CN111767801A - Remote sensing image water area automatic extraction method and system based on deep learning - Google Patents
Remote sensing image water area automatic extraction method and system based on deep learning Download PDFInfo
- Publication number
- CN111767801A CN111767801A CN202010493489.6A CN202010493489A CN111767801A CN 111767801 A CN111767801 A CN 111767801A CN 202010493489 A CN202010493489 A CN 202010493489A CN 111767801 A CN111767801 A CN 111767801A
- Authority
- CN
- China
- Prior art keywords
- water area
- remote sensing
- sensing image
- feature
- scale
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (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)
- Multimedia (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a remote sensing image water area automatic extraction method and system based on deep learning. Preprocessing remote sensing image data, obtaining different water area indexes through band operation, obtaining prior characteristic information extracted from a water area, fusing the remote sensing image data, Google map tile data and the like to realize multi-source characteristic information fusion, and then constructing a data set through visual interpretation and vectorization; training, verifying and testing a semantic segmentation model WE-Net built by a convolutional neural network; and calling a remote sensing image water area segmentation model WE-Net to realize automatic classification of the water area, and outputting a binary gray level map which is a classification and extraction result. The invention has the beneficial effects that: the water area in the research area can be extracted by calling the remote sensing image water area segmentation model, manual visual interpretation can be replaced, manpower and material resources are saved, and auxiliary technical support is provided for updating of a high-precision image map, including lake area change detection, water system transition and the like.
Description
Technical Field
The invention relates to the field of geographic information, in particular to the field of surface water areas, and particularly relates to a remote sensing image water area automatic extraction method and system based on deep learning.
Background
Rivers and lakes are the most common expression forms of surface water areas, are increased or reduced due to changes of factors such as climate change, land utilization, crustal activities and the like, and have important significance for detecting changes of the surface water areas, protecting and recovering wetland ecosystem, protecting aquatic animals and plants, supervising rivers, controlling pollution and the like. With the continuous development of remote sensing technology, remote sensing images gradually become an effective means for extracting surface water area changes. The traditional remote sensing image water body extraction method usually adopts manual visual interpretation, needs manual description, and is time-consuming and labor-consuming although the precision is very high. In addition, the single-band threshold value method and the water body index method have the problems of manual threshold value determination, same-spectrum foreign matter, low automation degree, poor real-time performance and the like, and the water domain range of a research area is difficult to acquire quickly in time. Subsequently, machine learning methods such as machine learning, support vector machine and K-means algorithm are widely applied to water extraction, but the problems of low precision, weak generalization capability and the like still exist.
With the continuous development of smart city construction, the requirement for automatic extraction of ground objects is continuously improved, and the traditional remote sensing water body extraction method obviously cannot meet the requirement, so that a mode with high precision, simple operation and low cost is urgently needed for realizing automatic classification and extraction of water areas. The rapid development of the deep learning technology, especially the application of the convolutional neural network in computer vision, makes the fields of target detection and semantic segmentation in image processing obtain great success, and directly promotes the research of applying the deep learning to the remote sensing field to solve the problems of ground feature classification, detection, extraction and the like. The water area is extracted in real time with high precision by a deep learning technology and combining a traditional water body index method. Finally, the whole remote sensing image water area extraction method is set as a set of complete water area automatic extraction method and system, and technical support and data support are provided for scientific research and project practice related to water area change detection.
Disclosure of Invention
The invention aims to solve the technical problem that the remote sensing image water area extraction method in the prior art is time-consuming and labor-consuming, and provides a remote sensing image water area automatic extraction method and system based on deep learning.
The invention solves the technical problem, and adopts the technical principle that: the invention discloses a remote sensing image water area automatic extraction method model based on deep learning, which is called WE-Net, and realizes automatic identification of a water area in a remote sensing image through training, testing and calling a remote sensing image water area segmentation model WE-Net. The automatic high-resolution remote sensing image classification method comprises the following steps: p1: firstly, preprocessing remote sensing image data, including radiation correction, geometric correction and research area cutting of the remote sensing image; p2: obtaining different water area indexes through band operation, and obtaining prior characteristic information extracted from a water area; p3: fusing remote sensing image data, Google map tile data and the like to realize multi-source characteristic information fusion, and then constructing a data set through visual interpretation and vectorization; training, verifying and testing a semantic segmentation model WE-Net built by a convolutional neural network P4; p5: calling a remote sensing image water area segmentation model WE-Net to realize automatic classification of the water area, and outputting a png-format binary gray level image which is a classification and extraction result; p6: and fine-tuning the classification result through a long-distance conditional random field. According to the method, only basic image processing and characteristic information fusion are needed to be carried out on spectral data and radar data of the multiband remote sensing image, the remote sensing image water area segmentation model is called, the water area in the research area can be extracted, the classification accuracy can reach 92.64% through expanding a data set in application, manual visual interpretation can be replaced, manpower and material resources are saved, and auxiliary technical support is provided for updating of a high-precision image map, including lake area change detection, water system transition and the like.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of a method for automatically extracting a water area of a remote sensing image based on deep learning according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an implementation block diagram of a remote sensing image water area segmentation model WE-Net for constructing the semantic segmentation of the remote sensing image in the embodiment of the present invention;
FIG. 3 is a block diagram of an embodiment of the present invention for constructing a residual error learning module RLU;
FIG. 4 is a block diagram of an embodiment of the invention for constructing a global attention module GAB;
fig. 5 is a block diagram of an implementation of the boundary learning module BLU according to the embodiment of the present invention.
Detailed Description
For a more clear understanding of the technical features, objects and effects of the present invention, embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
The embodiment of the invention provides a remote sensing image water area automatic extraction method and system based on deep learning.
In a first embodiment, the present embodiment is described with reference to fig. 1, and the method for automatically extracting a remote sensing image water area based on deep learning in the present embodiment includes the following steps:
step (1), downloading sentinel-2 data of European and space Bureau (S2A MSIL1C), opening a CMD console, performing atmospheric correction through a command L2A _ Process in a Sen2cor, and performing resampling (with filter- > geometric operations- > resampling) on the corrected data through SNAP software to obtain data of each waveband of a remote sensing image which can be processed by ENVI5.3 software.
Step (2), calculating normalized difference water body index NDWI (NDWI-NIR)/(Green + NIR)), improved water body index model NDWI3(NDWI 3-NIR-2)/(NIR + SWIR-2)), improved normalized difference water body index MNDWI (MNDWI-SWIR-1)/(Green + SWIR-1)), enhanced water body index EWI (EWI-NIR-SWIR-1)/(Green + NIR + SWIR-1)), normalized vegetation coverage index NDVI (NDVI-NIR-Red)/(NIR + Red)), wetland forest index WFI (WFI-NIR-Red)/SWIR-2)), Red Green Blue (Red, Green, Blue) three visible light bands, and near infrared band through a band calculation tool of remote sensing image processing software ENVI5.3, And the data or index data of each of the intermediate infrared SWIR-1 and SWIR-2 wave bands are output as a gray scale map, and the gray scale maps are 12 in total.
And (3) building a personal geographic database- > building a new element data set- > building a new surface vector file water.shp through ArcGIS software, loading red, green and blue three wave bands to form a true color image and the 12 gray maps, combining tile data of Google maps, vectorizing a water area distribution region according to a remote sensing visual interpretation method, storing the water area distribution region in the surface vector file water.shp, converting the water.shp file containing the true water area into a raster file through a tool ToRaster in the ArcToolBox, finally outputting the gray map water.png in a png format, binarizing the gray map water.png to enable the pixel value of the water area to be 1, the pixel value of the non-water area to be 0, and obtaining the binarized water.png file which is the label file of the manufactured remote sensing image area.
Step (4), invoking an imread function in an opencv-python library function in python to read the 12 gray-scale images and the 1 label file, and cutting the 12 gray-scale images and the 1 label file according to the step length of 128 and the image size of 256 in a one-to-one correspondence manner, so that the cut images with the size of 256 × 1 are respectively stored in 13 folders; calling an imgauge library function to perform one-to-one conversion and augmentation on the cut image according to a data enhancement method, such as cutting, rotation, mirror image change, Gaussian noise and the like, so as to expand a data set; finally, the pixel mean value and the standard deviation of all the data are counted, and the data are standardized; and then dividing the plurality of images after normalization processing to obtain a training set, a verification set and a test set.
Calling convolution layers, pooling layers, up-sampling layers, loss functions and activation functions in a deep learning framework TensorFlow and Keras to build a remote sensing image water area segmentation model WE-Net based on deep learning, wherein the segmentation model has 13 inputs in the process of training a sample, 12 gray-scale graphs with the respective dimension equal to 256 x 1 and a label file which corresponds to each gray-scale graph after binarization; the segmentation model WE-Net is implemented by an encoding step, a decoding step, a residual learning module step, a global attention module step, and a boundary learning unit step, which will be described in detail in the second embodiment.
Step (6), setting training batch size to be 16 and learning rate parameter learning to be 0.001 according to the calculation performance and model parameter number of the two NVIDIA GTX 1080Ti display cards, calling a train function, performing multi-round iterative training on a remote sensing image water area segmentation model WE-Net by using the training set, and performing iterative verification on the model after each round of training by using the verification set; the training process is visualized by taking the number of training rounds as a horizontal axis and the IOU value as a vertical axis, after dozens of rounds of training, the IOU rises firstly and then approaches a certain IOU value in a wireless mode, finally, fluctuation in a small range is kept near the value, and in the next dozens of rounds of training, the IOU value does not change along with the increase of the number of rounds, the segmentation model of the water area of the remote sensing image is considered to be converged, model parameters are stored, the training is stopped, and overfitting is prevented; otherwise, if the IOU of the training set and the verification set is changed continuously, returning to the step (4) to modify the batch and learning rate parameters, and loading the training set for retraining; and finally, calling the stored remote sensing image water area segmentation model WE-Net through a test function, and evaluating the precision of the remote sensing image water area segmentation model according to the IOU value calculated on the test set. In the embodiment, the precision evaluation index IOU of the remote sensing image water area segmentation model WE-Net reaches 0.9401 in the training set, the verification set reaches 0.9326, the model is stored after the precision evaluation index IOU does not descend, and finally a test function is called to detect that the precision IOU reaches 0.9264 in the test set.
Step (7), after the result of automatic extraction of the water area is output by the remote sensing image water area segmentation model, post-processing the result by a guide filter GF and a conditional random field model CRF; the label file is regarded as a guide graph by guide filtering, an original image is taken as an input image, and the boundary of a water area extraction result is optimized so as to eliminate salt and pepper noise; the binary potential function in the conditional random field restrains the color and the position between any two pixel points, so that the pixel points with similar color and adjacent positions can have the same classification more easily, and meanwhile, the smoothness between the adjacent pixel points is considered, the edge is smoothed, and the semantic segmentation result is finely adjusted.
And (8) after a satisfactory result is obtained by training and testing the remote sensing image water area segmentation model WE-Net, saving the weight parameter and the network model as WE-Net.h5, which is a weight file saved by ME-Net after training. The method comprises the steps that a local machine is used as a server to issue REST service through a flash framework; the client converts the remote sensing image into base64 format characters (img src ═ data: image/png) through a base64 coding tool base64.b64encode (); base64, "/>, passing the character to the local server via a post request; the server responds to a post request, data are obtained through request.get _ data (), a remote sensing image is decoded through base64.b64decode (), a remote sensing image water area segmentation model WE-Net and a post-processing algorithm are called to guide filtering GF and a conditional random field CRF, automatic water area extraction is achieved, and an extraction result is coded and returned through base64.
In a second embodiment, the present embodiment is described with reference to fig. 2, 3, 4 and 5, and the remote sensing image water area segmentation model WE-Net according to the present embodiment includes the following steps:
and (3) encoding: in the encoding stage, characteristic information of the water area is extracted through convolution and pooling. In the encoding stage, 12 gray maps are used as input data, feature maps are obtained by convolution and fusion of the input data, each feature map passes through one pooling layer to be one scale, and the feature maps comprise 5 scales of the original map, namely 256 × 32, 128 × 64, 64 × 128, 32 × 256 and 16 × 16 512; after the image passes through the pooling layer, the size of the feature map is reduced by half, the number of channels is doubled, and the water area feature information of the image is extracted through two convolutional neural networks.
And (3) decoding: in the decoding stage, the size of the image is restored through convolution and 4 times of up-sampling, and the result of water area extraction is obtained. Every time the feature map in the decoding stage is sampled once, the feature maps with the same size and size corresponding to the encoding stage are fused through the global attention module, and then boundary texture information of a water area part in the feature maps is integrated and extracted through the boundary learning unit; finally outputting a binary gray-scale map with the scale size of 256 × 1, wherein if the value of the gray-scale map is 1, the gray-scale map represents a water area part, and if the value of the gray-scale map is 0, the gray-scale map represents a non-water area part; wherein the loss function is set as a binary cross entropy loss function.
A residual error learning module: in the stage of a residual error learning module, a quick connection is added to the network, the speed of information circulation and the efficiency of network training are improved, and meanwhile, two convolution layers are added to improve the capability of extracting characteristic information of the model. The residual error learning module takes a feature map with the coding stage scale of 2w × 2h × c as input, the result of the feature map after being subjected to two times of convolution by c convolution kernels with 3 × c is directly added with the original feature map according to pixel one-to-one correspondence sum, then the sum is transformed and activated through softmax, the path is called as shortcut connection, and the finally obtained feature map is consistent with the scale of the original feature map and is 2w × 2h × c; wherein w, h and c represent the width, height and channel number of the characteristic diagram in turn.
Global attention module: in the global attention module stage, semantic segmentation information of the decoding stage and position information of the encoding stage are fused, and feature extraction information is compressed and enhanced in a global average pooling weighting mode. The global attention module takes the feature graph with the coding stage scale of 2w × 2h × c and the decoding stage scale of w × h × 2c as input, firstly obtains feature information after global average pooling of the feature graph of the decoding stage, then obtains new feature graph by performing multiplex weighting on the feature graph with the feature information as the coding stage, directly adds the new feature graph to the feature graph of the coding stage according to pixel sum, then changes the dimension after sampling on the feature graph of the decoding stage into 2w × 2h × 2c, and finally splices the feature graph obtained after sampling on the unsample and the feature graph after weighting according to sum of sum according to channels to obtain a fused feature graph, wherein the scale of the feature graph is 2w × 2h × 3 c.
A boundary learning module: in the stage of a boundary learning unit, a residual error learning module is formed by convolutional layers with different scales and shortcut connection shortcuts, and a convolution kernel is added on a branch, so that the aliasing effect generated in the process of upsampling and fusing feature maps with different scales due to the feature maps is eliminated, and the remote sensing image feature information with different scales is learned. The boundary learning unit takes the feature graph output by the global attention module stage as input, the feature graph needs to be subjected to information circulation through three different branches, the first branch is a shortcut connection and does not perform data transformation, the second branch changes the scale size of the feature graph from 2w 2h 3c to 2w 2h c after being subjected to c convolution with the size of 3 x 3c, the third branch changes the scale size of the feature graph from 2w 2h 3c to 2w 2h c after being subjected to c convolution with the size of 3 x 3c, and the feature graphs of the three branches are subjected to aliasing addition according to pixels to obtain the feature graph which is eliminated finally.
According to another aspect of the present invention, to solve the technical problem, the present invention further provides a system for automatically extracting a remote sensing image water area based on deep learning, comprising the following modules:
the data preprocessing module is used for downloading sentinel-2 data of the European space agency, performing atmospheric correction through a command in a Sen2cor, and resampling the corrected data through SNAP software to obtain data of each wave band of a remote sensing image;
the information extraction module is used for calculating a normalized difference water body index NDWI, an improved water body index model NDWI3, an improved normalized difference water body index MNDWI, an enhanced water body index EWI, a novel water body index NWI, a normalized vegetation coverage index NDVI, a wetland forest index, three visible light wave bands of Red, Green and Blue (Red, Green and Blue), near infrared NIR, middle infrared SWIR-1 and SWIR-2 wave bands through a wave band operation tool of remote sensing image processing software ENVI5.3, and data or index data of each wave band are output as a gray scale image;
the system comprises a tag file making module, a remote sensing image water area distribution area generating module, a remote sensing image water area vector generating module, a tag file generating module and a remote sensing image water area vector generating module, wherein the tag file making module is used for newly creating a water area vector file water.shp through ArcGIS software, loading 12 gray maps, vectorizing a water area distribution area according to a remote sensing visual interpretation method, finally outputting a gray map water.png, binarizing the gray map water.png, and obtaining a binarized water area vector file which is a tag file of the water area distribution area of the remote;
the data set generation module is used for calling a library function of opencv-python in python to read the 12 gray-scale images and the label files, and cutting the 12 gray-scale images and the label files in a one-to-one correspondence mode according to the step length of 128 and the image size of 256, so that the cut images with the size of 256 x 1 are respectively stored in 13 folders; calling an imgauge library function to correspondingly transform and amplify the cut images one by one according to a data enhancement method so as to expand a data set; finally, the pixel mean value and the standard deviation of all the data are counted, and the data are standardized; then dividing the plurality of pictures after normalization processing to obtain a training set, a verification set and a test set;
the classification model establishing module is used for calling a convolution layer, a pooling layer, an up-sampling layer, a loss function and an activation function in a deep learning framework TensorFlow and Keras so as to establish a remote sensing image water area segmentation model WE-Net based on deep learning, wherein the segmentation model has 13 inputs in the process of training a sample, and the inputs are 12 gray level images and a corresponding label file respectively; the segmentation model WE-Net is realized by the following steps of encoding, decoding, global attention module and boundary learning unit;
the coding module: the device is used for extracting characteristic information of a water area through convolution and pooling in an encoding stage; in the encoding stage, 12 gray maps are used as input data, feature maps are obtained by convolution and fusion of the input data, each feature map passes through one pooling layer to be one scale, and the feature maps comprise 5 scales of the original map, namely 256 × 32, 128 × 64, 64 × 128, 32 × 256 and 16 × 16 512; after the image passes through the pooling layer, the size of the feature map is reduced by half, the number of channels is doubled, and the water area feature information of the image is extracted through two convolutional neural networks;
a decoding module: the method is used for recovering the image size through convolution and 4 times of up-sampling in a decoding stage to obtain a water area extraction result. Every time the feature map in the decoding stage is sampled once, the feature maps with the same size and size corresponding to the encoding stage are fused through the global attention module, and then boundary texture information of a water area part in the feature maps is integrated and extracted through the boundary learning unit; finally outputting a binary gray-scale map with the scale size of 256 × 1, wherein if the value of the gray-scale map is 1, the gray-scale map represents a water area part, and if the value of the gray-scale map is 0, the gray-scale map represents a non-water area part;
a residual error learning module: the method is used for adding a quick connection to the network in the residual error learning module stage, so that the information circulation speed and the network training efficiency are improved, and meanwhile, two convolution layers are added to improve the capability of extracting the characteristic information of the model. The residual error learning module takes a feature map with the coding stage scale of 2w × 2h × c as input, the result of the feature map after passing through c convolution kernels with 3 × c twice is directly added with the original feature map in a one-to-one correspondence mode according to pixels, then transformation and activation are carried out through softmax, the path is called as shortcut connection, and the finally obtained feature map is consistent with the original feature map in scale and is 2w × 2h × c.
Global attention module: and the method is used for fusing semantic segmentation information of a decoding stage and position information of an encoding stage in a global attention module stage, and compressing and enhancing feature extraction information in a global average pooling weighting mode. The global attention module takes the feature graph with the coding stage scale of 2w x 2h x c and the decoding stage scale of w x h x 2c as input, firstly obtains feature information after the feature graph in the decoding stage is subjected to global average pooling, then weights the feature graph with the feature information as a weight value to obtain a new feature graph, then changes the dimension after the feature graph in the decoding stage is subjected to upsampling into 2w x 2h x 2c, and finally splices the feature graph obtained after the upsampling and the weighted feature graph according to channels to obtain a fused feature graph, wherein the scale of the feature graph is 2w x 2h x 3 c.
A boundary learning module: the method is used for forming a residual error learning module through convolution layers with different scales and shortcut connection short in a boundary learning unit stage, and adding a convolution kernel on a branch path, so that aliasing effect generated in the process of upsampling and fusing feature maps with different scales due to the feature maps is eliminated, and remote sensing image feature information with different scales is learned. The boundary learning unit takes the feature graph output by the global attention module stage as input, the feature graph needs to be subjected to information circulation through three different branches, the first branch is a shortcut connection and does not perform data transformation, the second branch changes the scale size of the feature graph from 2w 2h 3c to 2w 2h c after being subjected to c convolution with the size of 3 x 3c, the third branch changes the scale size of the feature graph from 2w 2h 3c to 2w 2h c after being subjected to c convolution with the size of 3 x 3c, and the feature graphs of the three branches are subjected to aliasing addition according to pixels to obtain the feature graph which is eliminated finally.
The model training module is used for setting training batch batchsize and learning rate parameter learning according to the calculation performance and model parameter number of the display card, calling a train function, performing iterative training on the remote sensing image water area segmentation model WE-Net by using the training set, and verifying the model after each round of training by using the verification set; the training process is visualized by taking the number of training rounds as a horizontal axis and the IOU value as a vertical axis, after dozens of rounds of training, the IOU rises firstly and then approaches a certain IOU value in a wireless mode, finally, fluctuation in a small range is kept around the value, and in the next dozens of rounds of training, the IOU value does not change along with the increase of the number of rounds, the model is considered to be converged, model parameters are stored, the training is stopped, and overfitting is prevented; otherwise, if the IOU of the training set and the verification set is continuously changed, the step S4 is returned to modify the batch and learning rate parameters, and the training set is loaded for retraining; and finally, calling the stored remote sensing image water area segmentation model WE-Net through a test function, and evaluating the precision of the model according to the IOU value calculated on the test set.
The model fine-tuning module is used for performing post-processing on the result through the guide filter GF and the conditional random field model CRF after the trained remote sensing image water area segmentation model outputs the result of automatic water area extraction, namely the segmentation result; the label file is regarded as a guide graph by guide filtering, an original image is taken as an input image, and the boundary of a water area extraction result is optimized so as to eliminate salt and pepper noise; the binary potential function in the conditional random field restrains the color and the position between any two pixel points, so that the pixel points with similar color and adjacent positions can have the same classification more easily, and meanwhile, the smoothness between the adjacent pixel points is considered, the edge is smoothed, and the semantic segmentation result is finely adjusted.
The model application module is used for training and testing the remote sensing image water area segmentation model WE-Net to obtain a satisfactory result, storing the weight parameter and the network model WE-Net.h5, and issuing REST service through a flash framework by taking the machine as a server; the client converts the remote sensing image into base64 format characters through a base64 encoding tool, and transmits the characters to the local server through a post request; and the server responds to the post request, decodes the remote sensing image, calls a remote sensing image water area segmentation model WE-Net and a post-processing algorithm guide filter GF and a conditional random field CRF to realize automatic water area extraction, and then returns an extraction result to the client through base64 coding.
The invention has the beneficial effects that: the method has the advantages that only basic image processing and characteristic information fusion are needed to be carried out on spectral data and radar data of the multiband remote sensing image, the remote sensing image water area segmentation model is called, the water area in the research area can be extracted, the classification accuracy can reach 92.64% through the expansion data set in the application of the model, manual visual interpretation can be replaced, manpower and material resources are saved, and auxiliary technical support is provided for updating of a high-precision image map, including lake area change detection, water system transition and the like.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. A remote sensing image water area automatic extraction method based on deep learning is characterized in that: the method comprises the following steps:
s1: atmospheric correction is carried out on the spectral data of a certain remote sensing image, and the corrected data are resampled to obtain data of each wave band of the remote sensing image;
s2: normalizing the data of each wave band of the remote sensing image by a wave band operation tool of remote sensing image processing software, respectively calculating a normalized difference water body index NDWI, an improved water body index model NDWI3, an improved normalized difference water body index MNDWI, an enhanced water body index EWI, a novel water body index NWI, a normalized vegetation coverage index NDVI and a wetland forest index, identifying three visible light wave bands of red, green and blue, and near infrared NIR, middle infrared SWIR-1 and SWIR-2 wave bands, and outputting the data of each wave band or the data of the indexes into a gray scale map, thereby obtaining 12 gray scale maps;
s3: newly building a water area vector file through ArcGIS software, loading the 12 gray maps, vectorizing a water area distribution area according to a remote sensing visual interpretation method, finally outputting 12 processed gray maps, and performing binarization processing on the processed gray maps to obtain a label file of the water area distribution area of the remote sensing image;
s4: calling an opencv-python library function in python to read the 12 processed gray-scale images and corresponding label files, and cutting the gray-scale images and the corresponding label files one by one according to step length m and image size n, so that the cut image size is n x 1, and both m and n are positive integers larger than 0 and are respectively stored under a plurality of folders; calling an imgauge library function to perform transformation and augmentation processing on all cut images one by one according to a data augmentation method to obtain an expanded data set; finally, respectively counting the pixel mean value and the standard deviation of each image in the data set, and carrying out standardization processing on the data; then, all images in the data set are subjected to normalization processing, and all images after normalization are divided to obtain a training set, a verification set and a test set;
s5: calling a convolution layer, a pooling layer, an up-sampling layer, a loss function and an activation function in a deep learning framework TensorFlow and Keras, and building a remote sensing image water area segmentation model based on deep learning, wherein the remote sensing image water area segmentation model has 13 inputs in the process of training a sample, and the input inputs are 12 gray-scale images and a corresponding label file respectively;
s6: setting training batch size and learning rate parameter learning according to the calculation performance and model parameter number of the display card, calling a train function, performing iterative training on the remote sensing image water area segmentation model by using the training set, and verifying and testing the remote sensing image water area segmentation model after each round of training by using a verification set and a test set; when the remote sensing image water area segmentation model is converged, obtaining and storing the trained remote sensing image water area segmentation model;
s7: after the trained remote sensing image water area segmentation model outputs segmentation results, fine adjustment processing is carried out on the segmentation results through a guide filter GF and a conditional random field model CRF; the guide filter GF takes the label file as a guide graph, takes an original image as an input image, and optimizes the boundary of a water area extraction result to eliminate salt and pepper noise; the binary potential function in the conditional random field model CRF restrains the color and the position between any two pixel points, so that the pixel points with similar color and adjacent positions can have the same classification more easily, and the edges are smoothed according to the smoothness between the adjacent pixel points;
s8: the method comprises the steps that a local machine is used as a server to issue REST service through a flash framework; the client converts the remote sensing image into base64 format characters through a base64 encoding tool, and transmits the characters to the local server through a post request; the local server responds to the post request, decodes an actual remote sensing image, calls a trained remote sensing image water area segmentation model and a post-processing algorithm guide filter GF and a conditional random field model CRF to the actual remote sensing image, realizes automatic extraction of the water area, and returns an extraction result to the client through base64 coding.
2. The remote sensing image water area automatic extraction method based on deep learning of claim 1, characterized in that: in step S5, the remote sensing image water area segmentation model is implemented by an encoding stage, a decoding stage, a residual learning module stage, a global attention module stage, and a boundary learning unit stage as follows:
in the encoding stage, extracting characteristic information of a water area through convolution and pooling;
in the decoding stage, the size of the image is restored through convolution and up-sampling for 4 times to obtain the result of water area extraction;
in the stage of a residual error learning module, a quick connection is arranged in a convolutional neural network so as to improve the speed of information circulation and the efficiency of network training, and two convolutional layers are added to improve the capability of extracting characteristic information of a model; the residual error learning module takes a feature map with the coding stage size of 2w × 2h × c as input, the result of the feature map after passing through the convolution kernel of two 3 × c is directly added with the original feature map in a one-to-one correspondence mode according to pixels, then the feature map is transformed and activated through softmax, the channel is called as shortcut connection, and the finally obtained feature map is consistent with the original feature map in size and is 2w × 2h × c;
in the global attention module stage, semantic segmentation information of the decoding stage and position information of the encoding stage are fused, and feature extraction information is compressed and enhanced in a global average pooling weighting mode; the global attention module takes the feature graph with the coding stage scale of 2w x 2h x c and the decoding stage scale of w x h x 2c as input, firstly obtains feature information after global average pooling of the feature graph in the decoding stage, then weights the feature graph with the feature information as a weight value to obtain a new feature graph, then changes the scale after the up-sampling of the feature graph in the decoding stage into 2w x 2h x 2c, and finally splices the feature graph obtained after the up-sampling and the weighted feature graph according to channels to obtain a fused feature graph, wherein the scale of the feature graph is 2w x 2h x 3 c;
in the stage of a boundary learning unit, a residual error learning module is formed by convolutional layers with different scales and shortcut connection shortcuts, and a convolution kernel is added on a branch, so that the aliasing effect generated in the process of upsampling and fusing feature maps with different scales due to the feature maps is eliminated, and the remote sensing image feature information with different scales is learned; the boundary learning unit takes the feature graph output by the global attention module stage as input, the feature graph needs to be subjected to information circulation through three different branches, the first branch is in quick connection and does not perform data transformation, the second branch is subjected to c convolution with the size of 3 × 3c to change the scale size of the feature graph from 2w × 2h × 3c to 2w × 2h × c, the third branch is subjected to c convolution with the size of 3 × 3c to change the scale size of the feature graph from 2w × 2h × 3c to 2w × 2h × c, then the feature graphs of the three branches are subjected to aliasing addition through c convolution with the size of 3 × c, and finally the feature graph which is eliminated is obtained;
wherein, w, h and c represent the width, height and channel number of the characteristic diagram respectively.
3. The remote sensing image water area automatic extraction method based on deep learning of claim 1, characterized in that: in step S4, the 12 gray-scale maps and the label files corresponding to the 12 gray-scale maps are all n × 1 in size.
4. The remote sensing image water area automatic extraction method based on deep learning of claim 2, characterized in that: in the encoding stage, 12 gray maps are used as input data, feature maps are obtained by convolution and fusion of the input data, each feature map passes through one pooling layer to be one scale, and the feature maps comprise 5 scales of the original map, namely 256 × 32, 128 × 64, 64 × 128, 32 × 256 and 16 × 16 512; after the image passes through the pooling layer, the size of the characteristic graph is halved, the number of channels is doubled, and the water area characteristic information of the image is extracted through two convolutional neural networks.
5. The remote sensing image water area automatic extraction method based on deep learning of claim 3, characterized in that: when n is 256, the feature map in the decoding stage is fused with the feature map with the same size corresponding to the encoding stage through the global attention module every time the feature map is sampled, and then boundary texture information of a water area part in the feature map is integrated and extracted through the boundary learning unit; and finally outputting a binary grayscale map with the size of 256 × 1, wherein if the value of the binary grayscale map is 1, the water area part is represented, and if the value of the binary grayscale map is 0, the non-water area part is represented.
6. The utility model provides a remote sensing image waters automatic extraction system based on degree of deep learning which characterized in that: the system comprises the following modules:
the data preprocessing module is used for performing atmospheric correction on spectral data of a certain remote sensing image and resampling the corrected data to obtain data of each wave band of the remote sensing image;
the information extraction module is used for carrying out normalization processing on data of each wave band of the remote sensing image through a wave band operation tool of remote sensing image processing software, respectively calculating a normalized difference water body index NDWI, an improved water body index model NDWI3, an improved normalized difference water body index MNDWI, an enhanced water body index EWI, a novel water body index NWI, a normalized vegetation coverage index NDVI and a wetland forest index, identifying three visible light wave bands of red, green and blue, near infrared NIR, intermediate infrared SWIR-1 and SWIR-2 wave bands, and outputting data of each wave band or the data of the indexes into a gray scale map so as to obtain 12 gray scale maps;
the system comprises a tag file making module, a remote sensing image water area distribution area generating module, a remote sensing image water area vector generating module, a tag file generating module and a remote sensing image water area vector generating module, wherein the tag file making module is used for newly creating a water area vector file water.shp through ArcGIS software, loading 12 gray maps, vectorizing a water area distribution area according to a remote sensing visual interpretation method, finally outputting a gray map water.png, binarizing the gray map water.png, and obtaining a binarized water area vector file which is a tag file of the water area distribution area of the remote;
the data set generation module is used for calling a library function of opencv-python in python to read the 12 gray-scale images and the label files, and cutting the 12 gray-scale images and the label files in a one-to-one correspondence mode according to the step length of 128 and the image size of 256, so that the cut images with the size of 256 x 1 are respectively stored in 13 folders; calling an imgauge library function to correspondingly transform and amplify the cut images one by one according to a data enhancement method so as to expand a data set; finally, the pixel mean value and the standard deviation of all the data are counted, and the data are standardized; then dividing the plurality of pictures after normalization processing to obtain a training set, a verification set and a test set;
the classification model establishing module is used for calling a convolution layer, a pooling layer, an up-sampling layer, a loss function and an activation function in a deep learning framework TensorFlow and Keras so as to establish a remote sensing image water area segmentation model WE-Net based on deep learning, wherein the segmentation model has 13 inputs in the process of training a sample, and the inputs are 12 gray level images and a corresponding label file respectively; the segmentation model WE-Net is realized by the following steps of encoding, decoding, residual learning, global attention module and boundary learning unit;
the coding module: the device is used for extracting characteristic information of a water area through convolution and pooling in an encoding stage; in the encoding stage, 12 gray maps are used as input data, feature maps are obtained by convolution and fusion of the input data, each feature map passes through one pooling layer to be one scale, and the feature maps comprise 5 scales of the original map, namely 256 × 32, 128 × 64, 64 × 128, 32 × 256 and 16 × 16 512; after the image passes through the pooling layer, the size of the feature map is reduced by half, the number of channels is doubled, and the water area feature information of the image is extracted through two convolutional neural networks;
a decoding module: the method is used for recovering the image size through convolution and 4 times of up-sampling in a decoding stage to obtain a water area extraction result; every time the feature map in the decoding stage is sampled once, the feature maps with the same size and size corresponding to the encoding stage are fused through the global attention module, and then boundary texture information of a water area part in the feature maps is integrated and extracted through the boundary learning unit; finally outputting a binary gray-scale map with the scale size of 256 × 1, wherein if the value of the gray-scale map is 1, the gray-scale map represents a water area part, and if the value of the gray-scale map is 0, the gray-scale map represents a non-water area part;
a residual error learning module: the method is used for setting a quick connection for the convolutional neural network in the residual error learning module stage, improving the speed of information circulation and the efficiency of network training, and simultaneously adding two convolutional layers to improve the capability of extracting characteristic information of a model; the residual error learning module takes a feature map with the coding stage scale of 2w × 2h × c as input, the feature map is directly added with the original feature map in a one-to-one correspondence manner according to pixels after the result of two times of convolution is carried out by c convolution kernels with 3 × c, then transformation and activation are carried out through softmax, the path is called as shortcut connection, and the finally obtained feature map is consistent with the original feature map in scale and is 2w × 2h × c;
global attention module: the system is used for fusing semantic segmentation information of a decoding stage and position information of an encoding stage at a global attention module stage, and compressing and enhancing feature extraction information in a global average pooling weighting mode; the global attention module takes the feature graph with the coding stage scale of 2w x 2h x c and the decoding stage scale of w x h x 2c as input, firstly obtains feature information after global average pooling of the feature graph in the decoding stage, then weights the feature graph with the feature information as a weight value to obtain a new feature graph, then changes the scale after the up-sampling of the feature graph in the decoding stage into 2w x 2h x 2c, and finally splices the feature graph obtained after the up-sampling and the weighted feature graph according to channels to obtain a fused feature graph, wherein the scale of the feature graph is 2w x 2h x 3 c;
a boundary learning module: the method is used for forming a residual error learning module through convolution layers with different scales and shortcut connection short in a boundary learning unit stage, and adding a convolution kernel on a branch path, so that aliasing effect generated in the process of up-sampling and fusing feature maps with different scales due to the feature maps is eliminated, and remote sensing image feature information with different scales is learned; the boundary learning unit takes the feature graph output by the global attention module stage as input, the feature graph needs to be subjected to information circulation through three different branches, the first branch is in quick connection and does not perform data transformation, the second branch is subjected to c convolution with the size of 3 × 3c to change the scale size of the feature graph from 2w × 2h × 3c to 2w × 2h × c, the third branch is subjected to c convolution with the size of 3 × 3c to change the scale size of the feature graph from 2w × 2h × 3c to 2w × 2h × c, then the feature graphs of the three branches are subjected to aliasing addition through c convolution with the size of 3 × c, and finally the feature graph which is eliminated is obtained;
the model training module is used for setting training batch size and learning rate parameter learning according to the calculation performance and model parameter number of the display card, calling a train function, performing iterative training on the remote sensing image water area segmentation model WE-Net by using the training set, and verifying the remote sensing image water area segmentation model WE-Net after each round of training by using the verification set; the training process is visualized by taking the number of training rounds as a horizontal axis and the IOU value as a vertical axis, after dozens of rounds of training, the IOU rises firstly and then approaches a certain IOU value in a wireless mode, then the remote sensing image water area segmentation model WE-Net is converged, the remote sensing image water area segmentation model WE-Net parameters are stored, and the training is stopped; finally, calling the stored remote sensing image water area segmentation model WE-Net through a test function, and evaluating the precision of the remote sensing image water area segmentation model WE-Net according to the IOU value calculated on the test set;
the model fine-tuning module is used for outputting a result of automatic water area extraction by the trained remote sensing image water area segmentation model and then post-processing the result by the guide filter GF and the conditional random field model CRF; the label file is regarded as a guide graph by guide filtering, an original image is taken as an input image, and the boundary of a water area extraction result is optimized so as to eliminate salt and pepper noise; the binary potential function in the conditional random field restrains the color and the position between any two pixel points, so that the pixel points with similar color and adjacent positions can easily have the same classification, and meanwhile, the smoothness between the adjacent pixel points is considered, the edges are smoothed, and the semantic segmentation result is finely adjusted;
the model application module is used for training and testing a remote sensing image water area segmentation model WE-Net, storing weight parameters after preset precision is met, obtaining and storing a network model WE-Net.h5, and issuing REST service through a flash framework by taking the local machine as a server; the client converts the remote sensing image into base64 format characters through a base64 encoding tool, and transmits the characters to the local server through a post request; and the server responds to the post request, decodes the remote sensing image, calls a remote sensing image water area segmentation model WE-Net and a post-processing algorithm guide filter GF and a conditional random field CRF to realize automatic water area extraction, and then returns an extraction result to the client through base64 coding.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010493489.6A CN111767801B (en) | 2020-06-03 | 2020-06-03 | Remote sensing image water area automatic extraction method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010493489.6A CN111767801B (en) | 2020-06-03 | 2020-06-03 | Remote sensing image water area automatic extraction method and system based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111767801A true CN111767801A (en) | 2020-10-13 |
| CN111767801B CN111767801B (en) | 2023-06-16 |
Family
ID=72719334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010493489.6A Active CN111767801B (en) | 2020-06-03 | 2020-06-03 | Remote sensing image water area automatic extraction method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111767801B (en) |
Cited By (45)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112258539A (en) * | 2020-11-13 | 2021-01-22 | 腾讯科技（深圳）有限公司 | Water system data processing method, device, electronic equipment and readable storage medium |
| CN112365508A (en) * | 2020-11-03 | 2021-02-12 | 云南电网有限责任公司昆明供电局 | SAR remote sensing image water area segmentation method based on visual attention and residual error network |
| CN112508986A (en) * | 2020-12-04 | 2021-03-16 | 武汉大学 | Water level measurement method based on deep convolutional network and random field |
| CN112508106A (en) * | 2020-12-08 | 2021-03-16 | 大连海事大学 | Underwater image classification method based on convolutional neural network |
| CN112561876A (en) * | 2020-12-14 | 2021-03-26 | 中南大学 | Image-based pond and reservoir water quality detection method and system |
| CN112883900A (en) * | 2021-03-12 | 2021-06-01 | 中科三清科技有限公司 | Method and device for bare-ground inversion of visible images of remote sensing images |
| CN112989919A (en) * | 2020-12-25 | 2021-06-18 | 首都师范大学 | Method and system for extracting target object from image |
| CN113269028A (en) * | 2021-04-07 | 2021-08-17 | 南方科技大学 | Water body change detection method and system based on deep convolutional neural network |
| CN113538425A (en) * | 2021-09-16 | 2021-10-22 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Passable water area segmentation equipment, image segmentation model training and image segmentation method |
| CN113591633A (en) * | 2021-07-18 | 2021-11-02 | 武汉理工大学 | Object-oriented land utilization information interpretation method based on dynamic self-attention Transformer |
| CN113642663A (en) * | 2021-08-24 | 2021-11-12 | 中国水利水电科学研究院 | Satellite remote sensing image water body extraction method |
| CN113643235A (en) * | 2021-07-07 | 2021-11-12 | 青岛高重信息科技有限公司 | Chip counting method based on deep learning |
| CN113840127A (en) * | 2021-08-12 | 2021-12-24 | 长光卫星技术有限公司 | Method for acquiring water area automatic mask processing DSM by satellite video image |
| CN114332637A (en) * | 2022-03-17 | 2022-04-12 | 北京航空航天大学杭州创新研究院 | Remote sensing image water body extraction method and interaction method for remote sensing image water body extraction |
| CN114565657A (en) * | 2021-12-24 | 2022-05-31 | 西安电子科技大学 | Method for extracting river width in remote sensing image based on edge gradient and direction texture |
| US11403069B2 (en) | 2017-07-24 | 2022-08-02 | Tesla, Inc. | Accelerated mathematical engine |
| US11409692B2 (en) | 2017-07-24 | 2022-08-09 | Tesla, Inc. | Vector computational unit |
| CN114998738A (en) * | 2022-06-10 | 2022-09-02 | 湖南大学 | Water body image extraction method and system based on confidence coefficient filtering |
| CN115170979A (en) * | 2022-06-30 | 2022-10-11 | 国家能源投资集团有限责任公司 | Mining area fine land classification method based on multi-source data fusion |
| WO2022222458A1 (en) * | 2021-04-19 | 2022-10-27 | 温州医科大学 | Artificial intelligence-assisted diagnosis model construction system for medical images |
| US11487288B2 (en) | 2017-03-23 | 2022-11-01 | Tesla, Inc. | Data synthesis for autonomous control systems |
| CN115423829A (en) * | 2022-07-29 | 2022-12-02 | 江苏省水利科学研究院 | Method and system for rapidly extracting water body from single-band remote sensing image |
| US11537811B2 (en) | 2018-12-04 | 2022-12-27 | Tesla, Inc. | Enhanced object detection for autonomous vehicles based on field view |
| US11562231B2 (en) | 2018-09-03 | 2023-01-24 | Tesla, Inc. | Neural networks for embedded devices |
| US11561791B2 (en) | 2018-02-01 | 2023-01-24 | Tesla, Inc. | Vector computational unit receiving data elements in parallel from a last row of a computational array |
| US11567514B2 (en) | 2019-02-11 | 2023-01-31 | Tesla, Inc. | Autonomous and user controlled vehicle summon to a target |
| US11610117B2 (en) | 2018-12-27 | 2023-03-21 | Tesla, Inc. | System and method for adapting a neural network model on a hardware platform |
| US11636333B2 (en) | 2018-07-26 | 2023-04-25 | Tesla, Inc. | Optimizing neural network structures for embedded systems |
| CN116030352A (en) * | 2023-03-29 | 2023-04-28 | 山东锋士信息技术有限公司 | Long-time-sequence land utilization classification method integrating multi-scale segmentation and super-pixel segmentation |
| US11665108B2 (en) | 2018-10-25 | 2023-05-30 | Tesla, Inc. | QoS manager for system on a chip communications |
| US11681649B2 (en) | 2017-07-24 | 2023-06-20 | Tesla, Inc. | Computational array microprocessor system using non-consecutive data formatting |
| CN116452901A (en) * | 2023-06-19 | 2023-07-18 | 中国科学院海洋研究所 | Automatic extraction method for ocean culture area of remote sensing image based on deep learning |
| CN116452985A (en) * | 2023-02-21 | 2023-07-18 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
| CN116481600A (en) * | 2023-06-26 | 2023-07-25 | 四川省林业勘察设计研究院有限公司 | Plateau forestry ecological monitoring and early warning system and method |
| US11734562B2 (en) | 2018-06-20 | 2023-08-22 | Tesla, Inc. | Data pipeline and deep learning system for autonomous driving |
| US11748620B2 (en) | 2019-02-01 | 2023-09-05 | Tesla, Inc. | Generating ground truth for machine learning from time series elements |
| CN116699096A (en) * | 2023-08-08 | 2023-09-05 | 凯德技术长沙股份有限公司 | Water quality detection method and system based on deep learning |
| US11790664B2 (en) | 2019-02-19 | 2023-10-17 | Tesla, Inc. | Estimating object properties using visual image data |
| US11816585B2 (en) | 2018-12-03 | 2023-11-14 | Tesla, Inc. | Machine learning models operating at different frequencies for autonomous vehicles |
| US11841434B2 (en) | 2018-07-20 | 2023-12-12 | Tesla, Inc. | Annotation cross-labeling for autonomous control systems |
| CN117392539A (en) * | 2023-10-13 | 2024-01-12 | 哈尔滨师范大学 | River water body identification method based on deep learning, electronic equipment and storage medium |
| CN117409203A (en) * | 2023-11-14 | 2024-01-16 | 自然资源部国土卫星遥感应用中心 | Shallow lake area extraction method |
| US11893774B2 (en) | 2018-10-11 | 2024-02-06 | Tesla, Inc. | Systems and methods for training machine models with augmented data |
| US11893393B2 (en) | 2017-07-24 | 2024-02-06 | Tesla, Inc. | Computational array microprocessor system with hardware arbiter managing memory requests |
| US12014553B2 (en) | 2019-02-01 | 2024-06-18 | Tesla, Inc. | Predicting three-dimensional features for autonomous driving |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050100220A1 (en) * | 2002-11-06 | 2005-05-12 | Keaton Patricia A. | Method and apparatus for automatically extracting geospatial features from multispectral imagery suitable for fast and robust extraction of landmarks |
| US10185891B1 (en) * | 2016-07-08 | 2019-01-22 | Gopro, Inc. | Systems and methods for compact convolutional neural networks |
| CN109325395A (en) * | 2018-04-28 | 2019-02-12 | 二十世纪空间技术应用股份有限公司 | The recognition methods of image, convolutional neural networks model training method and device |
| CN110781775A (en) * | 2019-10-10 | 2020-02-11 | 武汉大学 | Remote sensing image water body information accurate segmentation method supported by multi-scale features |
| CN110852225A (en) * | 2019-10-31 | 2020-02-28 | 中国地质大学（武汉） | Remote sensing image mangrove forest extraction method and system based on deep convolutional neural network |
-
2020
- 2020-06-03 CN CN202010493489.6A patent/CN111767801B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050100220A1 (en) * | 2002-11-06 | 2005-05-12 | Keaton Patricia A. | Method and apparatus for automatically extracting geospatial features from multispectral imagery suitable for fast and robust extraction of landmarks |
| US10185891B1 (en) * | 2016-07-08 | 2019-01-22 | Gopro, Inc. | Systems and methods for compact convolutional neural networks |
| CN109325395A (en) * | 2018-04-28 | 2019-02-12 | 二十世纪空间技术应用股份有限公司 | The recognition methods of image, convolutional neural networks model training method and device |
| CN110781775A (en) * | 2019-10-10 | 2020-02-11 | 武汉大学 | Remote sensing image water body information accurate segmentation method supported by multi-scale features |
| CN110852225A (en) * | 2019-10-31 | 2020-02-28 | 中国地质大学（武汉） | Remote sensing image mangrove forest extraction method and system based on deep convolutional neural network |
Non-Patent Citations (3)
| Title |
|---|
| WENQING FENG,ET.AL: "Water Body Extraction From Very High-Resolution Remote Sensing Imagery Using Deep U-Net and a Superpixel-Based Conditional Random Field Model", IEEE GEOSCIENCE AND REMOTE SENSING LETTERS * |
| 么鸿原等: "基于优化导向滤波算法的遥感图像预处理仿真", 计算机仿真 * |
| 陈前，等: "基于深度学习的高分遥感影像水体提取模型研究", 地理与地理信息科学 * |
Cited By (68)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US12020476B2 (en) | 2017-03-23 | 2024-06-25 | Tesla, Inc. | Data synthesis for autonomous control systems |
| US11487288B2 (en) | 2017-03-23 | 2022-11-01 | Tesla, Inc. | Data synthesis for autonomous control systems |
| US11403069B2 (en) | 2017-07-24 | 2022-08-02 | Tesla, Inc. | Accelerated mathematical engine |
| US11409692B2 (en) | 2017-07-24 | 2022-08-09 | Tesla, Inc. | Vector computational unit |
| US11893393B2 (en) | 2017-07-24 | 2024-02-06 | Tesla, Inc. | Computational array microprocessor system with hardware arbiter managing memory requests |
| US11681649B2 (en) | 2017-07-24 | 2023-06-20 | Tesla, Inc. | Computational array microprocessor system using non-consecutive data formatting |
| US11797304B2 (en) | 2018-02-01 | 2023-10-24 | Tesla, Inc. | Instruction set architecture for a vector computational unit |
| US11561791B2 (en) | 2018-02-01 | 2023-01-24 | Tesla, Inc. | Vector computational unit receiving data elements in parallel from a last row of a computational array |
| US11734562B2 (en) | 2018-06-20 | 2023-08-22 | Tesla, Inc. | Data pipeline and deep learning system for autonomous driving |
| US11841434B2 (en) | 2018-07-20 | 2023-12-12 | Tesla, Inc. | Annotation cross-labeling for autonomous control systems |
| US11636333B2 (en) | 2018-07-26 | 2023-04-25 | Tesla, Inc. | Optimizing neural network structures for embedded systems |
| US11562231B2 (en) | 2018-09-03 | 2023-01-24 | Tesla, Inc. | Neural networks for embedded devices |
| US11983630B2 (en) | 2018-09-03 | 2024-05-14 | Tesla, Inc. | Neural networks for embedded devices |
| US11893774B2 (en) | 2018-10-11 | 2024-02-06 | Tesla, Inc. | Systems and methods for training machine models with augmented data |
| US11665108B2 (en) | 2018-10-25 | 2023-05-30 | Tesla, Inc. | QoS manager for system on a chip communications |
| US11816585B2 (en) | 2018-12-03 | 2023-11-14 | Tesla, Inc. | Machine learning models operating at different frequencies for autonomous vehicles |
| US11908171B2 (en) | 2018-12-04 | 2024-02-20 | Tesla, Inc. | Enhanced object detection for autonomous vehicles based on field view |
| US11537811B2 (en) | 2018-12-04 | 2022-12-27 | Tesla, Inc. | Enhanced object detection for autonomous vehicles based on field view |
| US11610117B2 (en) | 2018-12-27 | 2023-03-21 | Tesla, Inc. | System and method for adapting a neural network model on a hardware platform |
| US12014553B2 (en) | 2019-02-01 | 2024-06-18 | Tesla, Inc. | Predicting three-dimensional features for autonomous driving |
| US11748620B2 (en) | 2019-02-01 | 2023-09-05 | Tesla, Inc. | Generating ground truth for machine learning from time series elements |
| US11567514B2 (en) | 2019-02-11 | 2023-01-31 | Tesla, Inc. | Autonomous and user controlled vehicle summon to a target |
| US11790664B2 (en) | 2019-02-19 | 2023-10-17 | Tesla, Inc. | Estimating object properties using visual image data |
| CN112365508A (en) * | 2020-11-03 | 2021-02-12 | 云南电网有限责任公司昆明供电局 | SAR remote sensing image water area segmentation method based on visual attention and residual error network |
| CN112258539A (en) * | 2020-11-13 | 2021-01-22 | 腾讯科技（深圳）有限公司 | Water system data processing method, device, electronic equipment and readable storage medium |
| CN112258539B (en) * | 2020-11-13 | 2023-08-01 | 腾讯科技（深圳）有限公司 | Water system data processing method, device, electronic equipment and readable storage medium |
| CN112508986A (en) * | 2020-12-04 | 2021-03-16 | 武汉大学 | Water level measurement method based on deep convolutional network and random field |
| CN112508106A (en) * | 2020-12-08 | 2021-03-16 | 大连海事大学 | Underwater image classification method based on convolutional neural network |
| CN112508106B (en) * | 2020-12-08 | 2024-05-24 | 大连海事大学 | Underwater image classification method based on convolutional neural network |
| CN112561876A (en) * | 2020-12-14 | 2021-03-26 | 中南大学 | Image-based pond and reservoir water quality detection method and system |
| CN112561876B (en) * | 2020-12-14 | 2024-02-23 | 中南大学 | Image-based water quality detection method and system for ponds and reservoirs |
| CN112989919A (en) * | 2020-12-25 | 2021-06-18 | 首都师范大学 | Method and system for extracting target object from image |
| CN112989919B (en) * | 2020-12-25 | 2024-04-19 | 首都师范大学 | Method and system for extracting target object from image |
| CN112883900A (en) * | 2021-03-12 | 2021-06-01 | 中科三清科技有限公司 | Method and device for bare-ground inversion of visible images of remote sensing images |
| CN113269028A (en) * | 2021-04-07 | 2021-08-17 | 南方科技大学 | Water body change detection method and system based on deep convolutional neural network |
| CN113269028B (en) * | 2021-04-07 | 2022-02-11 | 南方科技大学 | Water body change detection method and system based on deep convolutional neural network |
| WO2022222458A1 (en) * | 2021-04-19 | 2022-10-27 | 温州医科大学 | Artificial intelligence-assisted diagnosis model construction system for medical images |
| CN113643235B (en) * | 2021-07-07 | 2023-12-29 | 青岛高重信息科技有限公司 | Chip counting method based on deep learning |
| CN113643235A (en) * | 2021-07-07 | 2021-11-12 | 青岛高重信息科技有限公司 | Chip counting method based on deep learning |
| CN113591633A (en) * | 2021-07-18 | 2021-11-02 | 武汉理工大学 | Object-oriented land utilization information interpretation method based on dynamic self-attention Transformer |
| CN113591633B (en) * | 2021-07-18 | 2024-04-30 | 武汉理工大学 | Object-oriented land utilization information interpretation method based on dynamic self-attention transducer |
| CN113840127B (en) * | 2021-08-12 | 2024-02-27 | 长光卫星技术股份有限公司 | Method for automatically masking DSM (digital multimedia subsystem) in satellite video image acquisition water area |
| CN113840127A (en) * | 2021-08-12 | 2021-12-24 | 长光卫星技术有限公司 | Method for acquiring water area automatic mask processing DSM by satellite video image |
| CN113642663A (en) * | 2021-08-24 | 2021-11-12 | 中国水利水电科学研究院 | Satellite remote sensing image water body extraction method |
| CN113642663B (en) * | 2021-08-24 | 2022-03-22 | 中国水利水电科学研究院 | Satellite remote sensing image water body extraction method |
| CN113538425A (en) * | 2021-09-16 | 2021-10-22 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Passable water area segmentation equipment, image segmentation model training and image segmentation method |
| CN114565657B (en) * | 2021-12-24 | 2023-08-15 | 西安电子科技大学 | Method for extracting river width in remote sensing image based on edge gradient and directional texture |
| CN114565657A (en) * | 2021-12-24 | 2022-05-31 | 西安电子科技大学 | Method for extracting river width in remote sensing image based on edge gradient and direction texture |
| CN114332637B (en) * | 2022-03-17 | 2022-08-30 | 北京航空航天大学杭州创新研究院 | Remote sensing image water body extraction method and interaction method for remote sensing image water body extraction |
| CN114332637A (en) * | 2022-03-17 | 2022-04-12 | 北京航空航天大学杭州创新研究院 | Remote sensing image water body extraction method and interaction method for remote sensing image water body extraction |
| CN114998738A (en) * | 2022-06-10 | 2022-09-02 | 湖南大学 | Water body image extraction method and system based on confidence coefficient filtering |
| CN115170979B (en) * | 2022-06-30 | 2023-02-24 | 国家能源投资集团有限责任公司 | Mining area fine land classification method based on multi-source data fusion |
| CN115170979A (en) * | 2022-06-30 | 2022-10-11 | 国家能源投资集团有限责任公司 | Mining area fine land classification method based on multi-source data fusion |
| CN115423829B (en) * | 2022-07-29 | 2024-03-01 | 江苏省水利科学研究院 | Method and system for rapidly extracting water body of single-band remote sensing image |
| CN115423829A (en) * | 2022-07-29 | 2022-12-02 | 江苏省水利科学研究院 | Method and system for rapidly extracting water body from single-band remote sensing image |
| CN116452985A (en) * | 2023-02-21 | 2023-07-18 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
| CN116452985B (en) * | 2023-02-21 | 2023-10-31 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
| CN116030352A (en) * | 2023-03-29 | 2023-04-28 | 山东锋士信息技术有限公司 | Long-time-sequence land utilization classification method integrating multi-scale segmentation and super-pixel segmentation |
| CN116452901A (en) * | 2023-06-19 | 2023-07-18 | 中国科学院海洋研究所 | Automatic extraction method for ocean culture area of remote sensing image based on deep learning |
| CN116452901B (en) * | 2023-06-19 | 2023-09-15 | 中国科学院海洋研究所 | Automatic extraction method for ocean culture area of remote sensing image based on deep learning |
| CN116481600B (en) * | 2023-06-26 | 2023-10-20 | 四川省林业勘察设计研究院有限公司 | Plateau forestry ecological monitoring and early warning system and method |
| CN116481600A (en) * | 2023-06-26 | 2023-07-25 | 四川省林业勘察设计研究院有限公司 | Plateau forestry ecological monitoring and early warning system and method |
| CN116699096B (en) * | 2023-08-08 | 2023-11-03 | 凯德技术长沙股份有限公司 | Water quality detection method and system based on deep learning |
| CN116699096A (en) * | 2023-08-08 | 2023-09-05 | 凯德技术长沙股份有限公司 | Water quality detection method and system based on deep learning |
| CN117392539B (en) * | 2023-10-13 | 2024-04-09 | 哈尔滨师范大学 | River water body identification method based on deep learning, electronic equipment and storage medium |
| CN117392539A (en) * | 2023-10-13 | 2024-01-12 | 哈尔滨师范大学 | River water body identification method based on deep learning, electronic equipment and storage medium |
| CN117409203B (en) * | 2023-11-14 | 2024-04-02 | 自然资源部国土卫星遥感应用中心 | Shallow lake area extraction method |
| CN117409203A (en) * | 2023-11-14 | 2024-01-16 | 自然资源部国土卫星遥感应用中心 | Shallow lake area extraction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111767801B (en) | 2023-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111767801B (en) | Remote sensing image water area automatic extraction method and system based on deep learning | |
| CN110852225B (en) | Remote sensing image mangrove forest extraction method and system based on deep convolutional neural network | |
| CN111242037A (en) | Lane line detection method based on structural information | |
| CN113780296A (en) | Remote sensing image semantic segmentation method and system based on multi-scale information fusion | |
| CN110728197B (en) | Single-tree-level tree species identification method based on deep learning | |
| CN113469052B (en) | Super-resolution building fine identification method based on multi-scale feature deconvolution | |
| CN111008664B (en) | Hyperspectral sea ice detection method based on space-spectrum combined characteristics | |
| CN110570440A (en) | Image automatic segmentation method and device based on deep learning edge detection | |
| Du et al. | Segmentation and sampling method for complex polyline generalization based on a generative adversarial network | |
| CN110991430B (en) | Ground feature identification and coverage rate calculation method and system based on remote sensing image | |
| CN112950780A (en) | Intelligent network map generation method and system based on remote sensing image | |
| CN113887472A (en) | Remote sensing image cloud detection method based on cascade color and texture feature attention | |
| CN113239736A (en) | Land cover classification annotation graph obtaining method, storage medium and system based on multi-source remote sensing data | |
| CN116958827A (en) | Deep learning-based abandoned land area extraction method | |
| CN116645592A (en) | Crack detection method based on image processing and storage medium | |
| CN113205016B (en) | River and lake shoreline change detection method of constant residual error type Unet and remote sensing water body index | |
| Wu et al. | Improved mask R-CNN-based cloud masking method for remote sensing images | |
| CN113269028B (en) | Water body change detection method and system based on deep convolutional neural network | |
| CN115527027A (en) | Remote sensing image ground object segmentation method based on multi-feature fusion mechanism | |
| CN112800851B (en) | Water body contour automatic extraction method and system based on full convolution neuron network | |
| CN117115666A (en) | Plateau lake extraction method, device, equipment and medium based on multi-source data | |
| CN115019044A (en) | Individual plant segmentation method and device, terminal device and readable storage medium | |
| CN112036246B (en) | Construction method of remote sensing image classification model, remote sensing image classification method and system | |
| Shun et al. | Research on remote sensing image extraction based on deep learning | |
| CN114241194A (en) | Instrument identification and reading method based on lightweight network |
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 |