CN111476308B - Remote sensing image classification method and device based on priori geometric constraint and electronic equipment - Google Patents
Remote sensing image classification method and device based on priori geometric constraint and electronic equipment Download PDFInfo
- Publication number
- CN111476308B CN111476308B CN202010282973.4A CN202010282973A CN111476308B CN 111476308 B CN111476308 B CN 111476308B CN 202010282973 A CN202010282973 A CN 202010282973A CN 111476308 B CN111476308 B CN 111476308B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- historical
- sensing image
- classification
- distribution 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000013135 deep learning Methods 0.000 claims abstract description 118
- 238000013145 classification model Methods 0.000 claims abstract description 55
- 238000012549 training Methods 0.000 claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000001228 spectrum Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000903 blocking effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 239000003643 water by type Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a remote sensing image classification method, a remote sensing image classification device and electronic equipment based on priori geometric constraint; comprising the following steps: acquiring remote sensing images and historical spatial distribution data of the remote sensing images; the historical spatial distribution data is used for representing the spatial granularity condition of the historical classification result of the remote sensing image; inputting the remote sensing images and the historical space distribution data into a deep learning classification model which is trained in advance, and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing images, classification labels of the historical remote sensing images and the historical spatial distribution data polygons. In the mode, the historical space distribution data is used as the pattern spot granularity priori data of the remote sensing image, the space scale of the remote sensing image can be restrained, and the pattern spot granularity and the boundary trend of the remote sensing image are determined, so that the accuracy, the reliability and the popularization and application of the remote sensing image classification method based on the priori geometric constraint are improved.
Description
Technical Field
The invention relates to the technical field of remote sensing image classification, in particular to a remote sensing image classification method, device and electronic equipment based on priori geometric constraint.
Background
In recent years, the deep learning technology is rapidly developed, and the deep learning technology is widely applied to the field of deep learning classification of remote sensing images. The traditional remote sensing image deep learning takes an image block as an object, and the deep learning classification is completed through two processes of sample training and sample prediction: (1) sample training: training and learning under a deep learning engine by adopting a certain amount of representative image blocks and corresponding classification results, and establishing a prediction model between the image blocks and the classification results; (2) sample prediction: the method comprises the steps of dividing the image into image blocks with specified sizes according to requirements, training a built prediction model by using samples for each image block, taking the image blocks as input parameters, and obtaining output parameters, namely image block classification results, through calculation.
In the related art, the fundamental problem of deep learning classification of remote sensing images is a spatial scale problem. The traditional image deep learning method is used for performing deep learning classification based on image blocks, is limited by insufficient image information, and has the bottleneck that the target scale is difficult to define: the classification result boundary is inaccurate, the result has instability, the model generalization capability is poor, the deep learning classification result depends on the selection of training samples to a certain extent, and the actual cross-region popularization and application are difficult. For example: the goal is to extract building areas, and the results may be extracted as cells or as parts of houses, resulting in large difference between classification results and requirements, and even if a part of the building is extracted, the boundaries are blocked by trees and the like and become concave-convex saw teeth, which does not meet the actual requirements.
Disclosure of Invention
In view of the above, the present invention aims to provide a remote sensing image classification method, device and electronic equipment based on prior geometric constraint, so as to improve accuracy, reliability and popularization and application of the remote sensing image classification method based on prior geometric constraint.
In a first aspect, an embodiment of the present invention provides a remote sensing image classification method based on a priori geometric constraint, including: acquiring remote sensing images and historical spatial distribution data of the remote sensing images; the historical spatial distribution data is used for representing the spatial granularity condition of the historical classification result of the remote sensing image; inputting the remote sensing images and the historical space distribution data into a deep learning classification model which is trained in advance, and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon of the historical spatial distribution data; the deep learning classification model consists of three variables of images, historical space data and classification labels, and comprises pattern spectrum information classification under the known space object segmentation granularity.
In a preferred embodiment of the present invention, the step of obtaining the historical spatial distribution data of the remote sensing image includes: acquiring a historical classification result of the remote sensing image; determining the object in the history classification result and the position information of each object; and taking the position and the pattern spot range information of each object as historical space distribution data, wherein the historical space distribution data is a Polygon, the space data format of the historical space distribution data is SHAPEFILE, GEOJSON or WKT, and the historical space distribution data comprises a Polygon range.
In a preferred embodiment of the present invention, the step of inputting the remote sensing image and the historical spatial distribution data into the pre-trained deep learning classification model and outputting the classification result of the remote sensing image includes: inputting the remote sensing image and the historical space distribution data into a deep learning classification model which is trained in advance; wherein the framework of the deep learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j; the deep learning classification model divides the historical space distribution data and the remote sensing images, performs deep learning classification on the divided remote sensing images, and outputs classification results of the remote sensing images.
In a preferred embodiment of the present invention, the deep learning classification model divides the historical spatial distribution data and the remote sensing image, performs deep learning classification on the divided remote sensing image, and outputs a classification result of the remote sensing image, including: dividing the historical spatial distribution data into a plurality of data blocks; wherein each data block includes an object in the historical spatial distribution data; dividing the remote sensing image into a plurality of image blocks according to the same dividing mode as the historical space distribution data; performing deep learning classification on each image block to obtain a classification result of each image block; and combining the classification results of each image block and outputting the classification results of the remote sensing images.
In a preferred embodiment of the present invention, the step of dividing the historical spatial distribution data into a plurality of data blocks includes: according to the boundary of the object in the historical space distribution data, the historical space distribution data is divided into a plurality of data blocks, and each data block boundary corresponds to the circumscribed rectangle of one historical classification pattern spot.
In a second aspect, an embodiment of the present invention further provides a remote sensing image classification device based on a priori geometric constraint, including: the remote sensing image acquisition module is used for acquiring the remote sensing images and the historical spatial distribution data of the remote sensing images; the historical spatial distribution data is used for representing the spatial granularity condition of the historical classification result of the remote sensing image; the classification result output module is used for inputting the remote sensing images and the historical space distribution data into a deep learning classification model which is trained in advance and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon of the historical spatial distribution data; the deep learning classification model consists of three variables of images, historical space data and classification labels, and comprises pattern spectrum information classification under the known space object segmentation granularity.
In a preferred embodiment of the present invention, the remote sensing image acquisition module is configured to: acquiring a historical classification result of the remote sensing image; determining the object in the history classification result and the position information of each object; and taking the position and the pattern spot range information of each object as historical space distribution data, wherein the historical space distribution data is a Polygon, the space data format of the historical space distribution data is SHAPEFILE, GEOJSON or WKT, and the historical space distribution data comprises a Polygon range.
In a preferred embodiment of the present invention, the classification result output module is configured to: inputting the remote sensing image and the historical space distribution data into a deep learning classification model which is trained in advance; wherein the framework of the deep learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j; the deep learning classification model divides the historical space distribution data and the remote sensing images, performs deep learning classification on the divided remote sensing images, and outputs classification results of the remote sensing images.
In a third aspect, an embodiment of the present invention further provides an electronic device, including a processor and a memory, where the memory stores computer executable instructions executable by the processor, and the processor executes the computer executable instructions to implement the steps of the remote sensing image classification method based on the prior geometric constraint.
In a fourth aspect, an embodiment of the present invention further provides a computer readable storage medium, where computer executable instructions are stored, where the computer executable instructions, when invoked and executed by a processor, cause the processor to implement the steps of the remote sensing image classification method based on prior geometric constraints.
The embodiment of the invention has the following beneficial effects:
According to the remote sensing image classification method, device and electronic equipment based on the priori geometric constraint, the historical space distribution data is used as the pattern spot granularity priori data of the remote sensing image, the space scale of the remote sensing image can be constrained, and the pattern spot granularity and the boundary trend of the remote sensing image can be determined, so that the accuracy, the reliability and the popularization and application of the remote sensing image classification method based on the priori geometric constraint are improved.
Additional features and advantages of the disclosure will be set forth in the description which follows, or in part will be obvious from the description, or may be learned by practice of the techniques of the disclosure.
The foregoing objects, features and advantages of the disclosure will be more readily apparent from the following detailed description of the preferred embodiments taken in conjunction with the accompanying drawings.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a remote sensing image classification method based on prior geometric constraint provided by an embodiment of the invention;
FIG. 2 is a flowchart of another remote sensing image classification method based on prior geometric constraints according to an embodiment of the present invention;
Fig. 3 is a schematic structural diagram of a remote sensing image classification device based on prior geometric constraint according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, but 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.
At present, the traditional image deep learning method is used for performing deep learning classification based on image blocks, is limited by insufficient image information, and has the bottleneck that the target scale is difficult to define: the classification result boundary is inaccurate, the result has instability, the model generalization capability is poor, the deep learning classification result depends on the selection of training samples to a certain extent, and the actual cross-region popularization and application are difficult. Based on the above, the embodiment of the invention provides a remote sensing image classification method, a remote sensing image classification device and electronic equipment based on priori geometric constraint, and relates to the technical field of remote sensing design, in particular to a depth learning method based on space priori constraint.
For the convenience of understanding the present embodiment, the remote sensing image classification method based on prior geometric constraint disclosed in the present embodiment is first described in detail.
Example 1
The embodiment of the invention provides a remote sensing image classification method based on prior geometric constraint, referring to a flow chart of the remote sensing image classification method based on prior geometric constraint shown in fig. 1, the remote sensing image classification method based on prior geometric constraint comprises the following steps:
Step S102, acquiring a remote sensing image and historical space distribution data of the remote sensing image; the historical spatial distribution data is used for representing the spatial granularity condition of the historical classification result of the remote sensing image.
The remote sensing image is a film or a photo for recording the electromagnetic wave sizes of various ground objects, and is mainly divided into an aerial photo and a satellite photo. The type and the position of the object can be determined from the remote sensing image, and the object can be: people, buildings, natural resources (mountains, waters, lakes, etc.). These objects can be found from the remote sensing image and the boundaries and coordinates of these objects are determined as spatially distributed data for these objects.
The historical spatial distribution data refers to the information of the boundary and coordinates of the object in the same region before the shooting time of the current remote sensing image. For example: the remote sensing image was already photographed for this area 3 months before the photographing time of the current remote sensing image, which is called the last remote sensing image. Several objects have been extracted from the last remote sensing image, and the spatial distribution data of these objects. Then, the spatial distribution data extracted from the previous remote sensing image may be referred to as historical spatial distribution data.
Step S104, inputting the remote sensing images and the historical space distribution data into a deep learning classification model which is trained in advance, and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon of the historical spatial distribution data; the deep learning classification model consists of three variables of images, historical space data and classification labels, and comprises pattern spectrum information classification under the known space object segmentation granularity.
The deep learning classification model is used for performing deep learning classification on objects included in the remote sensing image, namely, firstly determining the objects included in the remote sensing image and the spatial distribution data of the objects, and then performing deep learning classification on the objects. For example: the object a deep learning is classified as a building and the object B deep learning is classified as a lake. In the deep learning classification model in this embodiment, the historical spatial distribution data is used as priori knowledge, the remote sensing images are primarily divided by the historical spatial distribution data, and then the primarily divided remote sensing images are subjected to deep learning classification.
The deep learning classification model in the embodiment has the classification of the pattern spectrum information under the known space object segmentation granularity, namely, the general deep learning needs to automatically identify the space granularity and the ground object type, while the deep learning model in the embodiment has the known space granularity, so that only the space type needs to be identified. Because the space granularity and the ground object type are directly related, when the space granularity and the ground object type are required to be identified simultaneously, the uncertainty of the space granularity and the ground object type is large, and the classification result is influenced.
For example, when the boundary of a house is known, the texture features of the house on the image are strong, and the deep learning classification can be directly accurate. However, since the general deep learning model is to determine the boundary and the type of the house at the same time, it is easy to divide the house into a polygon, and then the type is recognized as a type such as a woodland or other factory building, but is not easy to recognize as a building. The embodiment adopts the known space granularity of the pattern spots as the input quantity, and can greatly improve the accuracy of classification of the ground objects. According to the remote sensing image classification method based on the priori geometric constraint, the historical space distribution data is used as the pattern spot granularity priori data of the remote sensing image, the space scale of the remote sensing image can be constrained, and the pattern spot granularity and the boundary trend of the remote sensing image can be determined, so that the accuracy, the reliability and the popularization and application of the remote sensing image classification method based on the priori geometric constraint are improved.
Example 2
The embodiment of the invention also provides another remote sensing image classification method based on priori geometric constraint; the method is realized on the basis of the method of the embodiment; the method focuses on the specific implementation of determining the resident data of each base station based on the signaling data and the pre-acquired base station location table.
As shown in fig. 2, another flow chart of a remote sensing image classification method based on a priori geometric constraint includes the following steps:
Step S202, a remote sensing image is acquired.
The remote sensing image is a remote sensing image which needs to be subjected to deep learning classification, and may include a plurality of objects, and the objects need to be subjected to deep learning classification, so as to determine the boundary and position information of the objects. The remote sensing image may not include an object, and the remote sensing image is subjected to deep learning classification to confirm that the remote sensing image does not include an object.
Step S204, obtaining a history classification result of the remote sensing image.
The historical classification result of the remote sensing image refers to the classification result of the historical remote sensing image before the shooting time of the current remote sensing image in the unified area. In the historical classification result, the deep learning classification condition of the object included in the historical remote sensing image is described, and the spatial granularity condition of each object is also described.
Step S206, determining the object in the history classification result and the position information of each object.
The historical classification result in this embodiment includes not only the deep learning classification condition of the object in the historical remote sensing image, but also the spatial distribution condition of each object in the historical remote sensing image. The position information of the object, that is, the coordinate information of the object and the boundary information of the object can be determined from the spatial distribution.
In step S208, the position and the patch range information of each object are used as the historical spatial distribution data, the historical spatial distribution data is a Polygon, the spatial data format of the historical spatial distribution data is SHAPEFILE, GEOJSON or WKT, and the historical spatial distribution data includes a Polygon range.
The historical spatial distribution data can be determined by combining the position information of each object in the historical remote sensing image. Wherein, the historical space distribution data only has a polygon range and no attribute information. Step S210, inputting the remote sensing images and the historical space distribution data into a pre-trained deep learning classification model, and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing images, classification results of the historical remote sensing images and the polygons of the historical spatial distribution data; the deep learning classification model consists of three variables of images, historical space data and classification labels, and comprises pattern spectrum information classification under the known space object segmentation granularity.
In the process of training the deep learning classification model, firstly, the historical remote sensing image is required to be subjected to blocking treatment: namely: dividing the historical remote sensing image into a plurality of image blocks with uniform sizes according to the blocking requirement of the deep learning; however, a priori data chunking processing may be used: the historical space distribution data can be used as priori knowledge to divide the classification result of the historical remote sensing image into a plurality of data blocks with uniform sizes according to the same segmentation requirement: if the classification result of the historical remote sensing image is a grid, the historical remote sensing image is directly segmented, and if the classification result of the historical remote sensing image is a vector, the historical remote sensing image is segmented after being rasterized. Finally, model training is carried out: when the image deep learning modeling is carried out, on the basis of the original image blocks and the classification results, the spatial distribution of the image spots of the classification results is used as parameters to be input into a training model, and a three-parameter deep learning classification model is established: deep learning class type = f (image block, plaque boundary).
The step of inputting the remote sensing image and the historical spatial distribution data into the pre-trained deep learning classification model and outputting the classification result of the remote sensing image can be performed by steps A1-A2:
Step A1, inputting remote sensing images and historical space distribution data into a deep learning classification model which is trained in advance; wherein the framework of the deep learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j.
The input of the deep learning classification model is remote sensing images and historical spatial distribution data. The historical space distribution data is used as priori data for restraining the size of the image blocks divided by the remote sensing image, and can be used as a priori value of model prediction.
And step A2, the deep learning classification model divides the historical space distribution data and the remote sensing images, performs deep learning classification on the divided remote sensing images, and outputs classification results of the remote sensing images.
The remote sensing image needs to be divided into a plurality of image blocks, and the dividing method can be determined based on the boundary of the object in the historical space distribution data. For example, it may be performed by steps B1 to B4:
step B1, dividing historical space distribution data into a plurality of data blocks; wherein each data block comprises an object in the historical spatial distribution data.
Firstly, the deep learning classification model needs to divide the historical space distribution data into a plurality of data blocks, and the historical space distribution data can be divided into a plurality of data blocks according to the boundary (namely the pattern spot boundary) of the object in the historical space distribution data; wherein each data block boundary corresponds to a circumscribed rectangle of a historical classification map spot. This step may be referred to as a priori extraction: and extracting or collecting a historical image classification result for the target area needing deep learning, carrying out prior data block processing on the target area according to a prior data block processing method, and extracting a space boundary as a prior value of model prediction.
And step B2, dividing the remote sensing image into a plurality of image blocks according to the same dividing mode as the historical space distribution data.
The same division mode of the historical space distribution data is the same as the division mode of the remote sensing image into a plurality of image blocks. The historical space distribution data is divided by the boundary of the object, and each divided area comprises various objects. Then the remote sensing image is divided according to the same dividing mode, so that the moderate scale of the image block of the remote sensing image can be ensured, the condition that the scale is too large or the scale is too small can not occur, namely, the goal is to extract the building area of the house, and the result is extracted as a cell or a part of the house, so that the classification result and the requirement have large difference. Thereby increasing the accuracy of the deep learning classification and the efficiency of the deep learning classification.
And B3, performing deep learning classification on each image block to obtain a classification result of each image block.
And then, a deep learning classification model carries out deep learning classification on each image block, the image blocks before the deep learning classification correspond to the pattern boundary data blocks divided by the historical space distribution data one by one, the image blocks are input into the deep learning classification model, and the deep learning classification model can obtain the classification result of each image block.
And step B4, combining the classification results of each image block and outputting the classification results of the remote sensing images.
And combining the classification results of each image block to obtain the classification result of the remote sensing image, and finally outputting the classification result of the remote sensing image by the deep learning classification model.
The method provided by the embodiment of the invention can improve the accuracy of deep learning classification: in the image deep learning process, the historical deep learning classification map spot boundary is increased to be the prior space constraint, so that the map spot deep learning classification scale of the deep learning is determined, the accuracy and the efficiency of the deep learning classification calculation are greatly improved, and the problems of too thin deep learning classification and too thick deep learning classification caused by insufficient information in the traditional method are avoided.
The method provided by the embodiment of the invention can improve the boundary precision: by means of historical extraction of the boundary information of the image spots, the boundary saw tooth condition can be effectively processed in a new learning process, so that the result meets the actual requirement, the application requirement is met, and the saw tooth problem of the traditional pure image learning is solved.
The method provided by the embodiment of the invention can improve the stability: the complexity of the deep learning algorithm is greatly simplified through the prior constraint of the granularity of the pattern spots, the deep learning algorithm does not need to judge the deep learning classification among objects with different forces, such as a district, a house and a part, the difficulty of the deep learning classification is too high, and the deep learning algorithm is directly focused on the corresponding deep learning classification with the same granularity, such as a large house and a small house, the boundary characteristics of similar objects are strong, and the consistency of internal textures is strong. Under the condition, the stability of the model is greatly improved, and when the cross-region popularization and application are carried out, the granularity of the image object can be restrained by means of the prior knowledge of the map-patch boundary, so that the study depth learning classification of the map-patch level is focused, namely, the consistency of the features of the houses for dividing Beijing and the houses for dividing Shanghai is strong, the stability of the depth learning classification is improved, and the industrialization of the image depth learning can be promoted to fall into the application.
Example 3
Corresponding to the above method embodiment, the embodiment of the present invention provides a remote sensing image classification device based on a priori geometric constraint, as shown in fig. 3, which is a schematic structural diagram of the remote sensing image classification device based on the priori geometric constraint, and the remote sensing image classification device based on the priori geometric constraint includes:
The remote sensing image acquisition module 31 is configured to acquire a remote sensing image and historical spatial distribution data of the remote sensing image; the historical spatial distribution data is used for representing the spatial granularity condition of the historical classification result of the remote sensing image;
The classification result output module 32 is configured to input the remote sensing image and the historical spatial distribution data into a pre-trained deep learning classification model, and output a classification result of the remote sensing image; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon of the historical spatial distribution data; the deep learning classification model consists of three variables of images, historical space data and classification labels, and comprises pattern spectrum information classification under the known space object segmentation granularity.
According to the remote sensing image classification device based on the priori geometric constraint, the historical space distribution data is used as the pattern spot granularity priori data of the remote sensing image, the space scale of the remote sensing image can be constrained, and the pattern spot granularity and the boundary trend of the remote sensing image can be determined, so that the accuracy, the reliability and the popularization and application of the remote sensing image classification method based on the priori geometric constraint are improved.
In some embodiments, the remote sensing image acquisition module is configured to: acquiring a historical classification result of the remote sensing image; determining the object in the history classification result and the position information of each object; and taking the position and the pattern spot range information of each object as historical space distribution data, wherein the historical space distribution data is a Polygon, the space data format of the historical space distribution data is SHAPEFILE, GEOJSON or WKT, and the historical space distribution data comprises a Polygon range.
In some embodiments, the classification result output module is configured to: inputting the remote sensing image and the historical space distribution data into a deep learning classification model which is trained in advance; wherein the framework of the deep learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j; the deep learning classification model divides the historical space distribution data and the remote sensing images, performs deep learning classification on the divided remote sensing images, and outputs classification results of the remote sensing images.
In some embodiments, the classification result output module is configured to: dividing the historical spatial distribution data into a plurality of data blocks; wherein each data block includes an object in the historical spatial distribution data; dividing the remote sensing image into a plurality of image blocks according to the same dividing mode as the historical space distribution data; performing deep learning classification on each image block to obtain a classification result of each image block; and combining the classification results of each image block and outputting the classification results of the remote sensing images.
In some embodiments, the classification result output module is configured to: according to the boundary of the object in the historical space distribution data, the historical space distribution data is divided into a plurality of data blocks, and each data block boundary corresponds to the circumscribed rectangle of one historical classification pattern spot.
The remote sensing image classification device based on the priori geometric constraint provided by the embodiment of the invention has the same technical characteristics as the remote sensing image classification method based on the priori geometric constraint provided by the embodiment, so that the same technical problems can be solved, and the same technical effects can be achieved.
Example 4
The embodiment of the invention also provides electronic equipment for running the remote sensing image classification method based on the priori geometric constraint; referring to fig. 4, an electronic device includes a memory 100 and a processor 101, where the memory 100 is configured to store one or more computer instructions, and the one or more computer instructions are executed by the processor 101 to implement the above-described remote sensing image classification method based on prior geometric constraints.
Further, the electronic device shown in fig. 4 further includes a bus 102 and a communication interface 103, and the processor 101, the communication interface 103, and the memory 100 are connected through the bus 102.
The memory 100 may include a high-speed random access memory (RAM, random Access Memory), and may further include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the system network element and at least one other network element is implemented via at least one communication interface 103 (which may be wired or wireless), and may use the internet, a wide area network, a local network, a metropolitan area network, etc. Bus 102 may be an ISA bus, a PCI bus, an EISA bus, or the like. The buses may be divided into address buses, data buses, control buses, etc. For ease of illustration, only one bi-directional arrow is shown in FIG. 4, but not only one bus or type of bus.
The processor 101 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in the processor 101 or instructions in the form of software. The processor 101 may be a general-purpose processor, including a central processing unit (Central Processing Unit, abbreviated as CPU), a network processor (Network Processor, abbreviated as NP), and the like; but may also be a digital signal Processor (DIGITAL SIGNAL Processor, DSP), application Specific Integrated Circuit (ASIC), field-Programmable gate array (FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in the memory 100 and the processor 101 reads information in the memory 100 and in combination with its hardware performs the steps of the method of the previous embodiments.
The embodiment of the invention also provides a computer readable storage medium, which stores computer executable instructions that, when being called and executed by a processor, cause the processor to implement the remote sensing image classification method based on prior geometric constraint, and specific implementation can be seen in the method embodiment and will not be repeated here.
The remote sensing image classification method, device and computer program product of electronic equipment based on prior geometric constraint provided by the embodiment of the invention comprise a computer readable storage medium storing program codes, and instructions included in the program codes can be used for executing the method in the previous method embodiment, and specific implementation can be seen in the method embodiment and will not be repeated here.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the apparatus and/or the electronic device described above may refer to the corresponding process in the foregoing method embodiment, which is not described in detail herein.
Finally, it should be noted that: the above examples are only specific embodiments of the present invention, and are not intended to limit the scope of the present invention, but it should be understood by those skilled in the art that the present invention is not limited thereto, and that the present invention is described in detail with reference to the foregoing examples: any person skilled in the art may modify or easily conceive of the technical solution described in the foregoing embodiments, or perform equivalent substitution of some of the technical features, while remaining within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (7)
1. A remote sensing image classification method based on priori geometric constraint is characterized by comprising the following steps:
acquiring a remote sensing image and historical spatial distribution data of the remote sensing image; the historical spatial distribution data are used for representing the spatial granularity condition of the historical classification result of the remote sensing image;
Inputting the remote sensing images and the historical spatial distribution data into a deep learning classification model which is trained in advance, and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon range in the historical spatial distribution data of the historical remote sensing image; the input data of the deep learning classification model consists of three variables including a historical remote sensing image, a classification label of the historical remote sensing image and a polygonal range in the historical spatial distribution data of the historical remote sensing image, and the deep learning classification model comprises pattern spot spectrum information classification under the known space object segmentation granularity;
the step of obtaining the historical space distribution data of the remote sensing image comprises the following steps:
acquiring a history classification result of the remote sensing image;
determining an object in the historical classification result and position information of each object;
The polygon range of each object obtained according to the position information of each object is used as the historical space distribution data, and the data format of the historical space distribution data is SHAPEFILE, GEOJSON or WKT;
The step of inputting the remote sensing image and the historical space distribution data into a deep learning classification model which is trained in advance and outputting the classification result of the remote sensing image comprises the following steps:
dividing the historical spatial distribution data into a plurality of data blocks; wherein each of said data blocks comprises an object in said historical spatial distribution data;
dividing the remote sensing image into a plurality of image blocks according to the same dividing mode as the historical space distribution data;
Performing deep learning classification on each image block to obtain a classification result of each image block;
and carrying out space splicing and combination on the classification result of each image block, and outputting the classification result of the remote sensing image.
2. The method of claim 1, wherein the framework of the deep-learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j.
3. The method of claim 1, wherein the step of dividing the historical spatial distribution data into a plurality of data blocks comprises:
dividing the historical space distribution data into a plurality of data blocks according to the polygonal range in the historical space distribution data, wherein each data block boundary corresponds to an external rectangle of a historical classification map spot.
4. Remote sensing image classification device based on priori geometric constraint, characterized by comprising:
the remote sensing image acquisition module is used for acquiring the remote sensing image and the historical spatial distribution data of the remote sensing image; the historical spatial distribution data are used for representing the spatial granularity condition of the historical classification result of the remote sensing image;
The classification result output module is used for inputting the remote sensing images and the historical space distribution data into a deep learning classification model which is trained in advance and outputting classification results of the remote sensing images; the deep learning classification model is obtained by training based on the historical remote sensing image, the classification label of the historical remote sensing image and the polygon range in the historical spatial distribution data of the historical remote sensing image; the input data of the deep learning classification model consists of three variables including a historical remote sensing image, a classification label of the historical remote sensing image and a polygonal range in the historical spatial distribution data of the historical remote sensing image, and the deep learning classification model comprises pattern spot spectrum information classification under the known space object segmentation granularity;
the remote sensing image acquisition module is used for: acquiring a history classification result of the remote sensing image; determining an object in the historical classification result and position information of each object; the polygon range of each object obtained according to the position of each object is used as the historical space distribution data, and the data format of the historical space distribution data is SHAPEFILE, GEOJSON or WKT;
The step of inputting the remote sensing image and the historical space distribution data into a deep learning classification model which is trained in advance and outputting the classification result of the remote sensing image comprises the following steps: dividing the historical spatial distribution data into a plurality of data blocks; wherein each of said data blocks comprises an object in said historical spatial distribution data; dividing the remote sensing image into a plurality of image blocks according to the same dividing mode as the historical space distribution data; performing deep learning classification on each image block to obtain a classification result of each image block; and carrying out space splicing and combination on the classification result of each image block, and outputting the classification result of the remote sensing image.
5. The apparatus of claim 4, wherein the framework of the deep-learning classification model is TensorFlow, keras, pyTorch, caffe or DEEPLEARNING j.
6. An electronic device comprising a processor and a memory, the memory storing computer executable instructions executable by the processor to perform the steps of the remote sensing image classification method based on prior geometric constraints of any one of claims 1 to 3.
7. A computer readable storage medium storing computer executable instructions which, when invoked and executed by a processor, cause the processor to implement the steps of the remote sensing image classification method based on a priori geometric constraints of any of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010282973.4A CN111476308B (en) | 2020-04-10 | 2020-04-10 | Remote sensing image classification method and device based on priori geometric constraint and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010282973.4A CN111476308B (en) | 2020-04-10 | 2020-04-10 | Remote sensing image classification method and device based on priori geometric constraint and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111476308A CN111476308A (en) | 2020-07-31 |
| CN111476308B true CN111476308B (en) | 2024-06-11 |
Family
ID=71751765
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010282973.4A Active CN111476308B (en) | 2020-04-10 | 2020-04-10 | Remote sensing image classification method and device based on priori geometric constraint and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111476308B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221684B (en) * | 2021-04-27 | 2024-08-02 | 深圳前海微众银行股份有限公司 | Remote sensing image processing method, device, equipment and storage medium |
| CN117370844A (en) * | 2023-11-02 | 2024-01-09 | 重庆数字城市科技有限公司 | Space big data management method and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109446992A (en) * | 2018-10-30 | 2019-03-08 | 苏州中科天启遥感科技有限公司 | Remote sensing image building extracting method and system, storage medium, electronic equipment based on deep learning |
| CN109543630A (en) * | 2018-11-28 | 2019-03-29 | 苏州中科天启遥感科技有限公司 | Remote sensing image forest land extracting method and system, storage medium, electronic equipment based on deep learning |
| CN109840539A (en) * | 2017-12-31 | 2019-06-04 | 苏州中科天启遥感科技有限公司 | A kind of remote sensing space-time data fusion method based on plot figure spot |
| CN110084294A (en) * | 2019-04-18 | 2019-08-02 | 北京师范大学 | A kind of Remote Image Classification based on multiple dimensioned depth characteristic |
| CN110163303A (en) * | 2019-06-03 | 2019-08-23 | 中国农业大学 | A kind of remote sensing image parallel sorting method and system based on grid |
| WO2020063461A1 (en) * | 2018-09-30 | 2020-04-02 | 广州地理研究所 | Urban extent extraction method and apparatus based on random forest classification algorithm, and electronic device |
-
2020
- 2020-04-10 CN CN202010282973.4A patent/CN111476308B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109840539A (en) * | 2017-12-31 | 2019-06-04 | 苏州中科天启遥感科技有限公司 | A kind of remote sensing space-time data fusion method based on plot figure spot |
| WO2020063461A1 (en) * | 2018-09-30 | 2020-04-02 | 广州地理研究所 | Urban extent extraction method and apparatus based on random forest classification algorithm, and electronic device |
| CN109446992A (en) * | 2018-10-30 | 2019-03-08 | 苏州中科天启遥感科技有限公司 | Remote sensing image building extracting method and system, storage medium, electronic equipment based on deep learning |
| CN109543630A (en) * | 2018-11-28 | 2019-03-29 | 苏州中科天启遥感科技有限公司 | Remote sensing image forest land extracting method and system, storage medium, electronic equipment based on deep learning |
| CN110084294A (en) * | 2019-04-18 | 2019-08-02 | 北京师范大学 | A kind of Remote Image Classification based on multiple dimensioned depth characteristic |
| CN110163303A (en) * | 2019-06-03 | 2019-08-23 | 中国农业大学 | A kind of remote sensing image parallel sorting method and system based on grid |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111476308A (en) | 2020-07-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111652217A (en) | Text detection method and device, electronic equipment and computer storage medium | |
| CN111523414A (en) | Face recognition method and device, computer equipment and storage medium | |
| CN112233240B (en) | Three-dimensional vector data slicing method and device of three-dimensional vector map and electronic equipment | |
| CN110852949B (en) | Point cloud data completion method and device, computer equipment and storage medium | |
| CN116994140A (en) | Cultivated land extraction method, device, equipment and medium based on remote sensing image | |
| CN112949624B (en) | Water gauge-based water level detection method and device, electronic equipment and storage medium | |
| CN110287831B (en) | Method and device for acquiring control point based on landmark and electronic equipment | |
| CN111476308B (en) | Remote sensing image classification method and device based on priori geometric constraint and electronic equipment | |
| CN114254584A (en) | Comparison method, modeling method and device of chip products and storage medium | |
| CN114694005A (en) | Target detection model training method and device, and target detection method and device | |
| CN116682130A (en) | Method, device and equipment for extracting icon information and readable storage medium | |
| CN112527442B (en) | Environment data multi-dimensional display method, device, medium and terminal equipment | |
| CN112669432A (en) | Volume cloud rendering method and device, electronic equipment and storage medium | |
| CN116824090A (en) | Curved surface reconstruction method and device | |
| CN116863349A (en) | Remote sensing image change area determining method and device based on triangular network dense matching | |
| CN112419459B (en) | Method, apparatus, computer device and storage medium for baking model AO mapping | |
| CN115170679A (en) | Calibration method and device for road side camera, electronic equipment and storage medium | |
| CN116167925A (en) | Image restoration method, device, electronic equipment and computer readable storage medium | |
| CN113449620A (en) | Table detection method, device, equipment and medium based on semantic segmentation | |
| CN117152364B (en) | Method, device and equipment for three-dimensional reconstruction of water body boundary based on image data | |
| CN115861561B (en) | Contour line generation method and device based on semantic constraint | |
| CN115937454B (en) | Method and device for automatically placing tree models in large-scale city scene | |
| CN117876711B (en) | Image target detection method, device, equipment and medium based on image processing | |
| CN114625530B (en) | Gaussian beam imaging point rapid searching method and device based on dyeing algorithm | |
| CN117765186B (en) | Reconstruction method, device, equipment and storage medium of environment space |
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 |