cluster as 1, and marking the selection label of each satellite image in the unselected data set as 0;
The method comprises the steps of forming a feature vector by image parameters of each satellite image, taking the image parameters of each satellite image as an input of a remote sensing image screening model, taking a selection tag predicted value of each satellite image as an output, taking a selection tag predicted value of 0 or 1, taking a selection tag of each satellite image as a prediction target, taking the square of the difference between the selection tag predicted value and the selection tag as a first prediction error corresponding to each satellite image, minimizing the sum of the first prediction errors of all satellite images as a training target, training the remote sensing image screening model until the sum of the first prediction errors reaches convergence, stopping training, and training out whether the remote sensing image screening model is selected according to the image parameters of the satellite image, wherein the remote sensing image screening model is any one of a deep neural network model (DNN) or a Deep Belief Network (DBN) and the like;
further, in the collecting each regional cluster, the manner of collecting the display sequence number of each satellite image in each selected dataset is as follows:
When screening personnel manually screen the satellite images each time, collecting serial numbers which are sequenced by the screening personnel for the selected satellite images as display serial numbers for each satellite image in the selected data set; it can be understood that the display serial number shows the recognition degree of screening personnel on satellite image quality to a certain extent, and only the satellite image with the best quality can be ranked first, so that the display serial number can also be used for expressing satellite image quality;
Further, for each regional cluster, based on the image parameter and the display sequence number of each satellite image in each selected dataset, the image sequence number prediction model for evaluating the sequencing sequence number of each satellite image is trained in the following manner:
The method comprises the steps of forming a feature vector by image parameters of each satellite image, taking the image parameters of each satellite image as an input of an image sequence number prediction model, taking a display sequence number prediction value of each satellite image as an output, taking the display sequence number of each satellite image as a positive integer, taking the square of the difference between the display sequence number prediction value and the display sequence number as a second prediction error corresponding to each satellite image, taking the sum of the second prediction errors of all the satellite images as a training target, training the image sequence number prediction model until the sum of the second prediction errors reaches convergence, stopping training, and training out an image sequence number prediction model for predicting the display sequence number of the satellite image according to the image parameters of the satellite image, wherein the remote sensing image screening model is any one of a polynomial regression model, a support vector machine regression model (SVR) and the like;
Further, the task area parameters are area parameters of a ground area where the satellite images are required to be screened by the latest image screening task;
The task image set is all satellite images which need to be screened by the latest image screening task;
The mode of calculating the area cluster corresponding to the task area parameter as the task area cluster is as follows:
calculating the distance between the data point formed by the task area parameters and the center point of each area cluster, and taking the area cluster closest to the data point as the task area cluster;
the method for screening the task selected image set from the task image set comprises the following steps:
inputting each satellite image in the task image set into a remote sensing image screening model, obtaining a selection tag predicted value output by the remote sensing image screening model, and storing the satellite image with the selection tag predicted value of 1 into the task selected image set;
the method for numbering the satellite images in the task selected image set is as follows:
And inputting each satellite image in the task selected image set into the image sequence number prediction model to obtain a display sequence number predicted value output by the image sequence number prediction model, and numbering each satellite image according to the display sequence number predicted value from small to large.
Example 2
As shown in FIG. 2, the automatic screening system based on remote sensing image content comprises a sample data collection module, a model training module and an image screening module, wherein the modules are electrically connected;
The sample data collection module is mainly used for collecting a plurality of regional image data sets, dividing each regional image data set into a selected data set and an unselected data set, collecting regional parameters corresponding to each regional image data set, collecting image parameters of each satellite image in each selected data set and unselected data set, and collecting a display sequence number of each satellite image in each regional cluster;
The model training module is mainly used for carrying out cluster analysis on regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters, for each regional cluster, the remote sensing image screening model for judging whether the satellite image is selected or not is trained by using the image parameters of each satellite image in a selected data set and an unselected data set, and for each regional cluster, the image sequence number prediction model for evaluating the sequencing sequence number of each satellite image is trained based on the image parameters and the display sequence number of each satellite image in each selected data set;
The image screening module is mainly used for collecting task area parameters and task image sets corresponding to the latest image screening task after the remote sensing image screening model and the image sequence number prediction model are trained, calculating to obtain area clusters corresponding to the task area parameters as task area clusters, screening task selected image sets from the task image sets by using the remote sensing image screening model corresponding to the task area clusters, numbering satellite images in the task selected image sets by using the image sequence number prediction model corresponding to the task area clusters, and displaying satellite images in the task selected image sets according to the numbering sequence.
Example 3
Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 3, there is also provided an electronic device 100 according to yet another aspect of the present application. The electronic device 100 may include one or more processors, which may include a CPU102 and a GPU109, and one or more memories. Wherein the memory has stored therein computer readable code which, when executed by the one or more processors, can perform the remote sensing image content based autoscreening method as described above.
The method or system according to embodiments of the application may also be implemented by means of the architecture of the electronic device shown in fig. 3. As shown in fig. 3, the electronic device 100 may include a bus 101, one or more CPUs 102, a Read Only Memory (ROM) 103, a Random Access Memory (RAM) 104, a communication port 105 connected to a network, an input/output component 106, a hard disk 107, a GPU109, and the like. The storage device in the electronic device 100, such as the ROM103 or the hard disk 107, may store the remote sensing image content-based automatic screening method provided by the present application. The automatic screening method based on remote sensing image content can comprise the following steps of collecting a plurality of regional image data sets, dividing each regional image data set into a selected data set and an unselected data set, collecting regional parameters corresponding to each regional image data set, collecting image parameters of each satellite image in each selected data set and each unselected data set, carrying out clustering analysis on the regional parameters by using a clustering algorithm to obtain K regional cluster, K is the number of preset cluster, step four, training a remote sensing image screening model for judging whether a satellite image is selected by using the image parameters of each satellite image in the selected data set and the unselected data set, step five, collecting the display sequence number of each satellite image in each regional cluster, step six, for each regional cluster, training an image sequence number prediction model for evaluating each satellite image based on the image parameters and the display sequence number of each satellite image in each selected data set, step seven, carrying out clustering task number prediction on the task image corresponding to the task image in the cluster, step nine, and step nine, namely, using the cluster number of the image to be used as a task number prediction task image in the corresponding to the task image in the clustering task image, and step nine, and carrying out the task number prediction task number setting corresponding to the task image in the clustering task image screening region, displaying satellite images in the task selected image set according to the numbering sequence;
further, the electronic device 100 may also include a user interface 108. Of course, the architecture shown in fig. 3 is merely exemplary, and one or more components of the electronic device shown in fig. 3 may be omitted as may be practical in implementing different devices.
Example 4
FIG. 4 is a schematic diagram of a computer-readable storage medium according to one embodiment of the present application. As shown in fig. 4, is a computer-readable storage medium 200 according to one embodiment of the application. The computer-readable storage medium 200 has stored thereon computer-readable instructions. When the computer readable instructions are executed by the processor, the automatic screening method based on remote sensing image content according to the embodiment of the present application described with reference to the above drawings may be performed. Computer-readable storage medium 200 includes, but is not limited to, for example, volatile memory and/or nonvolatile memory. Volatile memory can include, for example, random Access Memory (RAM), cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.
In addition, according to embodiments of the present application, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions executable by a processor to perform instructions corresponding to the method steps provided by the present application, which when executed by a Central Processing Unit (CPU), perform the functions defined above in the method of the present application.
The methods and apparatus, devices of the present application may be implemented in numerous ways. For example, the methods and apparatus, devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, firmware. The above-described sequence of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the sequence specifically described above unless specifically stated otherwise. Furthermore, in some embodiments, the present application may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present application. Thus, the present application also covers a recording medium storing a program for executing the method according to the present application.
The above preset parameters or preset thresholds are set by those skilled in the art according to actual conditions or are obtained by mass data simulation.
The above embodiments are only for illustrating the technical method of the present invention and not for limiting the same, and it should be understood by those skilled in the art that the technical method of the present invention may be modified or substituted without departing from the spirit and scope of the technical method of the present invention.
Claims (10)
1. The automatic screening method based on the remote sensing image content is characterized by comprising the following steps of:
collecting a plurality of regional image data sets, and dividing each regional image data set into a selected data set and an unselected data set;
Collecting the regional parameters corresponding to each regional image dataset, and collecting the image parameters of each satellite image in each selected dataset and each unselected dataset;
Carrying out cluster analysis on the regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters;
For each regional cluster, training a remote sensing image screening model for judging whether the satellite image is selected or not by using image parameters of each satellite image in the selected data set and the unselected data set;
collecting the display sequence number of each satellite image in each selected data set in each regional cluster;
for each regional cluster, training an image sequence number prediction model for evaluating the sequencing sequence number of each satellite image based on the image parameters and the display sequence number of each satellite image in each selected data set;
After the remote sensing image screening model and the image sequence number prediction model are trained, collecting task area parameters and task image sets corresponding to the latest image screening task;
Calculating to obtain a region cluster corresponding to the task region parameter as a task region cluster;
screening a task selected image set from the task image set by using a remote sensing image screening model corresponding to the task region cluster;
and numbering the satellite images in the task selected image set by using an image sequence number prediction model corresponding to the task region cluster, and displaying the satellite images in the task selected image set according to the numbering sequence.
2. The method of claim 1, wherein the steps of collecting a plurality of regional image datasets and dividing each regional image dataset into a selected dataset and an unselected dataset are as follows:
Collecting all satellite images corresponding to the screening when the satellite images are screened manually each time; all satellite images when manually screening satellite images are used as a group of regional image data sets each time;
And when manual screening is performed each time, the partial satellite images screened by the screening personnel form a selected data set corresponding to the regional image data set, and other satellite images except the selected data set in the regional image data set form an unselected data set.
3. The method for automatically screening remote sensing image content according to claim 2, wherein the collecting the regional parameters corresponding to each regional image dataset is as follows:
For each regional image dataset, collecting the longitude median value, the latitude median value, the longitude span, the latitude span, the regional type and the regional area of the ground region corresponding to the regional image dataset together form regional parameters during manual screening.
4. The method of claim 3, wherein the collecting image parameters of each satellite image in each selected dataset and each unselected dataset is as follows:
And collecting the effective percentage, resolution, cloud cover, side view angle, shooting time difference and ground object resolution of the coverage selection area corresponding to each satellite image as image parameters for each satellite image in each selected data set and each unselected data set.
5. The method for automatically screening remote sensing image content according to claim 4, wherein the training of the remote sensing image screening model for determining whether the satellite image is selected is:
For each regional cluster:
Marking the selection label of each satellite image in the selected data set of all the regional image data sets in the regional cluster as 1, and marking the selection label of each satellite image in the unselected data set as 0;
The method comprises the steps of forming a feature vector by image parameters of each satellite image, taking the feature vector as input of a remote sensing image screening model, taking a selection tag predicted value of each satellite image as output, taking the selection tag predicted value as 0 or 1, taking the selection tag of each satellite image as a predicted target, taking the square of the difference between the selection tag predicted value and the selection tag as a first predicted error corresponding to each satellite image, minimizing the sum of the first predicted errors of all satellite images as a training target, and training the remote sensing image screening model until the sum of the first predicted errors reaches convergence.
6. The method for automatically screening remote sensing image content according to claim 5, wherein the training of the image sequence number prediction model for evaluating the sequence number of each satellite image is as follows:
The image parameter of each satellite image is formed into a feature vector and is used as an input of an image sequence number prediction model, the image sequence number prediction model takes a display sequence number prediction value of each satellite image as an output, the display sequence number prediction value is a positive integer, the display sequence number of each satellite image is used as a prediction target, the square of the difference between the display sequence number prediction value and the display sequence number is used as a second prediction error corresponding to each satellite image, the image sequence number prediction model takes the sum of the second prediction errors of all the minimized satellite images as a training target, and the image sequence number prediction model is trained until the sum of the second prediction errors reaches convergence.
7. The automatic screening method based on remote sensing image contents according to claim 6, wherein the method for screening the task selected image set from the task image set is as follows:
inputting each satellite image in the task image set into a remote sensing image screening model, obtaining a selection tag predicted value output by the remote sensing image screening model, and storing the satellite image with the selection tag predicted value of 1 into the task selected image set;
the method for numbering the satellite images in the task selected image set is as follows:
And inputting each satellite image in the task selected image set into the image sequence number prediction model to obtain a display sequence number predicted value output by the image sequence number prediction model, and numbering each satellite image according to the display sequence number predicted value from small to large.
8. An automatic screening system based on remote sensing image content, which is used for realizing the automatic screening method based on remote sensing image content according to any one of claims 1-7, and is characterized by comprising a sample data collection module, a model training module and an image screening module, wherein the modules are electrically connected;
The sample data collection module is used for collecting a plurality of regional image data sets, dividing each regional image data set into a selected data set and an unselected data set, collecting regional parameters corresponding to each regional image data set, collecting image parameters of each satellite image in each selected data set and unselected data set, and collecting a display sequence number of each satellite image in each regional cluster;
the model training module is used for carrying out cluster analysis on regional parameters by using a clustering algorithm to obtain K regional cluster clusters, wherein K is the number of preset cluster clusters, for each regional cluster, the remote sensing image screening model for judging whether the satellite image is selected or not is trained by using the image parameters of each satellite image in the selected data set and the unselected data set, and for each regional cluster, the image sequence number prediction model for evaluating the sequencing sequence number of each satellite image is trained based on the image parameters and the display sequence number of each satellite image in each selected data set;
The image screening module is used for collecting task area parameters and task image sets corresponding to the latest image screening task after the remote sensing image screening model and the image sequence number prediction model are trained, calculating to obtain area cluster corresponding to the task area parameters as task area cluster, screening task selected image sets from the task image sets by using the remote sensing image screening model corresponding to the task area cluster, numbering satellite images in the task selected image sets by using the image sequence number prediction model corresponding to the task area cluster, and displaying satellite images in the task selected image sets according to the numbering sequence.
9. An electronic device comprising a processor and a memory, wherein,
The memory stores a computer program which can be called by the processor;
The processor executes the remote sensing image content-based automatic screening method according to any one of claims 1 to 7 in the background by calling a computer program stored in the memory.
10. A computer readable storage medium having stored thereon a computer program that is erasable;
the computer program, when run on a computer device, causes the computer device to perform implementing the remote sensing image content-based autoscreening method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410233861.8A CN118691536B (en) | 2024-03-01 | 2024-03-01 | Automatic screening method and system based on remote sensing image content |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410233861.8A CN118691536B (en) | 2024-03-01 | 2024-03-01 | Automatic screening method and system based on remote sensing image content |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118691536A CN118691536A (en) | 2024-09-24 |
| CN118691536B true CN118691536B (en) | 2025-03-21 |
Family
ID=92770772
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410233861.8A Active CN118691536B (en) | 2024-03-01 | 2024-03-01 | Automatic screening method and system based on remote sensing image content |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118691536B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120045733B (en) * | 2025-04-23 | 2025-08-26 | 长光卫星技术股份有限公司 | A remote sensing data coverage screening method based on benefit value |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113327259A (en) * | 2021-08-04 | 2021-08-31 | 中国科学院空天信息创新研究院 | Remote sensing data screening method and system for area coverage |
| CN115223054A (en) * | 2022-07-15 | 2022-10-21 | 国家林业和草原局西南调查规划院 | Remote sensing image change detection method based on partition clustering and convolution |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116958810A (en) * | 2023-06-27 | 2023-10-27 | 中国四维测绘技术有限公司 | Automatic screening method and system for remote sensing image vegetation index based on deep learning |
-
2024
- 2024-03-01 CN CN202410233861.8A patent/CN118691536B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113327259A (en) * | 2021-08-04 | 2021-08-31 | 中国科学院空天信息创新研究院 | Remote sensing data screening method and system for area coverage |
| CN115223054A (en) * | 2022-07-15 | 2022-10-21 | 国家林业和草原局西南调查规划院 | Remote sensing image change detection method based on partition clustering and convolution |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118691536A (en) | 2024-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pally et al. | Application of image processing and convolutional neural networks for flood image classification and semantic segmentation | |
| CN114897779B (en) | Abnormal area positioning method and device for cervical cytology images based on fusion attention | |
| US11308714B1 (en) | Artificial intelligence system for identifying and assessing attributes of a property shown in aerial imagery | |
| CN113392775A (en) | Sugarcane seedling automatic identification and counting method based on deep neural network | |
| Li et al. | A comparison of deep learning methods for airborne lidar point clouds classification | |
| CN113486819A (en) | A Ship Target Detection Method Based on YOLOv4 Algorithm | |
| CN114463624B (en) | Method and device for detecting illegal building applied to urban management supervision | |
| CN111008956B (en) | Beam bottom crack detection method, system, device and medium based on image processing | |
| CN113160150A (en) | AI (Artificial intelligence) detection method and device for invasion of foreign matters in wire network based on multi-type sample fusion and multi-complex network | |
| CN116258956B (en) | Methods, equipment, storage media and devices for tree identification by drones | |
| Ruiz-Lendínez et al. | Deep learning methods applied to digital elevation models: state of the art | |
| CN116433596B (en) | Slope vegetation coverage measuring method and device and related components | |
| CN115953371A (en) | Insulator defect detection method, device, equipment and storage medium | |
| CN118691536B (en) | Automatic screening method and system based on remote sensing image content | |
| CN118097348B (en) | Unmanned ship recognition system based on improved YOLOv7 algorithm | |
| CN116310321B (en) | Target detection method in high-resolution aerial image | |
| KR20140053259A (en) | Techniques for feature extraction | |
| CN120126016A (en) | Mangrove ecological information extraction method and system based on drone | |
| CN116229419B (en) | Pedestrian detection method and device | |
| CN116091784B (en) | Target tracking method, device and storage medium | |
| CN116612382A (en) | Urban remote sensing image target detection method and device | |
| Nikitina | Automatic segmentation of tree crowns in pine forests using Mask R-CNN on RGB imagery from UAVs | |
| CN116844023B (en) | Marine small-scale target detection method and device | |
| CN120298825B (en) | Map resource identification method based on deep learning | |
| CN113076877B (en) | Remote sensing image target detection method, system and medium based on ground sampling distance |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |