CN110334719B - Method and system for extracting building image in remote sensing image - Google Patents
Method and system for extracting building image in remote sensing image Download PDFInfo
- Publication number
- CN110334719B CN110334719B CN201910623705.1A CN201910623705A CN110334719B CN 110334719 B CN110334719 B CN 110334719B CN 201910623705 A CN201910623705 A CN 201910623705A CN 110334719 B CN110334719 B CN 110334719B
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- convolutional neural
- image
- remote sensing
- 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
Images
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for extracting a building image in a remote sensing image, which comprises the following steps: acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output; acquiring a remote sensing image of a region to be acquired; inputting the remote sensing image into the convolutional neural network model, and extracting the building image of the area to be acquired to obtain a primary extraction result; and optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired. And also discloses a specific virtual system for implementing the method. The method and the system for extracting the building image from the remote sensing image have the characteristics of high image extraction precision and high efficiency.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method and a system for extracting a building image in a remote sensing image.
Background
High-resolution remote sensing image building extraction is a key technology in the fields of city change monitoring, city planning, three-dimensional modeling, digital city establishment and the like. With the continuous improvement of the resolution of the remote sensing satellite, higher requirements are put forward on the accuracy of the remote sensing image building extraction.
The remote sensing image building extraction is to perform semantic segmentation on ground objects in an image, namely to obtain building target and edge contour information by a two-pixel classification method. The extraction result intuitively and effectively reflects the position and distribution condition of the building target in the research area, so that the method becomes an important reference basis for evaluation and research in the fields of urban planning and the like. In the high-resolution remote sensing image in the urban area, the surrounding environment of the buildings in the urban area is complicated, the traffic network is dense in crossing, and the confusion and the unclear edge of the information of the edges of the buildings and the roads in the extracted result are easily caused. In addition, the resolution is continuously reduced in the forward propagation process of the convolutional neural network, and the edge effect is not ideal in the extraction result, so that the accuracy of building extraction is limited.
Disclosure of Invention
The invention aims to provide a method and a system for extracting a building image in a remote sensing image, which have the characteristics of high image extraction precision and high efficiency.
In order to achieve the purpose, the invention provides the following scheme:
a method for extracting a building image in a remote sensing image comprises the following steps:
acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
acquiring a remote sensing image of a region to be acquired;
inputting the remote sensing image into the convolutional neural network model, and extracting the building image of the area to be acquired to obtain a primary extraction result;
and optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired.
Optionally, before obtaining the convolutional neural network model, the method further includes:
obtaining a remote sensing image sample;
dividing the remote sensing image into a plurality of sub-images with the same size;
converting the sub-image into a true value image by taking the area pixel of the building as 1 and the area pixel of the non-building as 0;
acquiring a true value map of which the pixel ratio of the building area in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample;
and training the convolutional neural network model by using the training sample.
Optionally, before the obtaining the convolutional neural network model, the method further includes:
and improving the convolutional neural network model.
Optionally, the improving the convolutional neural network model includes:
improving the transverse connection layer in the convolutional neural network model, namely adding the feature map in the previous layer of the corresponding hierarchy for fusion when the feature maps between the corresponding hierarchies of the expansion path and the contraction path are fused;
introducing a Dropout strategy for fixing the node weights in the convolutional neural network model according to a specific probability;
a BN layer is added after each convolutional layer of the convolutional neural network model.
Optionally, the specific probability is 0.25.
A system for extracting an image of a building from a remote sensing image, comprising:
the convolutional neural network model acquisition module is used for acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
the remote sensing image acquisition module is used for acquiring a remote sensing image of an area to be acquired;
the first extraction module is used for inputting the remote sensing image into the convolutional neural network model, extracting the building image of the area to be acquired and obtaining a primary extraction result;
and the second extraction module is used for optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired.
Optionally, the system further includes:
the remote sensing image sample acquisition module is used for acquiring a remote sensing image sample;
the subimage dividing module is used for dividing the remote sensing image into a plurality of subimages with the same size;
the truth-value map conversion module is used for converting the sub-images into true-value maps by taking the pixels of the building areas as 1 and the pixels of the non-building areas as 0;
the training sample acquisition module is used for acquiring a true value map of which the pixel ratio of the building area region in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample;
and the optimization training module is used for performing optimization training on the convolutional neural network model by using the training sample.
Optionally, the system may further include:
and the convolutional neural network model improving module is used for improving the convolutional neural network model.
Optionally, the convolutional neural network model improving module includes:
the transverse connection layer improving unit is used for improving the transverse connection layer in the convolutional neural network model, namely adding a feature map in a layer before a corresponding layer when feature maps between corresponding layers of an expansion path and a contraction path are fused;
a Dropout strategy introducing unit, which is used for introducing a Dropout strategy for fixing the node weight in the convolutional neural network model according to a specific probability;
and the BN layer adding unit is used for adding BN layers after the convolution layers of the convolutional neural network model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the method and the system for extracting the building image in the remote sensing image, provided by the invention, firstly obtain a trained neural network model taking the remote sensing image as input and the building image as output, then input the obtained remote sensing image of the area to be acquired into the convolutional neural network model to extract the building image of the area to be acquired, obtain a primary extraction result, and then optimize the primary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired. In the method and the system for extracting the building image from the remote sensing image, the building image in the remote sensing image is extracted by adopting the convolutional neural network model, so that the efficiency of extracting the building image is improved; and the primary extraction result of the building image obtained by adopting the convolutional neural network model is optimized again, so that the accuracy of the extracted building image is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flowchart of a method for extracting a building image from a remote sensing image according to an embodiment of the present invention;
FIG. 2a is a schematic diagram of a remote sensing image divided according to an embodiment of the present invention;
FIG. 2b is a schematic view of a building labeled in accordance with an embodiment of the present invention;
FIG. 2c is a truth diagram of an embodiment of the present invention after conversion;
FIG. 3 is a schematic diagram of an improved convolutional neural network according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating the results of the preliminary building extraction according to an embodiment of the present invention;
FIG. 5 is a diagram of the final extraction results of the building according to the embodiment of the present invention;
FIG. 6 is a schematic structural diagram of a system for extracting a building image from a remote sensing image according to an embodiment of the present invention;
FIG. 7a is a diagram of the results of a classical convolutional neural network extraction building;
fig. 7b is a diagram of the result of extracting the building by the improved convolutional neural network according to the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method and a system for extracting a building image in a remote sensing image, which have the characteristics of high image extraction precision and high efficiency.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method for extracting a building image from a remote sensing image according to an embodiment of the present invention, and as shown in fig. 1, the method for extracting a building image from a remote sensing image specifically includes the following steps:
a method for extracting a building image in a remote sensing image comprises the following steps:
s1, acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
s2, obtaining a remote sensing image of the area to be acquired;
s3, inputting the remote sensing image into the convolutional neural network model, and extracting the building image of the area to be acquired to obtain a primary extraction result;
and S4, optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired.
In order to further improve the accuracy of extracting the building image by the obtained convolutional neural network model, before obtaining the convolutional neural network model, the method further comprises the following steps:
the method comprises the steps of obtaining preprocessed remote sensing image samples, wherein remote sensing images from a high-resolution second satellite and an L andsat8 satellite are used as remote sensing data sets, the imaging time of the images is 7 months in 2018, and the selected range is the remote sensing images of major urban areas of the State of Massachusetts.
Dividing the remote sensing image into a plurality of sub-images with the same size according to the grid distance of 572 × 572 pixels, as shown in fig. 2 a;
marking the building image in the obtained remote sensing image by means of manual semi-automatic marking, wherein the specific marking mode is as follows: firstly, manually marking the edge outline of a building in each remote sensing image, and extracting sub-images of a building area; then, the pixels of the building existing region in the image are set to 1 as shown in fig. 2b, the pixels of the non-building region are set to 0, and the sub-image is converted into a true value image as shown in fig. 2 c. Acquiring a true value map of which the pixel ratio of the building area in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample; in order to ensure that the positive samples and the negative samples in the training samples are distributed more uniformly, the preset conditions are manually set according to needs.
And training the convolutional neural network model by using the training sample.
In addition, in order to enhance the low-dimensional detail feature information of the feature map in the convolutional neural network transmission process, before the obtaining of the convolutional neural network model, the method further includes improving the convolutional neural network model, and the specific improvement process is as follows:
the classical convolutional neural network model consists of a downward contraction path and an upward expansion path. The input image firstly passes through a plurality of convolution layers and pooling layers to obtain a high-dimensional feature map with lower resolution, then reverse sampling is carried out for a plurality of times through a series of deconvolution to generate a feature map corresponding to the original feature pyramid step by step, and finally a pixel-level prediction result consistent with the resolution of the input image is output. In the process of up-sampling the high-dimensional feature map, the feature to be subjected to dimension reduction is fused with the feature map of the corresponding level in the feature pyramid in a matrix cascade mode, and the fused feature not only contains abstract data of the pyramid top layer, but also injects detail information extracted from each level of the low layer.
The invention further improves on the basis of the idea of the classical convolutional neural network model, and specifically comprises the following steps:
and improving the transverse connection layers in the convolutional neural network model, namely adding the feature maps of the previous layers of the corresponding layers for fusion when each layer (except the last layer) in the expansion path is fused with the feature map of the corresponding layer in the contraction path. Not only is the abstract feature of the layer guaranteed, but also the low-dimensional detail feature with higher resolution is enhanced. Fig. 3 is a schematic diagram of an improved convolutional neural network structure.
As the depth of the network increases, a large number of parameters and calculations can cause an overfitting situation to occur. The invention introduces a Dropout strategy in the network for fixing the node weights in the convolutional neural network model according to a specific probability to prevent overfitting. In the training process, the weights of partial nodes of the convolutional neural network model are not updated randomly according to a certain probability. The method can avoid that certain features only take effect under a fixed combination, destroy the correlation of the fixed combination, consciously enable the convolutional neural network to learn common commonalities, thereby improving the generalization capability of the model and preventing the overfitting of the network.
Specifically, Dropout is expressed as follows:
wherein, each output node does not update the parameter with the probability p, and U is a Bernoulli random number. In the present invention, p is preferably 0.25.
And adding BN layers after each convolution layer of the convolution neural network model, so that the training speed is increased and the generalization capability of the network is improved while the characteristics learned by the upper layer network are recovered.
Wherein, the BN network layer is a normalization layer and is a learnable X-X (X) with parameters (gamma, β)(1)...x(d)) Assuming that the layer has d-dimensional input, learning parameters (gamma, β) are introduced, and each dimensional feature is normalized by adopting a transformation reconstruction method:
when the improved convolutional neural network model is used for building extraction, since the image of the input image becomes smaller after each convolution operation, the SAME padding operation is used to fill 0 in the lost pixel position, so that the final output image is consistent with the input image in size, and a preliminary extraction result is obtained, as shown in fig. 4.
And in the step of optimizing the preliminary extraction result by adopting morphological closed operation to obtain the final extraction result of the building image in the area to be acquired, the morphological closed operation is mainly used for filling tiny holes and cracks formed due to missing detection. The final extraction results after filling are shown in fig. 5.
Fig. 6 is a schematic structural diagram of a system for extracting a building image from a remote sensing image according to an embodiment of the present invention, and as shown in fig. 6, a system for extracting a building image from a remote sensing image includes:
the convolutional neural network model acquisition module 1 is used for acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
the remote sensing image acquisition module 2 is used for acquiring a remote sensing image of an area to be acquired;
the first extraction module 3 is used for inputting the remote sensing image into the convolutional neural network model, extracting the building image of the area to be acquired and obtaining a primary extraction result;
and the second extraction module 4 is used for optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired.
The system further comprises:
the remote sensing image sample acquisition module is used for acquiring a remote sensing image sample;
the subimage dividing module is used for dividing the remote sensing image into a plurality of subimages with the same size;
the truth-value map conversion module is used for converting the sub-images into true-value maps by taking the pixels of the building areas as 1 and the pixels of the non-building areas as 0;
the training sample acquisition module is used for acquiring a true value map of which the pixel ratio of the building area region in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample;
and the optimization training module is used for performing optimization training on the convolutional neural network model by using the training sample.
The system may further include:
and the convolutional neural network model improving module is used for improving the convolutional neural network model.
The convolutional neural network model improvement module comprises:
the transverse connection layer improving unit is used for improving a transverse connection layer in the convolutional neural network model, namely adding a feature map in a previous layer corresponding to a current hierarchy to be fused in the contraction path for fusion when feature maps between corresponding levels of the expansion path and the contraction path are fused;
a Dropout strategy introducing unit, which is used for introducing a Dropout strategy for fixing the node weight in the convolutional neural network model according to a specific probability;
and the BN layer adding unit is used for adding BN layers after the convolution layers of the convolutional neural network model.
In addition, three indexes of an intersection ratio (IoU), a detection accuracy (pixel accuracy) and a kappa coefficient are used for evaluating the extraction result, the three indexes respectively represent the overall precision level of the detection result, the proportion coefficient of a correct part and the consistency with the truth value, and the calculation formula is as follows:
in the formula, TP is a pixel for correctly detecting a building, FP is a false detection pixel, FN is a missing detection pixel, and TN is a background pixel for correctly detecting. Through the calculation, the effect of extracting the building by using the classical convolutional neural network model method and the system for extracting the building image in the remote sensing image, which are provided by the invention, is obtained, and is shown in table 1; the result of extracting the building image by the above method is shown in fig. 7a and 7b, respectively, where green is the correctly detected pixel in fig. 7a and 7 b; red is a false detection pixel; blue is the missing pixel.
Table I, result quantitative analysis table
On the basis of the quantitative analysis comparison result shown in table one, by comparing fig. 7a with fig. 7b, it can be obtained that the result of extracting the building image in the remote sensing image by using the improved convolutional neural network model can be better optimized, and the extraction accuracy of the edge detail information is also improved to a certain extent.
Therefore, according to the method and the system for extracting the building image in the remote sensing image, the trained neural network model which takes the remote sensing image as input and the building image as output is obtained, the obtained remote sensing image of the area to be acquired is input into the convolutional neural network model to extract the building image of the area to be acquired, a preliminary extraction result is obtained, and then the preliminary extraction result is optimized by adopting morphological closed operation, so that a final extraction result of the building image in the area to be acquired is obtained. In the method and the system for extracting the building image from the remote sensing image, the building image in the remote sensing image is extracted by adopting the convolutional neural network model, so that the efficiency of extracting the building image is improved; and the primary extraction result of the building image obtained by adopting the convolutional neural network model is optimized again, so that the accuracy of the extracted building image is improved.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (5)
1. A method for extracting a building image in a remote sensing image is characterized by comprising the following steps:
acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
acquiring a remote sensing image of a region to be acquired;
inputting the remote sensing image into the convolutional neural network model, and extracting the building image of the area to be acquired to obtain a primary extraction result;
optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired;
before the obtaining the convolutional neural network model, further comprising:
improving the convolutional neural network model;
the improving the convolutional neural network model comprises:
improving the transverse connection layer in the convolutional neural network model, namely adding the feature map in the previous layer of the corresponding hierarchy for fusion when the feature maps between the corresponding hierarchies of the expansion path and the contraction path are fused;
introducing a Dropout strategy for fixing the node weights in the convolutional neural network model according to a specific probability;
a BN layer is added after each convolutional layer of the convolutional neural network model.
2. The method for extracting the building image in the remote sensing image according to claim 1, wherein before obtaining the convolutional neural network model, the method further comprises:
obtaining a remote sensing image sample;
dividing the remote sensing image into a plurality of sub-images with the same size;
converting the sub-image into a true value image by taking the area pixel of the building as 1 and the area pixel of the non-building as 0;
acquiring a true value map of which the pixel ratio of the building area in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample;
and training the convolutional neural network model by using the training sample.
3. The method for extracting the building image in the remote sensing image according to claim 1, wherein the specific probability is 0.25.
4. A system for extracting building images from remote sensing images, comprising:
the convolutional neural network model acquisition module is used for acquiring a convolutional neural network model; the convolutional neural network model is a trained neural network model which takes the remote sensing image as input and takes the building image as output;
the remote sensing image acquisition module is used for acquiring a remote sensing image of an area to be acquired;
the first extraction module is used for inputting the remote sensing image into the convolutional neural network model, extracting the building image of the area to be acquired and obtaining a primary extraction result;
the second extraction module is used for optimizing the preliminary extraction result by adopting morphological closed operation to obtain a final extraction result of the building image in the area to be acquired;
the convolutional neural network model improving module is used for improving the convolutional neural network model;
the convolutional neural network model improvement module comprises:
the transverse connection layer improving unit is used for improving the transverse connection layer in the convolutional neural network model, namely adding the feature map in the previous layer of the corresponding hierarchy for fusion when the feature maps between the corresponding hierarchies of the expansion path and the contraction path are fused;
a Dropout strategy introducing unit, which is used for introducing a Dropout strategy for fixing the node weight in the convolutional neural network model according to a specific probability;
and the BN layer adding unit is used for adding BN layers after the convolution layers of the convolutional neural network model.
5. The system for extracting building images from remote sensing images as claimed in claim 4, further comprising:
the remote sensing image sample acquisition module is used for acquiring a remote sensing image sample;
the subimage dividing module is used for dividing the remote sensing image into a plurality of subimages with the same size;
the truth-value map conversion module is used for converting the sub-images into true-value maps by taking the pixels of the building areas as 1 and the pixels of the non-building areas as 0;
the training sample acquisition module is used for acquiring a true value map of which the pixel ratio of the building area region in the true value map is lower than a preset condition, and taking the acquired true value map as a training sample;
and the optimization training module is used for performing optimization training on the convolutional neural network model by using the training sample.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910623705.1A CN110334719B (en) | 2019-07-11 | 2019-07-11 | Method and system for extracting building image in remote sensing image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910623705.1A CN110334719B (en) | 2019-07-11 | 2019-07-11 | Method and system for extracting building image in remote sensing image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110334719A CN110334719A (en) | 2019-10-15 |
CN110334719B true CN110334719B (en) | 2020-07-28 |
Family
ID=68146244
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910623705.1A Active CN110334719B (en) | 2019-07-11 | 2019-07-11 | Method and system for extracting building image in remote sensing image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110334719B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598101B (en) * | 2020-05-25 | 2021-03-23 | 中国测绘科学研究院 | Urban area intelligent extraction method, system and equipment based on remote sensing image scene segmentation |
CN111985372A (en) * | 2020-08-12 | 2020-11-24 | 兰州交通大学 | Remote sensing image water body extraction system for deep learning |
CN112883887B (en) * | 2021-03-01 | 2023-07-18 | 中央财经大学 | Building instance automatic extraction method based on high spatial resolution optical remote sensing image |
CN113112589B (en) * | 2021-04-13 | 2022-09-02 | 哈尔滨工程大学 | Three-dimensional reconstruction method of incremental remote sensing image based on space occupation probability fusion |
CN114387523B (en) * | 2022-03-23 | 2022-06-03 | 成都理工大学 | Remote sensing image building extraction method based on DCNN boundary guidance |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108537192A (en) * | 2018-04-17 | 2018-09-14 | 福州大学 | A kind of remote sensing image ground mulching sorting technique based on full convolutional network |
CN108681692A (en) * | 2018-04-10 | 2018-10-19 | 华南理工大学 | Increase Building recognition method in a kind of remote sensing images based on deep learning newly |
CN109871798A (en) * | 2019-02-01 | 2019-06-11 | 浙江大学 | A kind of remote sensing image building extracting method based on convolutional neural networks |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108629816B (en) * | 2018-05-09 | 2022-11-18 | 复旦大学 | Method for reconstructing thin-layer magnetic resonance image based on deep learning |
CN109685813B (en) * | 2018-12-27 | 2020-10-13 | 江西理工大学 | U-shaped retinal vessel segmentation method capable of adapting to scale information |
-
2019
- 2019-07-11 CN CN201910623705.1A patent/CN110334719B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108681692A (en) * | 2018-04-10 | 2018-10-19 | 华南理工大学 | Increase Building recognition method in a kind of remote sensing images based on deep learning newly |
CN108537192A (en) * | 2018-04-17 | 2018-09-14 | 福州大学 | A kind of remote sensing image ground mulching sorting technique based on full convolutional network |
CN109871798A (en) * | 2019-02-01 | 2019-06-11 | 浙江大学 | A kind of remote sensing image building extracting method based on convolutional neural networks |
Non-Patent Citations (4)
Title |
---|
"基于U型卷积神经网络的航空影像建筑物检测";伍广明等;《测绘学报》;20180630;第865-870页 * |
"组合全卷积神经网络和条件随机场的道路分割";宋青松等;《清华大学学报》;20181231;第1节 * |
Dropout Rademacher complexity of deep neural networks;Wei Gao等;《Science China Information Sciences》;20160731;论文正文 * |
伍广明等."基于U型卷积神经网络的航空影像建筑物检测".《测绘学报》.2018, * |
Also Published As
Publication number | Publication date |
---|---|
CN110334719A (en) | 2019-10-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110334719B (en) | Method and system for extracting building image in remote sensing image | |
CN110619282B (en) | Automatic extraction method for unmanned aerial vehicle orthoscopic image building | |
Yang et al. | Building extraction at scale using convolutional neural network: Mapping of the united states | |
CN113449594B (en) | Multilayer network combined remote sensing image ground semantic segmentation and area calculation method | |
CN113160234B (en) | Unsupervised remote sensing image semantic segmentation method based on super-resolution and domain self-adaptation | |
CN110889449A (en) | Edge-enhanced multi-scale remote sensing image building semantic feature extraction method | |
CN113628180B (en) | Remote sensing building detection method and system based on semantic segmentation network | |
CN111291826A (en) | Multi-source remote sensing image pixel-by-pixel classification method based on correlation fusion network | |
CN113379771B (en) | Hierarchical human body analysis semantic segmentation method with edge constraint | |
CN112347970A (en) | Remote sensing image ground object identification method based on graph convolution neural network | |
CN113610024B (en) | Multi-strategy deep learning remote sensing image small target detection method | |
CN113313094B (en) | Vehicle-mounted image target detection method and system based on convolutional neural network | |
CN113361528B (en) | Multi-scale target detection method and system | |
CN113255589A (en) | Target detection method and system based on multi-convolution fusion network | |
CN114973011A (en) | High-resolution remote sensing image building extraction method based on deep learning | |
CN116343103A (en) | Natural resource supervision method based on three-dimensional GIS scene and video fusion | |
CN115424017B (en) | Building inner and outer contour segmentation method, device and storage medium | |
CN115578615A (en) | Night traffic sign image detection model establishing method based on deep learning | |
CN115424059A (en) | Remote sensing land use classification method based on pixel level comparison learning | |
CN113361496B (en) | City built-up area statistical method based on U-Net | |
CN111881915A (en) | Satellite video target intelligent detection method based on multiple prior information constraints | |
CN110717960B (en) | Method for generating building rubbish remote sensing image sample | |
CN117078925A (en) | Building rubbish annual output accurate calculation method based on RDSA-deep LabV3+ network | |
CN114913475B (en) | Urban gridding management method and system based on GIS and machine vision | |
CN114445726B (en) | Sample library establishing method and device based on deep learning |
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 |