CN108509980B - Water level monitoring method based on dictionary learning - Google Patents

Water level monitoring method based on dictionary learning Download PDF

Info

Publication number
CN108509980B
CN108509980B CN201810188900.1A CN201810188900A CN108509980B CN 108509980 B CN108509980 B CN 108509980B CN 201810188900 A CN201810188900 A CN 201810188900A CN 108509980 B CN108509980 B CN 108509980B
Authority
CN
China
Prior art keywords
class
dictionary
matrix
training
images
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
CN201810188900.1A
Other languages
Chinese (zh)
Other versions
CN108509980A (en
Inventor
桂冠
潘金秋
熊健
范山岗
杨洁
孙颖异
周天
樊亚萍
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.)
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing University of Posts and Telecommunications
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 Nanjing University of Posts and Telecommunications filed Critical Nanjing University of Posts and Telecommunications
Priority to CN201810188900.1A priority Critical patent/CN108509980B/en
Publication of CN108509980A publication Critical patent/CN108509980A/en
Application granted granted Critical
Publication of CN108509980B publication Critical patent/CN108509980B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/28—Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01F—MEASURING VOLUME, VOLUME FLOW, MASS FLOW OR LIQUID LEVEL; METERING BY VOLUME
    • G01F23/00—Indicating or measuring liquid level or level of fluent solid material, e.g. indicating in terms of volume or indicating by means of an alarm
    • G01F23/04—Indicating or measuring liquid level or level of fluent solid material, e.g. indicating in terms of volume or indicating by means of an alarm by dip members, e.g. dip-sticks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques

Abstract

The invention discloses a water level monitoring method based on dictionary learning, which comprises the following steps: step 1, fixing a measuring marker post in a river area to be measured, and acquiring images of the measuring marker post and an area near a water surface in real time through a camera; step 2, collecting a plurality of images as training images, dividing each image into C types, collecting training data for each type of image block by adopting a sliding window, and marking corresponding type labels; step 3, training the training data by adopting a dictionary learning method to obtain a dictionary D; and 4, classifying the images acquired by the camera in real time by using the dictionary D to obtain class label vectors of the images so as to obtain water level values. The method can be realized only by installing the traditional marker post, overcomes the problem that the horizontal position is difficult to accurately identify in the traditional water level measurement scheme, and has higher stability and robustness.

Description

Water level monitoring method based on dictionary learning
Technical Field
The invention relates to the field of computer vision image processing, in particular to a water level monitoring method based on dictionary learning.
Background
Because the problems of water safety and water resource seriously affect the sustainable development of the economic society and the life and property safety of people, the real-time monitoring of the water level of rivers, lakes, rivers, reservoirs and the like is very important. The current water level monitoring technology is mainly to establish a video monitoring system in the areas and acquire the water level value through manual real-time observation.
The video monitoring method is mainly used only for utilizing the real-time browsing function of videos, particularly, the monitoring of the flood disasters in rainy seasons still needs 24 hours of manual uninterrupted observation, a large amount of manpower and material resources are consumed, the efficiency is low, the manual observation method is easily influenced by the surrounding environment, and if the water level is difficult to distinguish under the backlight condition or the condition of poor light at night, the current requirements cannot be met.
The invention provides a water level monitoring method based on dictionary learning, innovatively introduces a novel technology of dictionary learning, accurately determines the specific position of a horizontal plane by utilizing the dictionary learning, is less influenced by the environment, and liberates manpower and material resources.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a water level monitoring method based on dictionary learning, and solves the problems that the traditional water level monitoring method requires human eyes to observe water level, is low in efficiency and is difficult to distinguish.
In order to achieve the above purpose, the invention adopts the following technical scheme: a water level monitoring method based on dictionary learning comprises the following steps:
step 1, fixing a measuring marker post in a river area to be measured, and acquiring images of the measuring marker post and an area near a water surface in real time through a camera;
step 2, collecting a plurality of images as training images, dividing each image into C types, collecting training data for each type of image block by adopting a sliding window, and marking corresponding type labels;
step 3, training the training data by adopting a dictionary learning method to obtain a dictionary D;
and 4, classifying the images acquired by the camera in real time by using the dictionary D to obtain class label vectors of the images so as to obtain water level values.
The water level monitoring method based on dictionary learning is characterized in that: the step 2 specifically comprises the following steps:
step 2.1, collecting a plurality of images of the measuring marker post and the area near the water surface as training images, and converting the RGB training images into gray images;
2.2, dividing the marker post and the hydrology into C types based on the structural features and the hydrology texture features of the marker post in the image;
step 2.3, acquiring pixel gray values of the gray images as samples by adopting an mxn sliding window in each type of interested region, wherein the step length is s, m, n and s units are pixels;
step 2.4, elongating each m × n sample matrix into a column vector to obtain a sample vector;
and 2.5, synthesizing a training sample matrix Y by using all sample vectors, wherein each column of the matrix represents a sample, the class label corresponding to the sample collected in the C-th type interested area is C, C is 1-C, and the class label of each column of samples is adopted to form a class label vector L.
The water level monitoring method based on dictionary learning is characterized in that: the RGB training image is converted into a gray image, and particularly an algorithm for carrying out weighted average on R, G, B components is adopted.
The water level monitoring method based on dictionary learning is characterized in that: in the step 3, training data is trained by using a dictionary learning method to obtain a dictionary D, and the specific steps include:
step 3.1, inputting the training sample matrix Y and the class label vector L into a dictionary learning network for training, and obtaining a convergent dictionary D through iteration for a plurality of times;
the dictionary learning model is as shown in formula (1):
Figure BDA0001591117110000021
wherein j is an integer 1-NcY is a training sample matrix, [ Y ]1,Y2...YC]＝Y,Y1～YCRespectively represent training sample matrices belonging to classes 1 to C, X is a representation matrix of a training sample matrix Y based on a dictionary D, [ X [ ]1,X2...XC]＝X,X1～XCA representation matrix representing class 1 to class C samples,
Figure BDA0001591117110000031
Xca representation matrix representing class c samples, NcRepresents the number of class c samples,
Figure BDA0001591117110000032
respectively represent the 1 stcA representative vector of class c samples, [ d ]1,d2...dk]D, k being the number of columns in the dictionary D, D1～dkK column vectors, P, respectively representing the dictionary DcA matrix of the extraction coefficients is represented,
Figure BDA0001591117110000033
a laplacian matrix representing class c samples, β, λ, γ are three scalar factors,
Figure BDA0001591117110000034
the square of the F-norm of the matrix is represented,
Figure BDA0001591117110000035
represents the square of the 2 norm of the matrix, tr () represents the trace operator of the matrix, s.t. represents the constraint, | dk||2To representdk is a 2-norm of k,
Figure BDA0001591117110000036
represents an arbitrary k;
Figure BDA0001591117110000037
which is representative of the error of the reconstruction,
Figure BDA0001591117110000038
the representation represents a regularization of the representation,
Figure BDA0001591117110000039
is the cross-over suppression term or terms,
Figure BDA00015911171100000310
is a group regularization term;
and 3.2, outputting the trained dictionary D.
The water level monitoring method based on dictionary learning is characterized in that: extracting coefficient matrix P from the cross suppression termcExpressed as shown in equation (2):
Figure BDA00015911171100000311
wherein a, b are matrices PcIndex of row, column, ξcSet of indices, ξ, representing the columns in dictionary D associated with class c samples0The index set representing the column in dictionary D associated with the class 0 sample, ξ represents the index set of all columns of dictionary D.
The water level monitoring method based on dictionary learning is characterized in that: in the step 4, the images collected by the camera are classified in real time by using the dictionary D to obtain the label of each block of the image, so that the water level value is obtained, and the method specifically comprises the following steps:
step 4.1, acquiring a test image through a camera, acquiring a test matrix y from the test image by the method in the step 2, and calculating an expression matrix X' of the test matrix y by a formula (3):
X'＝(DTD+βI)-1DTy (7)
wherein, I is an identity matrix;
step 4.2, the class label vector label (y) of the test matrix y is obtained by the formula (4):
Figure BDA0001591117110000041
where h is an integer belonging to the index set of columns in dictionary D associated with class 0 and class c samples, X '(h) is the h-th column representing matrix X', ξcSet of indices, ξ, representing columns in dictionary D associated with class c samples0Indicating characterIndex set of columns in dictionary D associated with class 0 samples, DkK column vectors representing dictionary D;
step 4.3, the class label corresponding to the class C sample is C, and the class labels 1-C1Represents C1Personal tags, category tags C1+1 to C represent C2Traversing the class label vector label (y) to find out C1And C1The +1 dividing position is the position of the water level.
The water level monitoring method based on dictionary learning is characterized in that: the parts with similar structures of the marker post are classified into the same class, the parts with similar hydrological textures are classified into the same class, and the marker post is classified into C1Class is called flagpole class, and divides hydrology into C2The classes are called hydrologic classes, the benchmarks and hydrologic classes are combined to form C class, and C is equal to C1+C2。
The invention achieves the following beneficial effects: the method can be realized only by installing the traditional marker post, creatively introduces a dictionary learning method, extracts the characteristics of the marker post and the hydrology through computer iterative training, and can accurately identify the marker post and the hydrology so as to obtain the water level value. The invention overcomes the problem that the horizontal position is difficult to accurately identify in the traditional water level measuring scheme, and has higher stability and robustness.
Drawings
FIG. 1 is a flow chart of a water level monitoring method based on dictionary learning;
FIG. 2 is a schematic diagram of a training image feature extraction and dictionary training phase;
FIG. 3 is a flow chart of the water level real-time discrimination stage.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
As shown in fig. 1, a water level monitoring method based on dictionary learning includes the following steps:
step 1, fixing a measuring marker post in a river area to be measured, and acquiring images of the measuring marker post and an area near a water surface in real time through a camera;
step 2, collecting a plurality of images as training images, dividing each image into C types, collecting training data for each type of image block by adopting a sliding window, and marking corresponding type labels;
step 3, training the training data by adopting a dictionary learning method to obtain a dictionary D;
and 4, classifying the images acquired by the camera in real time by using the dictionary D to obtain class label vectors of the images so as to obtain water level values.
Step 2, specifically comprising the following steps:
step 2.1, collecting a plurality of images of the measuring marker post and the area near the water surface as training images, converting the RGB training images into gray images, and specifically adopting an algorithm for carrying out weighted average on R, G, B components;
step 2.2, dividing the marker post into C based on the structural features and the hydrologic texture features of the marker post in the image (namely, the parts with similar marker post structures are divided into the same class, and the parts with similar hydrologic textures are divided into the same class)1Class is called as the flagpole class, and divides hydrology into C2The classes are called hydrologic classes, the benchmarks and hydrologic classes are combined to form C class, and C is equal to C1+C2；
Step 2.3, acquiring pixel gray values of the gray images as samples by adopting an mxn sliding window in each type of interested region, wherein the step length is s, m, n and s units are pixels;
step 2.4, elongating each m × n sample matrix into a column vector to obtain a sample vector;
and 2.5, synthesizing a training sample matrix Y by using all sample vectors, wherein each column of the matrix represents a sample, the class label corresponding to the sample collected in the C-th type interest area is C, and C is 1 … C, and a class label vector L is formed by adopting the class labels of the samples in each column.
Step 3, training the training data by adopting a dictionary learning method to obtain a dictionary D, and the specific steps comprise:
step 3.1, inputting a training sample matrix Y and a class label vector L into a dictionary learning network for training, obtaining a convergent dictionary D through a plurality of iterations, wherein the dictionary D can represent most characteristics of training samples, and each training sample can be represented by a sparse representation matrix X and the dictionary D;
to obtain a compact and discriminating dictionary, each column of the dictionary should have a representative meaning, i.e., each column should be associated with a particular label. Specifically, each class of training samples has p columns associated with it within the dictionary. Furthermore, there may be similar background patterns between samples of different classes, for which reason q columns are reserved in the dictionary D to describe this similarity, and these columns are taken as class 0, q may be 0.
In particular, the amount of the solvent to be used,
the model of dictionary learning is as formula (1):
Figure BDA0001591117110000061
wherein j is an integer 1-NcY is a training sample matrix, [ Y ]1,Y2...YC]＝Y,Y1～YCRespectively representing training sample matrices belonging to classes 1 to C, X is a representation matrix of a sample matrix Y based on a dictionary D, [ X [ ]1,X2...XC]＝X,X1～XCRepresenting the representation matrixes of the class 1 to C training samples respectively,
Figure BDA0001591117110000062
Ncrepresents the number of class c training samples,
Figure BDA0001591117110000063
respectively represent the 1 stcA representative vector of class c training samples, [ d ]1,d2...dk]D, k is the number of columns in dictionary D, D1～dkK column vectors, respectively, representing the dictionary D, β, λ, γ are three scalar factors,
Figure BDA0001591117110000064
the square of the F-norm of the matrix is represented,
Figure BDA0001591117110000065
represents the square of the 2 norm of the matrix, PcRepresents an extraction coefficient matrix, and a specific expression is shown as formula (2), tr () represents a trace operator of the matrix,
Figure BDA0001591117110000066
laplace matrix representing class c samples, s.t. representing constraint condition, | dk||2Denotes dkThe 2-norm of (a) of (b),
Figure BDA0001591117110000067
represents an arbitrary k;
in the formula (1), the first and second groups,
Figure BDA0001591117110000068
which is representative of the error of the reconstruction,
Figure BDA0001591117110000069
the representation represents a regularization of the representation,
Figure BDA00015911171100000610
is the cross-over suppression term or terms,
Figure BDA00015911171100000611
is a group regularization term; the cross suppression items can amplify the difference of the expression matrixes X of different types of samples, and the group regular items can strengthen the similarity of the expression matrixes X of the same type of samples.
Extracting coefficient matrix P from cross suppression itemcExpressed as shown in equation (2):
Figure BDA00015911171100000612
wherein a and b are respectively a matrix PcIndex of row, column, ξcCorrelation with class c samples in representation dictionary DIndex set of linked columns, ξ0Index set representing columns in dictionary D associated with class 0 samples, xi represents index set of all columns in dictionary D, and expression matrix X' for some ith class test sample, i is 1-C, PcThe values representing the columns of matrix X' associated with the non-ith type of test sample can be suppressed so that the relatively large coefficients in the representation matrix are mainly concentrated near the columns associated with the ith type of test sample.
For a number of test samples of the l-th class, i ═ 1-C, the canonical terms are grouped such that their columns in the representation matrix X' associated with the test samples of the l-th class have a large degree of similarity.
And 3.2, outputting the trained dictionary D.
As shown in fig. 2, in step 4, the real-time image classification method specifically includes:
the method adopts the dictionary D obtained by learning to judge the category of the test sample.
Step 4.1, acquiring a test image through the camera in the step one, acquiring a test matrix y from the test image by the method in the step 2, and calculating an expression matrix X' of the test matrix y by a formula (3):
X'＝(DTD+βI)-1DTy (11)
wherein, I is an identity matrix;
step 4.2, as can be seen from the dictionary learning model formula (1), if the test matrix y belongs to the ith class, the larger coefficients in the representation matrix X' are mainly concentrated near the columns associated with the ith class and the columns shared by all the classes, and besides, the columns associated with other classes except the ith class are very small. Thus, the class label vector label (y) of the test matrix y can be obtained from equation (4):
Figure BDA0001591117110000071
where h is an integer belonging to the index set of columns in dictionary D associated with class 0 and class c samples, X '(h) being the h-th column representing matrix X';
step 4.3, due to class cThe class label corresponding to the sample is C, and the class labels 1-C1Represents C1Personal tags, category tags C1+ 1-C for C2Traversing the class label vector label (y) to find C1And C1The +1 dividing position is the position of the water level.
The method can be realized only by installing the traditional marker post, creatively introduces a dictionary learning method, extracts the characteristics of the marker post and the hydrology through computer iterative training, and can accurately identify the marker post and the hydrology so as to obtain the water level value. The invention overcomes the problem that the horizontal position is difficult to accurately identify in the traditional water level measuring scheme, and has higher stability and robustness. The invention is little influenced by natural environment, can accurately measure the real-time water level value under different weather and light conditions, solves the problems that the water level value needs to be manually read in the traditional scheme and a large amount of manpower is consumed, and has higher practical application value.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (6)

1. A water level monitoring method based on dictionary learning comprises the following steps:
step 1, fixing a measuring marker post in a river area to be measured, and acquiring images of the measuring marker post and an area near a water surface in real time through a camera;
step 2, collecting a plurality of images as training images, dividing each image into C types, collecting training data for each type of image block by adopting a sliding window, and marking corresponding type labels;
step 3, training the training data by adopting a dictionary learning method to obtain a dictionary D;
step 4, classifying the images acquired by the camera in real time by using the dictionary D to obtain class label vectors of the images so as to obtain water level values;
wherein, the step 4 comprises the following steps:
step 4.1, acquiring a test image through a camera, acquiring a test matrix y from the test image by the method in the step 2, and calculating an expression matrix X' of the test matrix y by a formula (3):
X'＝(DTD+βI)-1DTy (3)
wherein, I is an identity matrix;
step 4.2, the class label vector label (y) of the test matrix y is obtained by the formula (4):
Figure FDA0003634186330000011
where h is an integer belonging to the index set of columns in dictionary D associated with class 0 and class c samples, X '(h) is the h-th column representing matrix X', ξcSet of indices, ξ, representing the columns in dictionary D associated with class c samples0Set of indices representing columns in dictionary D associated with class 0 samples, DkK column vectors representing dictionary D;
step 4.3, the class label corresponding to the class-C sample is C, and the class labels 1-C1Represents C1Personal tags, category tags C1+ 1-C for C2Traversing the class label vector label (y) to find C1And C1The +1 dividing position is the position of the water level.
2. The water level monitoring method based on dictionary learning as claimed in claim 1, wherein: the step 2 specifically comprises the following steps:
step 2.1, collecting a plurality of images of the measuring marker post and the area near the water surface as training images, and converting the RGB training images into gray images;
2.2, dividing the marker post and the hydrology into C types based on the structural features and the hydrology texture features of the marker post in the image;
step 2.3, acquiring pixel gray values of gray images as samples by adopting an mxn sliding window in each type of region of interest, wherein the step length is s, m, n and s units are pixels;
step 2.4, elongating each m multiplied by n sample matrix into a column vector to obtain a sample vector;
and 2.5, synthesizing a training sample matrix Y by using all sample vectors, wherein each column of the matrix represents a sample, the class label corresponding to the sample collected in the C-th type interested area is C, C is 1-C, and the class label of each column of samples is adopted to form a class label vector L.
3. The water level monitoring method based on dictionary learning as claimed in claim 2, wherein: the RGB training image is converted into a gray image, and particularly an algorithm for carrying out weighted average on R, G, B components is adopted.
4. The water level monitoring method based on dictionary learning as claimed in claim 2, wherein: in the step 3, training data is trained by using a dictionary learning method to obtain a dictionary D, and the specific steps include:
step 3.1, inputting the training sample matrix Y and the class label vector L into a dictionary learning network for training, and obtaining a convergent dictionary D through iteration for a plurality of times;
the dictionary learning model is as shown in formula (1):
Figure FDA0003634186330000021
wherein j is an integer 1-NcY is a training sample matrix, [ Y ]1,Y2...YC]＝Y,Y1～YCRespectively represent training sample matrices belonging to classes 1 to C, X is a representation matrix of a training sample matrix Y based on a dictionary D, [ X [ ]1,X2...XC]＝X,X1～XCA representation matrix representing class 1 to class C samples,
Figure FDA0003634186330000022
Xca representation matrix representing class c samples, NcRepresents the number of class c samples,
Figure FDA0003634186330000023
respectively represent the 1 stcA representative vector of class c samples, [ d ]1,d2...dk]D, k is the number of columns in dictionary D, D1～dkK column vectors, P, respectively representing the dictionary DcRepresenting the matrix of extraction coefficients, LcA laplacian matrix representing class c samples, β, λ, γ are three scalar factors,
Figure FDA0003634186330000031
the square of the F-norm of the matrix is represented,
Figure FDA0003634186330000032
represents the square of the 2 norm of the matrix, tr () represents the trace operator of the matrix, s.t. represents the constraint, | dk||2Denotes dkThe 2-norm of (a) of (b),
Figure FDA0003634186330000033
represents an arbitrary k;
Figure FDA0003634186330000034
which is representative of the error of the reconstruction,
Figure FDA0003634186330000035
the representation represents a regularization of the representation,
Figure FDA0003634186330000036
is a cross-over suppression term, γ tr (X)cLc(Xc)T) Is a group regularization term;
and 3.2, outputting the trained dictionary D.
5. The water level monitoring method based on dictionary learning as claimed in claim 4, wherein: extracting coefficient matrix P from the cross suppression termcExpressed as shown in equation (2):
Figure FDA0003634186330000037
wherein a, b are matrices PcIndex of row and column, ξcSet of indices, ξ, representing the columns in dictionary D associated with class c samples0The index set representing the column in dictionary D associated with the class 0 sample, ξ represents the index set of all columns of dictionary D.
6. The water level monitoring method based on dictionary learning as claimed in claim 1, wherein: the parts with similar structures of the marker post are classified into the same class, the parts with similar hydrological textures are classified into the same class, and the marker post is classified into C1Class is called flagpole class, and divides hydrology into C2The classes are called hydrologic classes, the benchmarks and hydrologic classes are combined to form C class, and C is equal to C1+C2。
CN201810188900.1A 2018-03-08 2018-03-08 Water level monitoring method based on dictionary learning Active CN108509980B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810188900.1A CN108509980B (en) 2018-03-08 2018-03-08 Water level monitoring method based on dictionary learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810188900.1A CN108509980B (en) 2018-03-08 2018-03-08 Water level monitoring method based on dictionary learning

Publications (2)

Publication Number Publication Date
CN108509980A CN108509980A (en) 2018-09-07
CN108509980B true CN108509980B (en) 2022-07-01

Family

ID=63377350

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810188900.1A Active CN108509980B (en) 2018-03-08 2018-03-08 Water level monitoring method based on dictionary learning

Country Status (1)

Country Link
CN (1) CN108509980B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110705759B (en) * 2019-09-18 2022-06-03 平安科技（深圳）有限公司 Water level early warning monitoring method and device, storage medium and electronic equipment
CN111898459A (en) * 2020-07-07 2020-11-06 南瑞集团有限公司 Power transmission line vehicle approach detection method and device and storage medium
CN112991425B (en) * 2021-04-28 2021-08-06 武汉光谷信息技术股份有限公司 Water area water level extraction method and system and storage medium
CN113609602B (en) * 2021-06-21 2022-11-18 国网浙江省电力有限公司湖州供电公司 Unloading type multi-drive conveyor fault monitoring method based on dictionary learning
CN114387235B (en) * 2021-12-30 2024-02-09 重庆知行数联智能科技有限责任公司 Water environment monitoring method and system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN203721006U (en) * 2014-01-03 2014-07-16 南京信息工程大学 Water regimen monitoring early warning device, control center and system based on multiple sensors
CN107131925A (en) * 2017-04-28 2017-09-05 南京邮电大学 A kind of water level real-time monitoring method based on image procossing
CN107506798A (en) * 2017-08-31 2017-12-22 福建四创软件有限公司 A kind of water level monitoring method based on image recognition
CN107631782A (en) * 2017-07-18 2018-01-26 南京邮电大学 A kind of level testing methods based on Harris Corner Detections

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN203721006U (en) * 2014-01-03 2014-07-16 南京信息工程大学 Water regimen monitoring early warning device, control center and system based on multiple sensors
CN107131925A (en) * 2017-04-28 2017-09-05 南京邮电大学 A kind of water level real-time monitoring method based on image procossing
CN107631782A (en) * 2017-07-18 2018-01-26 南京邮电大学 A kind of level testing methods based on Harris Corner Detections
CN107506798A (en) * 2017-08-31 2017-12-22 福建四创软件有限公司 A kind of water level monitoring method based on image recognition

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于图像处理和稀疏表示的水位识别研究;高姗姗等;《人民黄河》;20161231;第38卷(第12期);第1-2章 *

Also Published As

Publication number Publication date
CN108509980A (en) 2018-09-07

Similar Documents

Publication Publication Date Title
CN108509980B (en) Water level monitoring method based on dictionary learning
Yang et al. Detecting and mapping tree crowns based on convolutional neural network and Google Earth images
CN110097115B (en) Video salient object detection method based on attention transfer mechanism
CN105989336B (en) Scene recognition method based on deconvolution deep network learning with weight
CN103735269B (en) A kind of height measurement method followed the tracks of based on video multi-target
CN109635726B (en) Landslide identification method based on combination of symmetric deep network and multi-scale pooling
CN113313082B (en) Target detection method and system based on multitask loss function
CN110598613B (en) Expressway agglomerate fog monitoring method
CN109145962B (en) Atmospheric parameter inversion observation method based on digital image
CN112766155A (en) Deep learning-based mariculture area extraction method
CN113111947B (en) Image processing method, apparatus and computer readable storage medium
CN107767416A (en) The recognition methods of pedestrian's direction in a kind of low-resolution image
Liang et al. Maximum likelihood classification of soil remote sensing image based on deep learning
CN111291818B (en) Non-uniform class sample equalization method for cloud mask
Yan et al. Glacier classification from Sentinel-2 imagery using spatial-spectral attention convolutional model
CN114881286A (en) Short-time rainfall prediction method based on deep learning
CN106980823A (en) A kind of action identification method based on interframe self similarity
Liu et al. TransCloudSeg: Ground-based cloud image segmentation with transformer
CN110909645B (en) Crowd counting method based on semi-supervised manifold embedding
CN114511787A (en) Neural network-based remote sensing image ground feature information generation method and system
CN109815889B (en) Cross-resolution face recognition method based on feature representation set
Guo et al. A water-level measurement method using sparse representation
Yu et al. An image-based automatic recognition method for the flowering stage of maize
CN114693600A (en) Semi-supervised learning method for carrying out nucleus segmentation on tissue pathology image
Robertson et al. Aerial image segmentation for flood risk analysis

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