CN111144487A - Method for establishing and updating remote sensing image sample library - Google Patents
Method for establishing and updating remote sensing image sample library Download PDFInfo
- Publication number
- CN111144487A CN111144487A CN201911372783.5A CN201911372783A CN111144487A CN 111144487 A CN111144487 A CN 111144487A CN 201911372783 A CN201911372783 A CN 201911372783A CN 111144487 A CN111144487 A CN 111144487A
- Authority
- CN
- China
- Prior art keywords
- sample
- remote sensing
- sensing image
- data
- sample data
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000012549 training Methods 0.000 claims abstract description 67
- 238000012545 processing Methods 0.000 claims abstract description 38
- 238000005457 optimization Methods 0.000 claims abstract description 26
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 238000002372 labelling Methods 0.000 claims description 24
- 238000013135 deep learning Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 8
- 238000010586 diagram Methods 0.000 claims description 7
- 230000007797 corrosion Effects 0.000 claims description 6
- 238000005260 corrosion Methods 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 6
- 238000013136 deep learning model Methods 0.000 claims description 6
- 238000009432 framing Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 6
- 230000010339 dilation Effects 0.000 claims description 5
- 230000003628 erosive effect Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000003384 imaging method Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000000265 homogenisation Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for establishing and updating a remote sensing image sample library, which comprises the following steps: when an operation instruction for establishing a remote sensing image sample library is received, establishing the remote sensing image sample library according to a first preset rule; when an instruction extracted from a remote sensing image sample database is received, judging whether sample data meeting a first preset condition exists in the remote sensing image sample database or not according to an extraction requirement; under the condition that sample data meeting a first preset condition does not exist in the sample library, receiving a mark of a first object in the remote sensing image data, and extracting a second object in the remote sensing image data by using an iterative training and preset method according to the mark so as to update the sample library; and under the condition that sample data meeting a first preset condition exists in the sample library, extracting the sample data in the sample library, performing model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a method for establishing and updating a remote sensing image sample library.
Background
In recent years, with the rapid development of Deep learning technology, great progress is made particularly in the field of remote sensing image information extraction and the like, and a Deep Neural Network (DNN) corresponding to Deep learning is a prerequisite for the network to be trained sufficiently, and a large number of samples are required as training data when the Deep Neural network is trained.
The sample labeling is an important part of training data in deep learning, and when the remote sensing image is labeled, various ground feature samples in the remote sensing image need to be labeled and are integrated to form a sample library. The sample library stores the labeling vector data and the base-period image files of various ground features, and corresponding data can be called from the sample library according to the sample types when the DNN training device is used, so that DNN training can be conveniently carried out.
However, in the related art, the labeling of the sample usually depends on manual labeling, a large amount of manpower and material resources are consumed, the labeling efficiency is low, and the effect is not ideal; the traditional sample organization management mode does not use a spatial database, and is inconvenient for sample data management and use.
Disclosure of Invention
In view of the above, the invention provides a method for establishing and updating a remote sensing image sample library, so as to solve the problems that in the related art, samples in remote sensing image data are artificially labeled, the labeling efficiency is low, and the effect is not ideal.
In order to achieve the above object, according to an aspect of the present invention, there is provided a method for establishing and updating a remote sensing image sample library, including:
(1) when an operation instruction for establishing a remote sensing image sample library is received, establishing the remote sensing image sample library according to a first preset rule;
(2) when an instruction extracted from the remote sensing image sample database is received, judging whether sample data meeting a first preset condition exists in the sample database according to an extraction requirement;
(3) under the condition that sample data meeting a first preset condition does not exist in the remote sensing image sample library, receiving a mark of a first object in remote sensing image data, and extracting a second object in the remote sensing image data by using an iterative training and preset method according to the mark so as to update the remote sensing image sample library;
(4) under the condition that sample data meeting a first preset condition exists in the remote sensing image sample library, extracting the sample data in the remote sensing image sample library, carrying out model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library.
In an optional manner, the creating the remote sensing image sample library according to a first preset rule includes:
(S1) storing the remote sensing image sample library as a spatial database;
(S2) designing a data table structure in the spatial database to be: image ID, image source, image geographic location, number of image bands, image resolution, imaging time, sample type, etc.;
the first preset rule is that sample data is stored according to a preset framing format or a first preset size data block according to requirements, the sample data stored according to standard framing is the sample data produced according to the requirements of deep learning application, and the sample data stored in the first preset size data block is directly used for training.
In an optional manner, the determining whether sample data meeting a first preset condition exists in the sample library according to the extraction requirement includes:
judging whether sample data meeting the training requirement of the deep learning model exists in a sample library or not according to the sample type; the first preset condition is whether the quantity and the quality of the sample data in the sample library meet the deep learning model training requirement or not.
In an optional mode, receiving a label of a first object in the remote sensing image data, and extracting a second object in the remote sensing image data by using an iterative training and preset method according to the label; the method comprises the following steps:
(T1) acquiring the preprocessed remote sensing image data;
(T2) marking a target object in the target remote sensing image data in the remote sensing image data to obtain sample data;
(T3) performing model training according to the sample data to obtain a primary prediction model;
(T4) performing prediction and optimization processing on remote sensing image data other than the target remote sensing image data one by one, and training a preliminary prediction model on prediction sample data satisfying a second preset condition and the corresponding remote sensing image data;
(T5) continuously repeating the step T4 until a third preset condition is met, and obtaining an optimized prediction model;
(T6) predicting the remote sensing image by using the trained optimized prediction model to obtain sample marking data;
the first object is labeled sample data, and the second object is unlabeled sample data except the first object.
In an alternative, the pre-processing comprises at least one of: orthorectification, image fusion, bit depth transformation, image homogenization and image mosaicing.
In an optional manner, the predicting and optimizing process includes training a primary prediction model on prediction labeling data and corresponding remote sensing image data that satisfy a second preset condition, and includes:
(M1) constructing a type rule set of the prediction sample data according to attribute information of the prediction sample data, the attribute information including at least one of: the remote sensing index, the sample shape, the structure, the texture and the sample space topological relation of the remote sensing image data corresponding to the prediction sample data;
(M2) calculating a confidence level for the prediction sample data in dependence on the rule set;
(M3) screening the prediction sample data according to the confidence.
In an optional mode, the constructing a type rule set of the prediction sample data according to the attribute information of the prediction sample data further includes: under the condition that the target object is a first-class object, receiving first processing on the outline of the target object, and deleting discrete points and holes in the prediction sample data; or, receiving second processing on the contour of the target object when the target object is the second class object.
In an alternative, the first processing includes: performing contour corrosion expansion operation and contour smoothing operation; the second processing includes: a contour erosion dilation operation and a contour optimization operation.
In an alternative, the contour erosion dilation operation comprises: carrying out convolution processing on the prediction sample data by using an operator with a characteristic size to complete the open operation of the prediction sample; the contour smoothing operation includes: and disconnecting the connection of the contours corresponding to the prediction samples, wherein the connection is less than or equal to a first preset threshold value, and deleting the protruding parts of the contours corresponding to the prediction samples, wherein the protruding degrees of the protruding parts are less than a second preset threshold value.
In an alternative, the method further comprises: and under the condition that a land use current situation vector diagram exists in the geographic area, adding the land use current situation vector diagram and corresponding remote sensing image data into a sample library.
In an optional mode, the current land utilization vector diagram is a land utilization vector diagram within a preset time period, and the preset time period is a time period which is located before the moment of acquiring the data of the remote sensing image and is within a preset interval with the moment interval of acquiring the remote sensing image.
In an alternative, the extracting sample data in the sample library includes:
under the condition that the sample data is first-class labeled sample data, cutting the remote sensing image data corresponding to the labeled sample data according to the geographic space position of the remote sensing image data corresponding to the labeled sample data to generate remote sensing image blocks with the same size; the first type of marking sample data is marking sample data of a target object corresponding to the sample data, which is distributed in the remote sensing image data and is less than or equal to a third preset threshold value; or,
under the condition that the sample data is second-class labeled sample data, cutting the remote sensing image data corresponding to the first labeled sample data from left to right and from top to bottom to generate remote sensing image blocks with the same size; and the second type of marking sample data is marking sample data of a target object corresponding to the marking sample data, which is distributed in the remote sensing image data and is larger than or equal to a fourth preset threshold value.
In an optional manner, the extracting sample data in the remote sensing image sample library, performing model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library includes:
(N1) performing model training according to the sample data to obtain a prediction model;
(N2) carrying out prediction and optimization processing on the unmarked remote sensing image by using the prediction model to obtain prediction sample data meeting a fourth preset condition;
(N3) updating sample data of which the precision meets the preset precision in the prediction sample data into a sample library.
The method for establishing and updating the remote sensing image sample library solves the problems of low marking efficiency, inaccurate marking result and unsatisfactory effect of manually marking the remote sensing image data in the related technology; the accuracy and the efficiency of marking the remote sensing image data are improved, and the sample manufacturing and management efficiency is improved.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings.
Fig. 1 is a flowchart illustrating an implementation of a method for establishing and updating a remote sensing image sample library according to an embodiment of the present disclosure;
fig. 2 is a flowchart illustrating an implementation of a method for creating and updating a remote sensing image sample library according to another embodiment of the present application;
fig. 3 is a flowchart illustrating an implementation of a method for creating and updating a remote sensing image sample library according to another embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the embodiments of the present invention clearer, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
In the description of the embodiments of the present invention, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Fig. 1 is a flowchart illustrating an implementation of a method for establishing and updating a remote sensing image sample library according to an embodiment of the present application.
Referring to fig. 1, the method for establishing and updating a remote sensing image sample library provided in the embodiment of the present application may be specifically applied to an intelligent terminal such as a desktop computer, a notebook computer, a mainframe computer, and a tablet computer; the method comprises the following steps:
Specifically, in the embodiment of the application, a remote sensing image sample library is stored as a spatial database; designing a data table structure in the spatial database as follows: image ID, image source, image geographic location, number of image bands, image resolution, imaging time, sample type, etc.; the first preset rule is that sample data is stored according to a preset framing format or a preset size data block according to requirements, the sample data can be stored according to standard framing, sample data with any size and format can be produced according to the requirements of deep learning application, the sample data stored in the data block with a specific size can be trained by directly using the sample data, and the sample data does not need to be cut.
And 102, judging whether sample data meeting a first preset condition exists in the sample database according to an extraction requirement when an extraction instruction from the remote sensing image sample database is received.
Specifically, in the embodiment of the application, whether sample data meeting the deep learning model training requirement exists in the sample library is judged according to the sample type; the first preset condition is whether the quantity and the quality of sample data in the sample library meet the deep learning model training requirement or not.
103, receiving a label of a first object in the remote sensing image data under the condition that no sample data meeting a first preset condition exists in the remote sensing image sample library, and extracting a second object in the remote sensing image data by using an iterative training and presetting method according to the label to update the remote sensing image sample library.
Specifically, in the embodiment of the application, the remote sensing image data after preprocessing is acquired. In the embodiment of the application, receiving manual labeling of a target object in target remote sensing image data in first remote sensing image data, in some specific examples, the manual labeling means that a spatial position where an interested ground object is located is sketched out from the first remote sensing image data through professional labeling software (for example, ARCGIS, MAPGIS and the like) and is stored in a vector mark file; specifically, in the embodiment of the present application, the target object in the remote sensing image data may be a specific ground object in the remote sensing image data; for example, in some possible application scenarios, the target object may be an artificial feature such as a greenhouse, a windmill, an oil tank pipe, a golf course, or a natural feature such as a mountain, a wetland, or a lake in the first remote sensing image data. Specifically, in the embodiment of the present application, the target remote sensing image data in the remote sensing image data may be a small portion of the remote sensing image data in the remote sensing image data.
Specifically, in the embodiment of the application, new sample data is extracted by using an iterative training and preset method, and model training is performed according to the sample data to obtain a primary prediction model; performing prediction optimization processing on remote sensing image data except the target remote sensing image data one by one, and training a primary prediction model on prediction sample data meeting a second preset condition and the corresponding remote sensing image data; continuously repeating the steps until a third preset condition is met to obtain an optimized prediction model; and predicting the remote sensing image by using the trained optimized prediction model to obtain sample marking data.
Specifically, in the embodiment of the application, model training is performed according to sample data, and different models can be selected for training according to different ground feature types; for example, in some specific examples, a Deep Supervised Significant (DSS) model or an RCF model may be selected for the extraction of road, water surface, building, etc. samples, and in other possible examples, a Yolo model may be used for windmills, oil pipes, etc.
Specifically, in the embodiment of the present application, the annotation data and the corresponding remote sensing image data may be stored in a remote sensing image sample library. Specifically, in the embodiment of the present application, the labeled sample data is vector labeled sample data stored in a vector format.
And 104, under the condition that sample data meeting a first preset condition exists in the remote sensing image sample library, extracting the sample data in the remote sensing image sample library, performing model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library.
Specifically, in the embodiment of the application, the remote sensing image data after preprocessing is acquired. Specifically, in the embodiment of the application, when the sample data is the first-class labeled sample data, the remote sensing image data corresponding to the labeled sample data is cut according to the geographic spatial position of the remote sensing image data corresponding to the labeled sample data, and remote sensing image blocks with the same size are generated; the class marking sample data is marking sample data of a target object corresponding to the sample data, which is distributed in the remote sensing image data and is less than or equal to a third preset threshold value; or, under the condition that the sample data is the second type of labeled sample data, cutting the remote sensing image data corresponding to the labeled sample data from left to right and from top to bottom to generate remote sensing image blocks with the same size; and the second type of marking sample data is marking sample data of which the distribution of the target object corresponding to the marking sample data in the remote sensing image data is greater than or equal to a fourth preset threshold value.
In some specific examples, the first type of labeling sample data may be sparse sample data, and specifically, the sparse sample data is sparsely distributed sample data in the remote sensing image data, such as less distributed ground objects like a windmill, an oil pipe, a golf course, and the like.
In some specific examples, the second type of labeling sample data may be non-sparse sample data, and specifically, the non-sparse sample data is sample data densely distributed in the remote sensing image data, such as features of roads, buildings, farmlands, and the like.
Specifically, in the embodiment of the present application, the training of the machine learning model, the prediction and optimization of the remote sensing image data are performed, and the updating of the sample data into the sample library includes: performing model training according to the sample data to obtain a prediction model; performing prediction optimization processing on the unmarked remote sensing image by using the prediction model to obtain prediction sample data meeting a fourth preset condition; and updating the incremental sample data which accords with the second preset rule into a sample library.
Specifically, in the embodiment of the application, the marking data meeting the precision and the corresponding remote sensing image data can be updated to the remote sensing image sample database.
The method for establishing and updating the remote sensing image sample library provided by the embodiment of the application is used for storing sample data by establishing the remote sensing image sample library; judging whether the sample data in the sample library meets a preset condition or not according to the requirement of model training, if not, obtaining the sample data by marking a small amount of the target object in the obtained remote sensing image data, extracting new sample data by using an iterative training and preset method, and updating the sample library; and if the preset conditions are met, extracting the sample data in the sample library, performing machine learning model training, remote sensing image data prediction and optimization processing, and updating the sample data into the sample library. The problems of low labeling efficiency, inaccurate labeling result and unsatisfactory effect of manually labeling the remote sensing image data in the related technology are solved; the accuracy and efficiency of marking the remote sensing image data are improved, the efficiency of establishing a sample library is improved, and the labor cost is saved; the method facilitates the management of the sample database, and improves the efficiency of extracting the marked sample vector data and the remote sensing image data from the sample database
Fig. 2 is a flowchart illustrating an implementation of a method for creating and updating a remote sensing image sample library according to another embodiment of the present application.
Based on the foregoing embodiment, referring to fig. 2, another embodiment of the present application provides a method for establishing and updating a remote sensing image sample library, including the following steps:
Specifically, in the embodiment of the present application, the sample data in the sample library does not meet the requirement of model training, and may be that there is no sample data of this type in the sample library or that the quantity and quality of the sample data do not meet the requirement.
And 202, obtaining a small amount of labels of target objects in the remote sensing image data to obtain sample data.
Specifically, in the embodiment of the present application, the preprocessing includes at least one of the following processes: orthorectification, image fusion, bit depth transformation, image homogenization and image mosaicing.
Specifically, in the embodiment of the application, a target object is taken as a greenhouse and is used as a specific example for explanation, a small amount of labeled greenhouse sample data is used for deep learning network training, and attention is paid to guarantee accuracy when the samples are labeled. And cutting the remote sensing image data and the greenhouse vector mark data into 800 × 800 training data, marking the corresponding position of the greenhouse area in the training label as 255, and setting the value of the surrounding area as 0. And amplifying the training data by using methods such as rotation, translation, mirror image and the like, and increasing sample data.
And 203, performing model training according to the sample data to obtain a primary prediction model.
Specifically, in the embodiment of the present application, a primary prediction model of the greenhouse is obtained according to the vector labeled data files of a small number of greenhouses labeled in step 202 and the corresponding remote sensing image data training models. Specifically, in the embodiment of the present application, a DSS deep learning network is used for extracting the remote sensing image greenhouse, and the DSS model is improved on the VGGNet model, so that several short connections from the output of the deeper side to the output of the shallower side are added. In this way, higher level functions can help lower the side outputs to better locate significant areas, while lower level functions can help enrich the higher level side outputs and finer details. And inputting the manually marked initial greenhouse sample data into a DSS model for greenhouse feature extraction training, and learning and training to obtain a primary prediction model of the greenhouse.
And 204, performing prediction optimization processing on the remote sensing image data except the target remote sensing image data one by one, and training the prediction sample data meeting a second preset condition and the corresponding remote sensing image data into a primary prediction model.
Specifically, in the embodiment of the present application, the currently processed remote sensing image data may be any remote sensing image data other than the target remote sensing image data; after the remote sensing image data which is processed currently is subjected to prediction processing, whether the labeled data obtained by performing prediction labeling on the remote sensing image data which is processed currently through the primary prediction model meets a first preset condition or not is judged, and the primary prediction model is trained through the predicted labeled data which meets the first preset condition and the corresponding remote sensing image data. Specifically, in this embodiment of the application, the second preset condition may be a condition for screening the prediction tagging data, for example, in some specific examples, the prediction tagging result of the primary prediction model on the remote sensing image data may be inaccurate, prediction tagging data with inaccurate prediction needs to be removed, and the prediction tagging data with more accurate prediction is used as input data for model training to continue training the prediction model, so as to improve the accuracy of the prediction model on processing the remote sensing image data. Specifically, in this embodiment of the present application, the second preset condition may be set according to an actual application requirement of a user, and in this embodiment of the present application, the second preset condition is not specifically limited.
Specifically, in the embodiment of the present application, training the primary prediction model with the prediction sample data when the prediction sample data meets the second preset condition includes:
(1) according to the attribute information of the prediction sample data, constructing a type rule set of the prediction sample data, wherein the attribute information comprises at least one of the following information: predicting the remote sensing index, the sample shape, the structure, the texture and the sample space topological relation of the remote sensing image data corresponding to the sample data;
(2) calculating the confidence coefficient of the predicted sample data according to the rule set;
(3) and (4) screening prediction sample data by confidence degree.
In some optional manners, before constructing the type rule set of the prediction sample data according to the attribute information of the prediction sample data, the method further includes:
under the condition that the target object is a first-class object, receiving first processing on the outline of the target object, and deleting discrete points and cavities in remote sensing image data corresponding to prediction sample data; alternatively, in the case where the target object is an object of a second type, second processing of the contour of the target object is received.
In some alternatives, the first processing includes: performing contour corrosion expansion operation and contour smoothing operation; the second process includes: performing contour corrosion expansion operation and contour optimization operation;
wherein the contour erosion dilation operation comprises: and (4) carrying out convolution processing on the prediction sample data by using an operator with a characteristic size to complete the open operation of the prediction sample. Carrying out convolution processing on remote sensing image data corresponding to the prediction sample data to obtain a local maximum value and a local minimum value of the remote sensing image data; opening an operation processing prediction sample according to the local maximum value and the local minimum value;
the contour smoothing operation includes: and disconnecting the connection of the predicted sample remote sensing image data in the corresponding outline, wherein the connection is less than or equal to a first preset threshold value, and deleting the protruding part of the predicted sample remote sensing image data in the corresponding outline, wherein the protruding degree of the protruding part is less than a second preset threshold value.
The contour optimization operation includes: and carrying out specific optimization operation on the target boundary according to the contour boundary characteristics of the second class of objects.
Specifically, in the embodiment of the present application, the first preset threshold and the second preset threshold may be set according to an actual processing requirement of the user on the remote sensing image data.
Specifically, in the embodiment of the application, the trained primary prediction model is used for predicting other unmarked remote sensing image areas to obtain the greenhouse prediction result. In some specific examples, an unmarked image is cropped into 800 × 800 predicted data, in order to ensure the boundary accuracy of the predicted result, a boundary of 50 pixels is overlapped between two adjacent pieces of data, all the 800 × 800 remote sensing image data are predicted to obtain predicted result data, then all the predicted result data are spliced into an image of the size before cropping, if the predicted result is high in precision, the predicted result is directly input into a DSS model for continuous training, and if the precision is low, sample screening processing is required.
In some specific examples, sample screening work is performed on the prediction result data of the greenhouse, and the sample screening work comprises false-positive sample rejection and sample boundary optimization processing. Specifically, in the embodiment of the application, the extracted greenhouse result is rejected by mistake by using the sample shape elements, the greenhouse shape is regular and mostly rectangular, a rule set is constructed by using the rectangular length-width ratio, and the extraction result which does not meet the condition is rejected. And then, the sample boundary optimization is carried out by utilizing the rule, and the sample boundary optimization can be carried out because the greenhouse is an artificial ground object and the boundary is regular. Firstly, eliminating sample noise by using an expansion corrosion algorithm, wherein an object region which does not contain structural elements can be deleted by corrosion and expansion, the object contour is smoothed, narrow connection is disconnected, and a small protruding part is removed; processing the extraction result by utilizing the principle that the greenhouse boundary is regular, removing large obtuse angles and sharp angles in internal angles of a greenhouse polygon, keeping the polygonal rectangular shape, adjusting sides which are possibly perpendicular to or parallel to the polygon in the polygon, extracting angular points in a grid diagram of the greenhouse result, adjusting the positions of polygonal angular points, and adjusting the positions of the polygon boundaries by linearly fitting the polygon boundaries to enable the boundaries to better meet the actual conditions; if the precision of the screened greenhouse samples is not high, the sample precision can be improved through manual intervention treatment.
In some specific examples, the screened greenhouse sample prediction result data is input into the DSS model as a sample to continue training the model. Specifically, firstly, sample preparation is carried out on the extracted greenhouse vector marking data and the remote sensing image data corresponding to the greenhouse vector marking data to obtain sample data with the size of 800 × 800; keeping the same with the sample format in the initial training of the model, inputting the greenhouse vector marking data with the size of 800 × 800 and the corresponding remote sensing image data into a DSS network for training, predicting the image data of other unmarked areas by using the trained DSS model, continuing training and learning if the prediction result is better, and performing boundary optimization and manual intervention processing if the prediction result is not ideal.
And step 205, continuously repeating the step 204 until a third preset condition is met, and obtaining an optimized prediction model.
Specifically, in some specific examples, the third preset condition may be a number of cycles of the predictive optimization loop process; in other specific examples, the third preset condition may also be an accuracy rate of the primary prediction model for predicting the first remote sensing image data. In the embodiment of the present application, the specific form of the third preset condition is not limited.
Specifically, in some specific examples, the step 205 is repeated until the precision of the greenhouse predicted by the greenhouse prediction model meets the requirement, so as to obtain the optimized greenhouse prediction model.
And step 206, predicting the remote sensing image by using the trained optimized prediction model to obtain sample marking data.
Specifically, in the embodiment of the application, the remote sensing image data labeled by the optimized prediction model can be target remote sensing image data, and the target remote sensing image data is labeled at one time by optimizing the preset model, so that the accuracy of labeling a target object in the target remote sensing image data is improved; in some possible examples, the remote sensing image data marked by the optimized prediction model can also be other remote sensing image data except the target remote sensing image data, and the marking efficiency and the marking accuracy of the remote sensing image data are improved by marking the remote sensing image data by the optimized prediction model.
In the embodiment of the application, a primary prediction model is trained by marking a small amount of remote sensing image data; and predicting the remote sensing image data by using the primary prediction model, repeatedly training the primary prediction model according to the prediction result meeting the conditions to obtain an optimized training model, labeling the remote sensing image data by using the optimized training model, and updating the labeled data and the corresponding remote sensing image data into a database. The problems of low labeling efficiency, inaccurate labeling result and unsatisfactory effect of manually labeling the remote sensing image data in the related technology are solved; the accuracy and the efficiency of marking the remote sensing image data are improved, the efficiency of establishing a sample library is improved, and the labor cost is saved.
Fig. 3 is a flowchart illustrating an implementation of a method for creating and updating a remote sensing image sample library according to another embodiment of the present application.
Based on the foregoing embodiment, referring to fig. 3, another embodiment of the present application provides a method for establishing and updating a remote sensing image sample library, including the following steps:
Specifically, in the embodiment of the present application, the sample data in the sample library meets the requirement of model training, that is, the quantity and quality of the sample data meet the requirement of model training.
Specifically, in the embodiment of the present application, a specific example is described in which a target object is a greenhouse, in an application of a greenhouse sample library, the distribution of greenhouses is concentrated, and for sparse samples, labeled sample vector data corresponding to the greenhouse is extracted from the sample library, then image cropping is performed according to an area where the labeled sample vector data of the greenhouse is located, an image block with a size of 800 × 800 is generated, and a labeled file with a size of 800 × 800 is generated, a value of a greenhouse area in a sample label is 255, and values of other areas are 0.
And 303, performing model training according to the sample data to obtain a prediction model.
Specifically, in the embodiment of the application, the prediction model of the greenhouse is obtained according to the vector marking data file of the greenhouse extracted from the sample library and the corresponding remote sensing image data training model.
And step 304, carrying out prediction and optimization processing on the remote sensing image data to obtain sample marking data.
Specifically, in the embodiment of the application, a greenhouse prediction model is used for predicting and optimizing an unmarked remote sensing image to obtain prediction sample data meeting a fourth preset condition; specifically, in the embodiment of the present application, the fourth preset condition is the same as the second preset condition in embodiment 2.
And 305, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library.
Specifically, in the embodiment of the application, the sample screening is performed on the result of the greenhouse prediction, and if the prediction result is good, the greenhouse extraction result can be directly updated to the database as the sample. Specifically, in some specific examples, the second preset rule means that the accuracy of the extracted sample meets the application requirement.
In the embodiment of the application, a prediction model is trained by extracting sample data in a sample library; and then, marking the remote sensing image data by using the prediction model, and updating the marked data and the corresponding remote sensing image data into a database. The problems of low labeling efficiency, inaccurate labeling result and unsatisfactory effect of manually labeling the remote sensing image data in the related technology are solved; the accuracy and the efficiency of marking the remote sensing image data are improved, the efficiency of establishing a sample library is improved, and the labor cost is saved.
Claims (12)
1. A method for establishing and updating a remote sensing image sample library is characterized by comprising the following steps:
(1) when an operation instruction for establishing a remote sensing image sample library is received, establishing the remote sensing image sample library according to a first preset rule;
(2) when an instruction extracted from the remote sensing image sample database is received, judging whether sample data meeting a first preset condition exists in the sample database according to an extraction requirement;
(3) under the condition that sample data meeting a first preset condition does not exist in the remote sensing image sample library, receiving a mark of a first object in remote sensing image data, and extracting a second object in the remote sensing image data by using an iterative training and preset method according to the mark so as to update the remote sensing image sample library;
(4) under the condition that sample data meeting a first preset condition exists in the remote sensing image sample library, extracting the sample data in the remote sensing image sample library, carrying out model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library.
2. The method according to claim 1, wherein the creating the remote sensing image sample library according to a first preset rule comprises:
(S1) storing the remote sensing image sample library as a spatial database;
(S2) designing a data table structure in the spatial database to be: image ID, image source, image geographic location, number of image bands, image resolution, imaging time, sample type, etc.;
the first preset rule is that sample data is stored according to a preset framing format or a first preset size data block according to requirements, the sample data stored according to standard framing is the sample data produced according to the requirements of deep learning application, and the sample data stored in the first preset size data block is directly used for training.
3. The method according to claim 1, wherein said determining whether there is sample data satisfying a first preset condition in the sample library according to the extraction requirement comprises:
judging whether sample data meeting the training requirement of the deep learning model exists in a sample library or not according to the sample type; the first preset condition is whether the quantity and the quality of the sample data in the sample library meet the deep learning model training requirement or not.
4. The method of claim 1, wherein the receiving a label for a first object in the remote sensing image data, and extracting a second object in the remote sensing image data according to the label by using an iterative training and presetting method; the method comprises the following steps:
(T1) acquiring the preprocessed remote sensing image data;
(T2) marking a target object in the target remote sensing image data in the remote sensing image data to obtain sample data;
(T3) performing model training according to the sample data to obtain a primary prediction model;
(T4) performing prediction and optimization processing on remote sensing image data other than the target remote sensing image data one by one, and training a preliminary prediction model on prediction sample data satisfying a second preset condition and the corresponding remote sensing image data;
(T5) continuously repeating the step T4 until a third preset condition is met, and obtaining an optimized prediction model;
(T6) predicting the remote sensing image by using the trained optimized prediction model to obtain sample marking data;
the first object is labeled sample data, and the second object is unlabeled sample data except the first object.
5. The method of claim 4, wherein the predicting and optimizing process of training the preliminary prediction model with the prediction labeling data and the corresponding remote sensing image data satisfying a second preset condition comprises:
(M1) constructing a type rule set of the prediction sample data according to attribute information of the prediction sample data, the attribute information including at least one of: the remote sensing index, the sample shape, the structure, the texture and the sample space topological relation of the remote sensing image data corresponding to the prediction sample data;
(M2) calculating a confidence level for the prediction sample data in dependence on the rule set;
(M3) screening the prediction sample data according to the confidence.
6. The method according to claim 5, wherein said constructing a type rule set of said prediction sample data according to attribute information of said prediction sample data, further comprises: under the condition that the target object is a first-class object, receiving first processing on the outline of the target object, and deleting discrete points and holes in the prediction sample data; or, receiving second processing on the contour of the target object when the target object is the second class object.
7. The method of claim 6, wherein the first processing comprises: performing contour corrosion expansion operation and contour smoothing operation; the second processing includes: a contour erosion dilation operation and a contour optimization operation.
8. The method of claim 7, wherein the contour erosion dilation operation comprises: carrying out convolution processing on the prediction sample data by using an operator with a characteristic size to complete the open operation of the prediction sample; the contour smoothing operation includes: and disconnecting the connection of the contours corresponding to the prediction samples, wherein the connection is less than or equal to a first preset threshold value, and deleting the protruding parts of the contours corresponding to the prediction samples, wherein the protruding degrees of the protruding parts are less than a second preset threshold value.
9. The method of claim 1, further comprising: and under the condition that a land use current situation vector diagram exists in the geographic area, adding the land use current situation vector diagram and corresponding remote sensing image data into a sample library.
10. The method according to claim 10, wherein the current land use situation vector map is a land use vector map within a preset time period, and the preset time period is a time period which is located before the time of acquiring the remote sensing image data and is within a preset interval from the time of acquiring the remote sensing image data.
11. The method of claim 1, wherein said extracting sample data in said sample library comprises:
under the condition that the sample data is first-class labeled sample data, cutting the remote sensing image data corresponding to the labeled sample data according to the geographic space position of the remote sensing image data corresponding to the labeled sample data to generate remote sensing image blocks with the same size; the first type of marking sample data is marking sample data of a target object corresponding to the sample data, which is distributed in the remote sensing image data and is less than or equal to a third preset threshold value; or,
under the condition that the sample data is second-class labeled sample data, cutting the remote sensing image data corresponding to the labeled sample data from left to right and from top to bottom to generate remote sensing image blocks with the same size; and the second type of marking sample data is marking sample data of a target object corresponding to the marking sample data, which is distributed in the remote sensing image data and is larger than or equal to a fourth preset threshold value.
12. The method according to claim 1, wherein the extracting sample data in the remote sensing image sample library, performing model training, remote sensing image data prediction and optimization processing, extracting an incremental sample according to a second preset rule, and updating the incremental sample into the sample library comprises:
(N1) performing model training according to the sample data to obtain a prediction model;
(N2) carrying out prediction and optimization processing on the unmarked remote sensing image by using the prediction model to obtain prediction sample data meeting a fourth preset condition;
(N3) updating sample data of which the precision meets the preset precision in the prediction sample data into a sample library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911372783.5A CN111144487B (en) | 2019-12-27 | 2019-12-27 | Method for establishing and updating remote sensing image sample library |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911372783.5A CN111144487B (en) | 2019-12-27 | 2019-12-27 | Method for establishing and updating remote sensing image sample library |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111144487A true CN111144487A (en) | 2020-05-12 |
| CN111144487B CN111144487B (en) | 2023-09-26 |
Family
ID=70521152
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911372783.5A Active CN111144487B (en) | 2019-12-27 | 2019-12-27 | Method for establishing and updating remote sensing image sample library |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111144487B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112884791A (en) * | 2021-02-02 | 2021-06-01 | 重庆市地理信息和遥感应用中心 | Method for constructing large-scale remote sensing image semantic segmentation model training sample set |
| CN113223042A (en) * | 2021-05-19 | 2021-08-06 | 自然资源部国土卫星遥感应用中心 | Intelligent acquisition method and equipment for remote sensing image deep learning sample |
| CN113705593A (en) * | 2020-05-21 | 2021-11-26 | 孙民 | Method for generating training data and electronic device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105279519A (en) * | 2015-09-24 | 2016-01-27 | 四川航天系统工程研究所 | Remote sensing image water body extraction method and system based on cooperative training semi-supervised learning |
| CN108764263A (en) * | 2018-02-12 | 2018-11-06 | 北京佳格天地科技有限公司 | The atural object annotation equipment and method of remote sensing image |
| CN108921025A (en) * | 2018-06-01 | 2018-11-30 | 苏州中科天启遥感科技有限公司 | A kind of object level classification samples automatic selecting method of collaborative variation detection |
| CN109670060A (en) * | 2018-12-10 | 2019-04-23 | 北京航天泰坦科技股份有限公司 | A kind of remote sensing image semi-automation mask method based on deep learning |
| CN110298348A (en) * | 2019-06-12 | 2019-10-01 | 苏州中科天启遥感科技有限公司 | Remote sensing image building sample areas extracting method and system, storage medium, equipment |
| CN110309809A (en) * | 2019-07-09 | 2019-10-08 | 广西壮族自治区基础地理信息中心 | High Resolution Remote Sensing Satellites image sugarcane extracting method based on deep neural network |
| CN110502654A (en) * | 2019-08-26 | 2019-11-26 | 长光卫星技术有限公司 | A kind of object library generation system suitable for multi-source heterogeneous remotely-sensed data |
-
2019
- 2019-12-27 CN CN201911372783.5A patent/CN111144487B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105279519A (en) * | 2015-09-24 | 2016-01-27 | 四川航天系统工程研究所 | Remote sensing image water body extraction method and system based on cooperative training semi-supervised learning |
| CN108764263A (en) * | 2018-02-12 | 2018-11-06 | 北京佳格天地科技有限公司 | The atural object annotation equipment and method of remote sensing image |
| CN108921025A (en) * | 2018-06-01 | 2018-11-30 | 苏州中科天启遥感科技有限公司 | A kind of object level classification samples automatic selecting method of collaborative variation detection |
| CN109670060A (en) * | 2018-12-10 | 2019-04-23 | 北京航天泰坦科技股份有限公司 | A kind of remote sensing image semi-automation mask method based on deep learning |
| CN110298348A (en) * | 2019-06-12 | 2019-10-01 | 苏州中科天启遥感科技有限公司 | Remote sensing image building sample areas extracting method and system, storage medium, equipment |
| CN110309809A (en) * | 2019-07-09 | 2019-10-08 | 广西壮族自治区基础地理信息中心 | High Resolution Remote Sensing Satellites image sugarcane extracting method based on deep neural network |
| CN110502654A (en) * | 2019-08-26 | 2019-11-26 | 长光卫星技术有限公司 | A kind of object library generation system suitable for multi-source heterogeneous remotely-sensed data |
Non-Patent Citations (1)
| Title |
|---|
| 邵振峰;张源;黄昕;朱秀丽;吴亮;万波;: "基于多源高分辨率遥感影像的2 m不透水面一张图提取" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113705593A (en) * | 2020-05-21 | 2021-11-26 | 孙民 | Method for generating training data and electronic device |
| CN112884791A (en) * | 2021-02-02 | 2021-06-01 | 重庆市地理信息和遥感应用中心 | Method for constructing large-scale remote sensing image semantic segmentation model training sample set |
| CN113223042A (en) * | 2021-05-19 | 2021-08-06 | 自然资源部国土卫星遥感应用中心 | Intelligent acquisition method and equipment for remote sensing image deep learning sample |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111144487B (en) | 2023-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11727163B2 (en) | System and method for construction estimation using aerial images | |
| CN107610131B (en) | Image clipping method and image clipping device | |
| CN111144487A (en) | Method for establishing and updating remote sensing image sample library | |
| CN116994140A (en) | Cultivated land extraction method, device, equipment and medium based on remote sensing image | |
| CN109684428A (en) | Spatial data building method, device, equipment and storage medium | |
| CN110688756B (en) | Collection and drawing integrated completion drawing manufacturing method | |
| CN113223042B (en) | Intelligent acquisition method and equipment for remote sensing image deep learning sample | |
| CN110083720B (en) | Method and device for constructing real scene semantic structure model | |
| CN115546116B (en) | Full-coverage type rock mass discontinuous surface extraction and interval calculation method and system | |
| CN108765445B (en) | Lung trachea segmentation method and device | |
| Gold et al. | Map generalization by skeleton retraction | |
| Guilbert et al. | Terrain generalisation | |
| US20220391553A1 (en) | Building information modeling systems and methods | |
| CN112991490A (en) | Informatization chart production method and system | |
| CN111414907A (en) | Data set labeling method, data set labeling device and computer-readable storage medium | |
| CN116956838A (en) | Project determining method and device for engineering quantity list | |
| CN112990001A (en) | Remote sensing image monomer building rapid and accurate extraction method applied to digital collection | |
| Tan et al. | Semantic enrichment for rooftop modeling using aerial LiDAR reflectance | |
| CN113158856A (en) | Processing method and device for extracting target area in remote sensing image | |
| CN111382645B (en) | Method and system for identifying overdue building in electronic map | |
| CN116343254A (en) | Cross-section identifying method, cross-section identifying device, electronic equipment and storage medium | |
| Chen et al. | A new DEM generalization method based on watershed and tree structure | |
| CN113850254B (en) | Building vector contour reduction method based on deep learning, model and model building method | |
| CN115600732A (en) | City planning method and system based on big data analysis | |
| CN116956837A (en) | Project determining method and device for engineering quantity list |
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 |