CN111985368B - Convolutional neural network water body extraction method for container cloud - Google Patents

Convolutional neural network water body extraction method for container cloud Download PDF

Info

Publication number
CN111985368B
CN111985368B CN202010789924.XA CN202010789924A CN111985368B CN 111985368 B CN111985368 B CN 111985368B CN 202010789924 A CN202010789924 A CN 202010789924A CN 111985368 B CN111985368 B CN 111985368B
Authority
CN
China
Prior art keywords
water body
neural network
spectral
convolutional neural
body extraction
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
Application number
CN202010789924.XA
Other languages
Chinese (zh)
Other versions
CN111985368A (en
Inventor
张东映
梁忠壮
黄伟
洪志明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan Shanlai Technology Co ltd
Original Assignee
Wuhan Shanlai Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuhan Shanlai Technology Co ltd filed Critical Wuhan Shanlai Technology Co ltd
Priority to CN202010789924.XA priority Critical patent/CN111985368B/en
Publication of CN111985368A publication Critical patent/CN111985368A/en
Application granted granted Critical
Publication of CN111985368B publication Critical patent/CN111985368B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—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
    • 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
    • Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
    • Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (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)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Astronomy & Astrophysics (AREA)
  • Remote Sensing (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a container cloud-oriented convolutional neural network water body extraction method, which relates to the technical field of interpretation and classification of remote sensing images and comprises the following steps: acquiring spectral characteristics of a water body in advance, and selecting wave bands to form spectral vectors; converting the spectrum vector to obtain a spectrum characteristic matrix, and using the spectrum characteristic matrix as an input characteristic of a convolutional neural network model; taking the spectral feature matrix as a sample, and obtaining a water body extraction model; and carrying out object segmentation on the remote sensing image needing water body extraction by using a multi-resolution segmentation algorithm, and identifying each object by using a unique ID. The method extracts the water body by comprehensively utilizing the spectral characteristics and the spatial characteristics, and can effectively inhibit the influence of the shadow on the water body extraction; meanwhile, the container cloud and Spark are utilized for parallel optimization, the efficiency is obviously superior to that in a single-machine mode, and the efficiency is obviously improved along with the increase of the data volume.

Description

Convolutional neural network water body extraction method for container cloud
Technical Field
The invention relates to the technical field of interpretation and classification of remote sensing images, in particular to a container cloud-oriented convolutional neural network water body extraction method.
Background
With the rapid development of remote sensing technology, the time resolution and the spatial resolution of remote sensing data are higher and higher, and abundant data source support is provided for business monitoring application in the fields of water conservancy, agriculture, environment and the like. Rivers and lakes in China are numerous, and how to fully utilize remote sensing images to quickly acquire water body information is one of important tasks of water conservancy monitoring. Especially for the region with water accumulation flood disasters all the year round, the method has important significance for preventing and reducing flood by accurately and quickly acquiring the flood range.
The traditional remote sensing water body extraction method has low parallelism or a plurality of algorithms are all serial, so that the water body extraction efficiency is low. At present, cloud computing and big data processing technologies are widely applied to remote sensing processing, but research on remote sensing water bodies by combining cloud computing, a container virtualization technology, Spark and a deep learning model is less, so that a water body extraction algorithm cannot be efficiently, accurately and flexibly applied.
Therefore, a convolution neural network water body extraction method facing to container cloud is needed.
An effective solution to the problems in the related art has not been proposed yet.
Disclosure of Invention
Aiming at the problems in the related art, the invention provides a convolutional neural network water body extraction method facing to a container cloud, so as to overcome the technical problems in the prior art.
The technical scheme of the invention is realized as follows:
a convolutional neural network water body extraction method facing to container cloud comprises the following steps:
step S1, acquiring the spectral characteristics of the water body in advance, and selecting wave bands to form spectral vectors;
step S2, converting the spectral vector to obtain a spectral feature matrix, and using the spectral feature matrix as an input feature of the convolutional neural network model;
step S3, taking the spectral feature matrix as a sample to obtain a water body extraction model;
step S4, carrying out object segmentation on the remote sensing image needing water body extraction by using a multi-resolution segmentation algorithm, and identifying each object by using a unique ID;
step S5, converting the data format of the remote sensing image needing water body extraction into the data format of parquet, reading the converted data in a Spark DataSet type by using a Spark Read method, and re-partitioning the Read DataSet in a line-based manner;
in step S6, each object is traversed in the DataSet of the classification result, information on the classification of pixels in the object range is extracted, the category with the largest number of votes is found by the voting method, and the category is used as the classification result of the object.
Further, the composing the spectral vector comprises the following steps:
acquiring a spectral reflectivity curve of a ground object in advance;
screening wave bands used for water body extraction in the images, and synthesizing the wave bands into a multi-band image;
collecting pixels of a water body and other ground objects for the synthesized multiband images, and carrying out category marking on the selected pixels;
a pixel is determined to contain a plurality of image band values that together form a spectral vector for the pixel.
Further, the spectral feature matrix further includes the following steps:
multiplying the spectrum vector by the transpose of the spectrum vector to obtain a spectrum characteristic matrix;
a spectral feature matrix, represented as:
S＝s T ·s
where s is the spectral vector in the ground feature pixel, s T S is the spectral feature matrix for each pixel, which is the transposed vector of the spectral vector.
Further, the water body extraction model comprises the following steps:
calibrating the spectral feature matrix as a sample;
and (3) adding 7:3, randomly dividing the sample into training sample data and verification sample data;
training a convolutional neural network water body model by using training sample data;
and verifying the water body model of the convolutional neural network by using the verification sample data.
Further, traversing each object in the DataSet of the classification result, further comprising the following steps:
packing the Spark application program into an executable program and a trained convolutional neural network water body model;
and calling an interface of the convolutional neural network model to classify images of each pixel in the DataSet of each partition, and outputting the DataSet of a classification result.
The invention has the beneficial effects that:
the invention relates to a convolutional neural network water body extraction method oriented to container cloud, which comprises the steps of selecting proper wave bands to form spectral vectors by analyzing the spectral characteristics of a water body, and converting the spectrum characteristic matrix into a spectrum characteristic matrix, using the spectrum characteristic matrix as the input characteristic of the convolutional neural network to improve the accuracy of water body extraction, realizing that cloud computing is used as a basic resource pool, using a container virtualization technology to provide elastic resources for providing computing resources for a convolutional neural network water body extraction method based on Spark memory parallel, meanwhile, the number of the spare data partitions and the number of the actuators are automatically planned for the submitted task according to the current resource condition of the system, the method is beneficial to cloud service management of remote sensing image water body extraction algorithms based on deep learning or traditional index method and the like, the water body is extracted by comprehensively utilizing the spectral characteristics and the spatial characteristics, so that the influence of shadow on the water body extraction can be effectively inhibited; meanwhile, the container cloud and Spark are utilized for parallel optimization, the efficiency is obviously superior to that in a single-machine mode, and the efficiency is obviously improved along with the increase of the data volume.
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 schematic flow chart of a convolutional neural network water body extraction method facing a container cloud according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a scene application of a convolutional neural network water body extraction method facing a container cloud according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a multi-resolution segmented scale variation curve of a convolutional neural network water body extraction method facing a container cloud according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a water body extraction result of the convolutional neural network water body extraction method facing the container cloud according to the embodiment of the 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 that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.
According to the embodiment of the invention, a convolutional neural network water body extraction method facing to container cloud is provided.
As shown in fig. 1, the convolutional neural network water body extraction method facing to the container cloud according to the embodiment of the present invention includes the following steps:
step S1, acquiring the spectral characteristics of the water body in advance, and selecting wave bands to form spectral vectors;
step S2, converting the spectral vector to obtain a spectral feature matrix, and using the spectral feature matrix as an input feature of the convolutional neural network model;
step S3, taking the spectral feature matrix as a sample to obtain a water body extraction model;
step S4, carrying out object segmentation on the remote sensing image needing water body extraction by using a multi-resolution segmentation algorithm, and identifying each object by using a unique ID;
step S5, converting the data format of the remote sensing image needing water body extraction into the data format of parquet, reading the converted data in a Spark DataSet type by using a Spark Read method, and re-partitioning the Read DataSet in a line-based manner;
in step S6, each object is traversed in the DataSet of the classification result, information on the classification of pixels in the object range is extracted, the category with the largest number of votes is found by the voting method, and the category is used as the classification result of the object.
Wherein the composing of the spectral vectors comprises the steps of:
acquiring a spectral reflectivity curve of a ground object in advance;
screening wave bands used for water body extraction in the images, and synthesizing the wave bands into a multi-band image;
collecting pixels of a water body and other ground objects for the synthesized multiband images, and carrying out category marking on the selected pixels;
a pixel is determined to contain a plurality of image band values that together constitute a spectral vector for the pixel.
Wherein the spectral feature matrix further comprises the steps of:
multiplying the spectrum vector by the transpose of the spectrum vector to obtain a spectrum characteristic matrix;
a spectral feature matrix, represented as:
S＝s T ·s
wherein s is the spectral vector in the ground feature pixel, s T S is the spectral feature matrix for each pixel, which is the transposed vector of the spectral vector.
The water body extraction model comprises the following steps:
calibrating the spectral feature matrix as a sample;
and (3) adding the following components in percentage by weight of 7:3, randomly dividing the sample into training sample data and verification sample data;
training a convolutional neural network water body model by using training sample data;
and verifying the water body model of the convolutional neural network by using the verification sample data.
Wherein traversing each object in the DataSet of the classification result further comprises:
packing the Spark application program into an executable program and a trained convolutional neural network water body model;
and calling an interface of the convolutional neural network model to classify images of each pixel in the DataSet of each partition, and outputting the DataSet of a classification result.
By means of the technical scheme, through analyzing the spectral characteristics of the water body, selecting proper wave bands to form spectral vectors, converting the spectral vectors into spectral characteristic matrixes, using the spectral characteristic matrixes as the input characteristics of the convolutional neural network to improve the accuracy of water body extraction, realizing that cloud computing is used as a basic resource pool, using a container virtualization technology to provide elastic resources to provide computing resources for a convolutional neural network water body extraction method based on Spark memory parallelism, and automatically planning the number of Spark data partitions and the number of actuators for submitting tasks according to the current resource condition of the system, the method is favorable for cloud service management of remote sensing image water body extraction algorithms based on deep learning or a traditional exponential method and the like, and can effectively inhibit the influence of shadows on water body extraction by comprehensively using the spectral characteristics and the spatial characteristics to extract the water body; meanwhile, the efficiency is obviously superior to that in a single machine mode by utilizing parallel optimization of the container cloud and the Spark, and the efficiency is improved more obviously along with the increase of the data volume.
In addition, specifically, as shown in fig. 2 to 4, a convolutional neural network water body extraction model is constructed in advance; then carrying out multi-scale segmentation on the image to be subjected to water body extraction; and finally, combining a Spark big data processing technology and a convolutional neural network model in the container cloud to extract the water body.
Selecting bands B02, B03, B04, B05, B06, B07, B08, B11 and B12 of the Sentinel-2 by analyzing the reflectivity difference of the water body and the shadow on the water body; meanwhile, in order to better highlight the difference between the water body and the shadow, the normalized water body index (NDWI) and the urban shadow index USI are used as independent wave bands, and finally the multiband remote sensing image is synthesized.
And selecting pixels of the water body and other ground objects from the synthesized multiband remote sensing image, and marking the selected pixels with corresponding category labels, such as water bodies, shadows, roads, vegetation and the like.
Reading the spectral values of all the wave bands of the selected feature vector of each pixel according to the wave Band sequence in the synthetic image, and forming vectors according to the sequence of [ Band2, Band3, Band4, Band5, Band6, Band7, Band8, Band11, Band12, NDWI and USI ], wherein the vectors respectively correspond to a blue wave Band, a green wave Band, a red wave Band, three vegetation red edges, near infrared and two infrared short waves.
The marked data is randomly scrambled, then the sample data is divided into two parts according to the proportion of 7:3, 70% of the data size is used as training data, and 30% of the data is used as test data.
Multiplying the spectrum vector of the training sample data by the transposed vector of the training sample data to obtain a corresponding spectrum characteristic matrix, which is expressed as: s ═ S T ·s。
A convolutional neural network water body extraction model is constructed, the obtained spectral feature characteristic matrix is used as an input layer, and the specific model parameters are configured as follows:
an input layer: 11x11 spectral feature matrix;
first convolutional layer: using 32 convolution kernels of 3x3 with the step size of 1 to obtain 32 feature mapping groups of 10x 10;
a second convolutional layer: using 32 convolution kernels of 3x3 with the step size of 1 to obtain 32 feature mapping groups of 8x 8;
a pooling layer: performing mean pooling on the 32 feature mapping groups obtained in the second step to obtain 32 feature mapping groups of 4 × 4;
a third convolutional layer: using 32 convolution kernels of 3x3 with the step size of 1 to obtain 24 feature mapping groups of 4x 4;
an output layer: and a full connection layer, and 29 neurons are obtained after the activation function.
The method comprises the steps of utilizing a multi-resolution segmentation algorithm to segment a synthetic image to be subjected to water body extraction, counting the change rate of a multi-resolution segmentation result, and when a peak value appears in a change rate curve, the change rate curve is possibly the optimal segmentation scale, and the change curve is shown in figure 3.
In addition, in order to extract the fine water body better, the minimum scale is selected as the segmentation scale, and the image to be subjected to water body extraction is subjected to multi-resolution segmentation by using the selected segmentation scale. Preprocessing an image to be subjected to water body extraction into a data format of a partial, reading the data format of the partial into a DataSet of Spark by Spark, re-partitioning by using a replay method of Spark, calling a water body extraction model in mapFunction, and storing a water body extraction result in a new DataSet.
Further, the classification information of each pixel in each object of the newly generated DataSet is obtained by traversing each object after the multi-scale division, and the class having the largest number of votes is obtained by the voting method, and is used as the classification result of the object. And finally packaging the program into an executable program.
And copying the trained convolutional neural network water model and the packaged Spark application program into a Docker container through a script of the Dockerfile.
And randomly combining the number of the actuators of the Spark and the number of the data partitions in the task submitting script to submit the task and execute the result. And obtaining the data partition size and the execution time consumption of the number of actuators through multiple experiments.
In addition, the resources required for each container to run are estimated, and then the maximum number of containers supported is calculated from all the resources in the cluster. Assuming that the available resources of the cluster are n and the resources required for each container to run are s, the maximum number of containers that the cluster can support is calculated as follows: and T is n/s.
The method comprises the steps of calculating the size of a data partition and the number of actuators of an image to be classified submitted by Matser, and setting the number of Spark executors, the size of the data partition and the storage position of the remote sensing image to be processed in a task submission script.
Compared with the operation mode of a single machine through Spark in a Local mode, the execution efficiency of water body extraction is remarkably improved under the condition of increasing the data volume.
In addition, as shown in fig. 4, the invention provides that the efficiency of the convolutional neural network water extraction facing to the container cloud is obviously improved by combining the container blurring technology. The water body extraction method can be executed in a single container, is very suitable for the current popular micro service, and can provide water body extraction service for a plurality of users under the condition of resource permission.
In summary, with the aid of the technical solution of the present invention, the Spark-based memory-parallel water body extraction model is migrated from the traditional Spark cluster to a container cloud environment, and a finer-grained container virtualization is used for scheduling, so as to improve the resource utilization rate. On the basis of a model of a parallel remote sensing water body extraction algorithm, the algorithm is abstracted into an independent container process in a container cloud environment. And computing resources are distributed according to the size of the calculated amount as required, so that the resource utilization rate can be improved. The data are distributed to different containers for execution, so that the concurrency of the algorithm and the data processing efficiency can be improved. The container is used as a carrier of the algorithm to provide service, and the API is opened in the form of remote sensing micro service and provided for users, so that the threshold of remote sensing use can be reduced.
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 (3)

1. A convolutional neural network water body extraction method facing to container cloud is characterized by comprising the following steps:
acquiring spectral characteristics of a water body in advance, and selecting wave bands to form spectral vectors;
converting the spectrum vector to obtain a spectrum characteristic matrix, and using the spectrum characteristic matrix as an input characteristic of a convolutional neural network model;
calibrating the spectral feature matrix as a sample;
and (3) adding the following components in percentage by weight of 7:3, randomly dividing the sample into training sample data and verification sample data;
training a convolutional neural network water body model by using training sample data;
verifying the water body model of the convolutional neural network by verification sample data;
carrying out object segmentation on the remote sensing image needing water body extraction by using a multi-resolution segmentation algorithm, and using a unique ID to identify each object;
converting the data format of the remote sensing image needing to be subjected to water body extraction into a data format of a request, reading the converted data in a Spark DataSet type by using a Spark Read method, and re-partitioning the Read DataSet in a line mode;
copying the trained convolutional neural network water model and the packed Spark application program into a Docker container through a Dockerfile script; calling an interface of the convolutional neural network model to classify images of each pixel in the DataSet of each partition, and outputting a DataSet of a classification result;
traversing each object in the DataSet of the classification result, extracting the classification information of pixels in the object range, obtaining the category with the most votes according to a voting method, and taking the category as the classification result of the object;
randomly combining the number of actuators of Spark and the number of data partitions in a task submitting script to submit a task, and executing results to obtain the size of the data partitions and the execution time consumption condition of the number of the actuators through multiple experiments;
estimating the resources required by the operation of each container, and then calculating the maximum supported number of containers according to all the resources in the cluster;
the method comprises the steps of calculating the size of a data partition and the number of actuators of an image to be classified submitted by Matser, and setting the number of Spark executors, the size of the data partition and the storage position of a remote sensing image to be processed in a task submission script;
the comparison is made by Local mode Spark and stand-alone mode of operation.
2. The vessel cloud oriented convolutional neural network water extraction method of claim 1, wherein the composition of the spectral vectors comprises the steps of:
acquiring a spectral reflectivity curve of a ground object in advance;
screening wave bands used for water body extraction in the images, and synthesizing the wave bands into a multi-band image;
collecting pixels of a water body and other ground objects for the synthesized multiband images, and carrying out category marking on the selected pixels;
a pixel is determined to contain a plurality of image band values that together form a spectral vector for the pixel.
3. The vessel cloud oriented convolutional neural network water body extraction method of claim 1, wherein the spectral feature matrix further comprises the following steps:
multiplying the spectrum vector by the transpose of the spectrum vector to obtain a spectrum characteristic matrix;
a spectral feature matrix, represented as:
S＝s T ·s
where s is the spectral vector in the ground feature pixel, s T S is the spectral feature matrix for each pixel, which is the transposed vector of the spectral vector.
CN202010789924.XA 2020-08-07 2020-08-07 Convolutional neural network water body extraction method for container cloud Active CN111985368B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010789924.XA CN111985368B (en) 2020-08-07 2020-08-07 Convolutional neural network water body extraction method for container cloud

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010789924.XA CN111985368B (en) 2020-08-07 2020-08-07 Convolutional neural network water body extraction method for container cloud

Publications (2)

Publication Number Publication Date
CN111985368A CN111985368A (en) 2020-11-24
CN111985368B true CN111985368B (en) 2022-08-30

Family

ID=73445709

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010789924.XA Active CN111985368B (en) 2020-08-07 2020-08-07 Convolutional neural network water body extraction method for container cloud

Country Status (1)

Country Link
CN (1) CN111985368B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113960152B (en) * 2021-10-18 2023-06-06 北方工业大学 Trace explosive identification method and system based on ion mobility spectrometry
CN115690593A (en) * 2022-03-31 2023-02-03 中国科学院空天信息创新研究院 Land classification method and device and cloud server

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104156964A (en) * 2014-08-14 2014-11-19 陈荣元 Remote sensing image area segmentation method integrating Markov random field (MRF) and Bayesian network (BN)
WO2015077493A1 (en) * 2013-11-20 2015-05-28 Digimarc Corporation Sensor-synchronized spectrally-structured-light imaging
CN107688776A (en) * 2017-07-21 2018-02-13 同济大学 A kind of urban water-body extracting method
CN109598202A (en) * 2018-11-05 2019-04-09 中国科学院遥感与数字地球研究所 A kind of object-based satellite image multi objective built-up areas extraction method
CN111104889A (en) * 2019-12-04 2020-05-05 山东科技大学 Water body remote sensing identification method based on U-net
CN111476170A (en) * 2020-04-09 2020-07-31 首都师范大学 Remote sensing image semantic segmentation method combining deep learning and random forest

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015077493A1 (en) * 2013-11-20 2015-05-28 Digimarc Corporation Sensor-synchronized spectrally-structured-light imaging
CN104156964A (en) * 2014-08-14 2014-11-19 陈荣元 Remote sensing image area segmentation method integrating Markov random field (MRF) and Bayesian network (BN)
CN107688776A (en) * 2017-07-21 2018-02-13 同济大学 A kind of urban water-body extracting method
CN109598202A (en) * 2018-11-05 2019-04-09 中国科学院遥感与数字地球研究所 A kind of object-based satellite image multi objective built-up areas extraction method
CN111104889A (en) * 2019-12-04 2020-05-05 山东科技大学 Water body remote sensing identification method based on U-net
CN111476170A (en) * 2020-04-09 2020-07-31 首都师范大学 Remote sensing image semantic segmentation method combining deep learning and random forest

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
自然保护区遥感信息提取技术研究进展;吴见等;《世界林业研究》;20130228(第01期);第55-60页 *

Also Published As

Publication number Publication date
CN111985368A (en) 2020-11-24

Similar Documents

Publication Publication Date Title
Li et al. Superpixel segmentation using linear spectral clustering
US20230185532A1 (en) Exploiting activation sparsity in deep neural networks
CN107169956B (en) Color woven fabric defect detection method based on convolutional neural network
Sameen et al. Classification of very high resolution aerial photos using spectral‐spatial convolutional neural networks
Lu et al. Adaptive object detection using adjacency and zoom prediction
CN111985368B (en) Convolutional neural network water body extraction method for container cloud
CN112819110B (en) Incremental small sample target detection method and system based on weight generation
CN109558806A (en) The detection method and system of high score Remote Sensing Imagery Change
CN103177265B (en) High-definition image classification method based on kernel function Yu sparse coding
CN117274608B (en) Remote sensing image semantic segmentation method based on space detail perception and attention guidance
CN104951791A (en) Data classification method and apparatus
Giveki Scale-space multi-view bag of words for scene categorization
CN113435254A (en) Sentinel second image-based farmland deep learning extraction method
CN110163294A (en) Remote Sensing Imagery Change method for detecting area based on dimensionality reduction operation and convolutional network
Pu et al. Design and implementation of a parallel geographically weighted k-nearest neighbor classifier
CN105069459B (en) One kind is directed to High Resolution SAR Images type of ground objects extracting method
CN112966815A (en) Target detection method, system and equipment based on impulse neural network
CN117152105A (en) Method, device, equipment and medium for estimating yield of litchi in orchard
CN116524357A (en) High-voltage line bird nest detection method, model training method, device and equipment
CN116051532A (en) Deep learning-based industrial part defect detection method and system and electronic equipment
Huang et al. A flower classification framework based on ensemble of cnns
CN109711376B (en) Multi-scale sparse blue noise sampling method based on optimal transmission theory
Anggiratih et al. Ship identification on satellite image using convolutional neural network and random forest
Le Goff et al. Distributed boosting for cloud detection
CN111353976A (en) Sand grain target detection method based on convolutional neural 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