CN110223341B - Intelligent water level monitoring method based on image recognition - Google Patents
Intelligent water level monitoring method based on image recognition Download PDFInfo
- Publication number
- CN110223341B CN110223341B CN201910517818.3A CN201910517818A CN110223341B CN 110223341 B CN110223341 B CN 110223341B CN 201910517818 A CN201910517818 A CN 201910517818A CN 110223341 B CN110223341 B CN 110223341B
- Authority
- CN
- China
- Prior art keywords
- water gauge
- water
- image
- network
- gauge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 234
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012544 monitoring process Methods 0.000 title claims abstract description 18
- 238000005520 cutting process Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000013528 artificial neural network Methods 0.000 claims abstract description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 25
- 238000000605 extraction Methods 0.000 claims description 20
- 238000010586 diagram Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000005259 measurement Methods 0.000 abstract description 4
- 238000001514 detection method Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 238000005286 illumination Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 239000013589 supplement Substances 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000004438 eyesight Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 102000010637 Aquaporins Human genes 0.000 description 1
- 208000034699 Vitreous floaters Diseases 0.000 description 1
- 101100524346 Xenopus laevis req-a gene Proteins 0.000 description 1
- 101100524347 Xenopus laevis req-b gene Proteins 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 239000003673 groundwater Substances 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000012625 in-situ measurement Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004297 night vision Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000013049 sediment Substances 0.000 description 1
- 230000009469 supplementation Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Fluid Mechanics (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an intelligent water level monitoring method based on image recognition, which comprises the following steps: receiving image data transmitted from an image pickup apparatus in a region to be detected; detecting and identifying a water gauge part above the water surface in the image data by adopting a depth neural network, and dividing a water gauge image above the water surface by adopting a graph cutting algorithm on the basis; calculating the height of the water gauge above the water surface according to the divided water gauge images, and combining the overall height of the water gauge to obtain the height of the water gauge below the water surface; the water level value can be calculated by acquiring the elevation of the bottom of the water gauge of the actual station and the elevation of the water gauge below the water surface. According to the invention, the water level is not needed to be surveyed manually on site, the image acquisition is realized in a remote mode, the analysis processing is carried out locally, the water level value is calculated, the manual operation cost is greatly reduced, and the measurement accuracy is improved.
Description
Technical Field
The invention relates to the technical field of water level monitoring, in particular to an intelligent water level monitoring method based on image recognition.
Background
The water level observation means an in-situ measurement of water level of rivers, lakes, groundwater, etc. The water level data is closely related to the life and production of human society. The planning, design, construction and management of hydraulic engineering require water level data. The engineering construction of bridges, ports, channels, water supply and drainage and the like also needs water level data. In flood control and drought resistance, water level data is more important, and is the basis of hydrologic forecasting and hydrologic information. The water level data is important basic data in the research of water level flow relation and the analysis of river sediment, ice condition, etc.
Typically measured using a water gauge. The water gauge is a traditional and effective direct observation device. In real time, the water level is obtained by adding the zero point elevation of the water gauge to the reading on the water gauge. The observation time and the observation times are suitable for the water level change process in one day, and the requirements of hydrologic forecast and hydrologic information are met. In general, the daily measurement is carried out1 to 2 times. When flood, icing, ice flowing, ice dam generation and ice and snow melting are carried out to supply the river, the observation times are increased, so that the measured result can completely reflect the water level change process.
The existing mode is that manual site survey is adopted, so that labor is consumed, personal safety is threatened, and efficiency is low.
Disclosure of Invention
The object of the present invention is to solve at least one of the technical drawbacks.
Therefore, the invention aims to provide an intelligent water level monitoring method based on image recognition.
In order to achieve the above object, an embodiment of the present invention provides an intelligent water level monitoring method based on image recognition, including the steps of:
Step S1, receiving image data sent by camera equipment in a detected area, wherein the image data comprises a water gauge image;
s2, detecting and identifying the water gauge in the image data by adopting a deep neural network, and dividing the water gauge image above the water surface by adopting a graph cutting algorithm on the basis;
Step S3, calculating the height of the water gauge above the water surface according to the divided water gauge image, and combining the integral height of the water gauge to obtain the height of the water gauge below the water surface;
and S4, acquiring the elevation of the bottom of the water gauge of the actual station and the elevation of the water gauge below the water surface obtained in the step S3, and calculating the water level value.
Further, in the step S2, a water gauge in the image data is detected and identified by using a fast R-CNN algorithm.
Further, extracting features of the water gauge image from the received image data by using a feature extraction network of a Faster R-CNN algorithm to generate a feature map;
Processing on the feature map by using an RPN network as a candidate region network, and outputting rectangular target candidate regions with various scales and aspect ratios;
and inputting the feature map and the generated target candidate region into a classification regression network, and outputting the category of the generated water gauge candidate region and a water gauge boundary box according to the features in the candidate region.
Further, according to a preset region of interest, using the upper left corner coordinate and the lower right corner coordinate of the region of interest as input coordinates of a graph cutting algorithm, and adopting the graph cutting algorithm to divide a water gauge image above the water surface from the water gauge bounding box, wherein the region of interest is a region of a preset frame for defining a water gauge.
Further, the upper left corner coordinates of the region of interest are read from a site water gauge data table, and the lower right corner coordinates are consistent with the lower left corner coordinates of the water gauge bounding box.
Further, the feature extraction network of the fast R-CNN algorithm adopts one of the following: ZF network, VGG16 network, and AlexNet network.
Further, in the step S3, the step of calculating the height of the water gauge above the water surface from the segmented water gauge image includes the steps of: dividing the number of pixels of the water gauge above the water surface by the number of pixels in unit length according to the divided water gauge image above the water surface to obtain the height of the water gauge above the water surface, and subtracting the height of the water gauge above the water surface from the overall height of the water gauge to obtain the height of the water gauge below the water surface.
Further, the number of pixels in the unit length is calculated by a preset site water gauge data table.
According to the intelligent water level monitoring method based on image recognition, the intelligent water level monitoring system based on intelligent image recognition utilizes a related computer vision technology and a machine learning (including deep learning) algorithm, a developed intelligent water gauge reading technology based on images (videos) can receive images or videos containing water gauges and transmitted by imaging equipment installed at reservoirs, tunnels and the like, and the water gauges in the images are detected and recognized, the water gauge parts above the water level are calibrated, the height of the water gauge parts above the water level is calculated and the like through the combination of an image processing technology, the deep learning and a traditional method, so that water level data are obtained. The calculated water level data is automatically stored, displayed or returned to the request end. If the water level value exceeds the warning value, water level early warning can be carried out. According to the invention, the water level is not needed to be surveyed manually on site, the image acquisition is realized in a remote mode, the analysis processing is carried out locally, the water level height is calculated, the manual operation cost is greatly reduced, and the measurement accuracy is improved.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the invention will become apparent and may be better understood from the following description of embodiments taken in conjunction with the accompanying drawings in which:
FIG. 1 is a flow chart of intelligent water level monitoring based on image recognition according to an embodiment of the present invention;
FIG. 2 is a block diagram of a Faster R-CNN network model according to an embodiment of the invention;
FIG. 3 is a block diagram of a Faster R-CNN feature extraction network according to an embodiment of the invention;
FIG. 4 is a block diagram of a Faster R-CNN candidate area generating network and a classification regression network according to an embodiment of the invention;
FIG. 5 is a flowchart of Grabcut algorithm according to an embodiment of the present invention;
FIG. 6 is a hardware environment diagram of a smart image recognition based water level monitoring system deployment in accordance with an embodiment of the present invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.
As shown in fig. 1, the intelligent water level monitoring method based on image recognition according to the embodiment of the invention comprises the following steps:
Step S1, receiving image data sent by the image pickup equipment in the detected area, wherein the image data comprises a water gauge image.
Specifically, a plurality of imaging devices are installed around the area to be detected, images containing water gauges are shot by the imaging devices, and then the images are remotely transmitted to an upper computer for analysis and processing.
And S2, detecting and identifying the water gauge in the image data by the upper computer through a deep neural network, and dividing the water gauge image above the water surface through a graph cutting algorithm on the basis.
In this step, referring to fig. 2 to 4, the water gauge in the image data is detected and identified by using the fast R-CNN algorithm. The fast R-CNN is a classical deep network structure for target detection, and replaces the traditional SELECTIVESEARCH target extraction method with network training to realize, so that the detection, classification speed and detection accuracy of the whole process are greatly improved.
The Faster R-CNN algorithm is mainly divided into two steps:
(1) The location of the target is determined and then the category of the target is identified. In a specific implementation, an image is input first, and then a series of convolution and pooling operations are performed to extract features of the image and generate a feature map.
(2) And processing on the feature map by using a candidate region generation network to generate target candidate regions with different scales and aspect ratios.
(3) And inputting the feature map and the generated target candidate region into a classification regression network, and outputting the category and the target boundary box of the generated target candidate region according to the features in the target candidate region.
The Faster R-CNN network structure mainly comprises three sub-networks: the method comprises a feature extraction network, a candidate region generation network and a classification regression network.
Specifically, the water gauge image processing corresponding to the invention comprises the following steps:
firstly, extracting features of a water gauge image from received image data by utilizing a feature extraction network of a Faster R-CNN algorithm to generate a feature map.
Then, using the RPN network as a candidate region network, processing is performed on the feature map, outputting rectangular target candidate regions having various scales and aspect ratios. The RPN network is a feature map in which the input to the RPN network is the feature extraction network output, outputting rectangular target candidate regions with multiple dimensions and aspect ratios. The RPN network firstly carries out convolution operation through a 3X 3 convolution check feature diagram to form a feature vector, then uses two convolution kernels with the size of 1X 1 to simulate two full-connection layers to obtain the score and correction parameters of the candidate region, and finally normalizes the score through a Softmax layer to obtain the confidence that whether the candidate region contains a target to be tested.
Finally, after extracting the candidate region, a classification regression operation is required for the candidate region. The input of the classification regression network is a feature map output by the feature extraction network and a candidate region output by the candidate region extraction network, and the output is the confidence coefficient of the candidate region corresponding to each category and the correction parameter of the candidate region. And inputting the feature map and the generated target candidate region into a classification regression network, and outputting the category of the generated water gauge candidate region and a water gauge boundary box according to the features in the candidate region.
In one embodiment of the invention, the feature extraction network in the Faster R-CNN algorithm is a convolutional neural network, and the network structure can be used as the feature extraction network of the Faster R-CNN algorithm, and the extracted image features provide input for the subsequent network. The feature extraction network of the fast R-CNN algorithm adopts one of the following: ZF network, VGG16 network, and AlexNet network. It should be noted that the feature extraction network is not limited to the above example, and other types of networks may be used, and will not be described herein.
After the water gauge image is detected by using the fast R-CNN algorithm, the left upper corner coordinate and the right lower corner coordinate of the region of interest are used as input coordinates of a graph cutting algorithm according to a preset region of interest, and the graph cutting algorithm is adopted to cut the water gauge image above the water level from a water gauge boundary frame, wherein the region of interest is a region of the preset frame for defining the water gauge. The upper left corner coordinates of the region of interest are read from the site water gauge data table, and the lower right corner coordinates are consistent with the lower left corner coordinates of the water gauge bounding box.
As shown in fig. 5, the graph-cut technique is an image segmentation algorithm based on graph theory, an image is regarded as a graph, each pixel point in the image represents a node on the graph, the relationship between the nodes is regarded as an edge, and the similarity between the nodes is used for representing the weight value of the edge. In the process of each segmentation, deleting the connection with smaller weight, so that the pixel points with higher similarity are positioned in the same graph, and the pixel points with lower similarity are positioned in different graphs, thereby realizing continuous segmentation of the graphs and finally realizing segmentation of the whole image.
The following describes in detail a coordinate calibration method of a preset region of interest:
Due to camera shake and the like, small movements of the position of the water gauge in the image often occur. When the graph cutting method is adopted to calculate the height of the water gauge, the left upper corner coordinate and the right lower corner coordinate of the region of interest are required to be used. The lower right corner coordinates can be obtained by the target detection method without initializing, and the upper left corner coordinates need to be initialized. The position of the upper left corner of the region of interest is generally taken at the upper left part of the water gauge, so that when the camera shakes, the X coordinate and the Y coordinate of the upper left corner of the water gauge are smaller than or equal to the X coordinate and the Y coordinate of the upper left corner of the region of interest.
And precisely dividing the water gauge part above the water surface by adopting a graph cutting algorithm. The problem that the water gauge target frame is not accurate enough is found in the water gauge detected by the fast R-CNN, namely the detected boundary frame of the water gauge above the water surface is different from the actual boundary frame of the water gauge above the water surface in the image. Therefore, the method further adopts a graph cutting algorithm to accurately divide the water gauge part above the water surface according to the region of interest (the region of framing the water gauge) in the image. The upper left corner coordinate of the region of interest is read from a site water gauge data table, and the lower right corner coordinate is consistent with the lower left corner coordinate of a target frame detected when the water gauge is detected by using a fast R-CNN. For example, a graph cutting algorithm is applied to the water gauge frame to divide the water gauge part above the water surface, and the foreground part is the water gauge part above the water surface.
And S3, calculating the height of the water gauge above the water surface according to the segmented water gauge image, and combining the overall height of the water gauge to obtain the height of the water gauge below the water surface.
Specifically, the method for calculating the height of the water gauge above the water surface according to the divided water gauge image comprises the following steps: dividing the number of pixels of the water gauge above the water surface by the number of pixels in a unit length (the number of pixels of the water gauge above the water surface/the number of pixels in the unit length) according to the partitioned water gauge image above the water surface to obtain the height of the water gauge above the water surface, and subtracting the height of the water gauge above the water surface from the overall height of the water gauge to obtain the height of the water gauge below the water surface. The number of pixels in unit length is calculated by a preset site water gauge data table (table 1).
Table 1 site water gauge data sheet
Specifically, the method for calculating the number of pixels per unit length is as follows: the number of pixels in unit length in the station water gauge data table is calculated. The calculation process is as follows: and selecting a water gauge with a certain length on the image, and setting the length of the corresponding actual water gauge as x (calculated from the scale difference of the water gauge). If the number of pixels in the image corresponding to the length is y, the number of pixels in the unit length is equal to y/x.
And S4, acquiring the elevation of the bottom of the water gauge of the actual station and the elevation of the water gauge below the water surface obtained in the step S3, and calculating the water level value.
The intelligent water level monitoring method based on image recognition comprises the following technical parameters:
(1) Use environment
Hardware environment: windows 10 (Ubuntu 16.4.3), 8G memory (16G memory optimization), CPU (Intel i 7) or GPU
Software environment: tensorFlow, keras framework, openCV library, python language, mySQL database
(2) Software function
1) Initializing a database table, wherein the structure of the database table is shown in table 2;
Table 2 database table structure
NODE: number identifying test river
Position: the water level value of the water level at the last time the station was tested
PixelSise number of pixels per unit length, in pixels per centimeter
Tm: time when the station water level value is tested
X: x-coordinate of upper left corner of region of interest applied in graph cut algorithm
Y: the Y-coordinate database connection for the upper left corner of the region of interest applied in the graph cut algorithm:
host＝"127.0.0.1",user＝"root",passwd＝"123456",db＝"waterline"
2) Receiving a request image identification request (the request comprises interface parameters, and specifically refers to interface description), performing intelligent water level detection (including water gauge detection and water level value calculation) on the synthetic image or performing error reporting on the illegal image, and returning detection results and the like.
(3) Providing data requirements
Data: the provided image data comprises images of different water qualities, different application scenes, different illumination conditions and different weather conditions.
(4) Interface
Web site:
119.3.204.121:8888/delectstcd＝1&imageUrl＝http://www.slhzt.com/img/1/2.jpg&mId＝1&tm＝1233
server address: 119.3.204.121
Port number: 8888
Main program entry: delect A
Station number: stcd A
Picture address: imageUrl = http:// www.slhzt.com/img/1/2.Jpg
Id number of picture: mId
Time: tm (tm)
Data transfer http:// ip: port/XXXXXX/XX
The post parameter was used and the parameters are shown in Table 3.
Table 3 transfer parameter table
| Attribute type | Attribute name | Attribute value and meaning thereof |
| String | token | Token authentication |
| String | stcd | Station code |
| String | imageUrl | Picture address |
| String | mId | Picture id |
| Date | tm | Time of |
| String | req1 | Request parameter 1 (leave for supplement) |
| String | req2 | Request parameters 2 (leave for supplement) |
| String | req3 | Request parameters 3 (leave for supplement) |
| String | req4 | Request parameters 4 (leave for supplementation) |
The return parameters are returned in json format and table 4 shows the return parameters.
Table 4 return parameter table
(5) Deployment
The intelligent image recognition-based water level monitoring system is already deployed on a cloud server leased by a company, and the cloud server is configured as shown in fig. 6.
According to the intelligent water level monitoring method based on image recognition, the intelligent water level monitoring system based on intelligent image recognition utilizes a related computer vision technology and a machine learning (including deep learning) algorithm, a developed intelligent water gauge reading technology based on images (videos) can receive images or videos containing water gauges and transmitted by imaging equipment installed at reservoirs, tunnels and the like, and the water gauges in the images are detected and recognized, the water gauge parts above the water level are calibrated, the height of the water gauge parts above the water level is calculated and the like through the combination of an image processing technology, the deep learning and a traditional method, so that water level data are obtained. The calculated water level data is automatically stored, displayed or returned to the request end. If the water level value exceeds the warning value, water level early warning can be carried out. According to the invention, the water level is not needed to be surveyed manually on site, the image acquisition is realized in a remote mode, the analysis processing is carried out locally, the water level height is calculated, the manual operation cost is greatly reduced, and the measurement accuracy is improved.
The performance description of the intelligent water level monitoring method based on image recognition in the embodiment of the invention:
1) Can adapt to different water quality, application scenes, illumination, weather and other conditions. The method specifically comprises the following steps:
(1) Can adapt to different water quality conditions, such as turbidity, containing floaters and the like;
(2) Meanwhile, the method can adapt to different application scenes;
(3) The illumination condition needs to be based on night vision, and can be identified in the daytime and the night;
(4) Is suitable for complex weather conditions, such as heavy rainfall, fog, haze, cloudiness and the like.
2) The detection precision error is +5cm to-5 cm.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Although embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives, and variations may be made in the above embodiments by those skilled in the art without departing from the spirit and principles of the invention. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (1)
1. An intelligent water level monitoring method based on image recognition is characterized by comprising the following steps:
Step S1, receiving image data sent by camera equipment in a detected area, wherein the image data comprises a water gauge image;
s2, detecting and identifying the water gauge in the image data by adopting a deep neural network, and dividing the water gauge image above the water surface by adopting a graph cutting algorithm on the basis; detecting and identifying the water gauge in the image data by adopting a Faster R-CNN algorithm:
The characteristic extraction network of the Faster R-CNN algorithm extracts characteristics of the water gauge image from the received image data to generate a characteristic map;
Processing on the feature map by using an RPN network as a candidate region network, and outputting rectangular target candidate regions with various scales and aspect ratios; the RPN network is a full convolution network, the input of the RPN network is a feature map output by the feature extraction network, and rectangular target candidate areas with various scales and aspect ratios are output; the RPN network firstly carries out convolution operation through a 3X 3 convolution check feature diagram to form a feature vector, then uses two convolution kernels with the size of 1X 1 to simulate two full-connection layers to obtain the score and correction parameters of a candidate region, and finally normalizes the score through a Softmax layer to obtain the confidence that whether the candidate region contains a target to be tested;
After extracting the candidate region, performing a classification regression operation on the candidate region; the input of the classification regression network is a feature map output by the feature extraction network and a candidate region output by the candidate region extraction network, and the output is the confidence coefficient of the candidate region corresponding to each category and the correction parameter of the candidate region; inputting the feature map and the generated target candidate region into a classification regression network, and outputting the category of the generated water gauge candidate region and a water gauge boundary frame according to the features in the candidate region;
the characteristic extraction network in the fast R-CNN algorithm is a convolutional neural network, the network structure can be used as the characteristic extraction network of the fast R-CNN algorithm, and the extracted image characteristics provide input for the subsequent network; the feature extraction network of the fast R-CNN algorithm adopts one of the following: ZF, VGG16, and AlexNet networks;
After a water gauge image is detected by using a fast R-CNN algorithm, using the upper left corner coordinate and the lower right corner coordinate of the region of interest as input coordinates of a graph cutting algorithm according to a preset region of interest, and dividing the water gauge image above the water surface from a water gauge boundary frame by using the graph cutting algorithm, wherein the region of interest is a region of the preset frame for setting the water gauge; reading the left upper corner coordinate of the region of interest from a site water gauge data table, wherein the right lower corner coordinate is consistent with the left lower corner coordinate of the water gauge boundary frame;
Step S3, calculating the height of the water gauge above the water surface according to the divided water gauge image, and combining the overall height of the water gauge to obtain the height of the water gauge below the water surface; the water gauge height above the water surface is calculated according to the divided water gauge image, and the method comprises the following steps: dividing the number of pixels of the water gauge above the water surface by the number of pixels in a unit length according to the divided water gauge image above the water surface to obtain the height of the water gauge above the water surface, and subtracting the height of the water gauge above the water surface from the overall height of the water gauge to obtain the height of the water gauge below the water surface;
the pixel number in the unit length is calculated by a preset station water gauge data table;
and S4, acquiring the elevation of the bottom of the water gauge of the actual station and the elevation of the water gauge below the water surface obtained in the step S3, and calculating the water level value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910517818.3A CN110223341B (en) | 2019-06-14 | 2019-06-14 | Intelligent water level monitoring method based on image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910517818.3A CN110223341B (en) | 2019-06-14 | 2019-06-14 | Intelligent water level monitoring method based on image recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110223341A CN110223341A (en) | 2019-09-10 |
| CN110223341B true CN110223341B (en) | 2024-05-28 |
Family
ID=67817208
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910517818.3A Active CN110223341B (en) | 2019-06-14 | 2019-06-14 | Intelligent water level monitoring method based on image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110223341B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110763297A (en) * | 2019-10-09 | 2020-02-07 | 中水淮河规划设计研究有限公司 | Intelligent water gauge water level identification system based on video |
| CN111680606B (en) * | 2020-06-03 | 2021-11-23 | 淮河水利委员会水文局(信息中心) | Low-power-consumption water level remote measuring system based on artificial intelligence cloud identification water gauge |
| CN113819971A (en) * | 2020-07-07 | 2021-12-21 | 湖北亿立能科技股份有限公司 | Artificial intelligence water level monitoring system based on water, scale and floater semantic segmentation |
| CN111967419B (en) * | 2020-08-25 | 2023-06-20 | 深圳大学 | Dam bank dangerous case prediction method, dam bank dangerous case prediction device, computer equipment and storage medium |
| CN112163496B (en) * | 2020-09-22 | 2024-05-28 | 广东华南水电高新技术开发有限公司 | Embedded terminal reservoir water level early warning method based on semantic segmentation |
| CN112101282B (en) * | 2020-09-25 | 2024-04-26 | 北京瞰天科技有限公司 | Water target identification method and device, electronic equipment and storage medium |
| CN112508986B (en) * | 2020-12-04 | 2022-07-05 | 武汉大学 | Water level measurement method based on deep convolutional network and random field |
| CN112523804B (en) * | 2020-12-10 | 2022-06-14 | 广东电网有限责任公司佛山供电局 | Cable tunnel drainage monitoring system and method |
| CN112907506B (en) * | 2021-01-11 | 2023-07-07 | 昆明理工大学 | Water gauge color information-based variable-length water gauge water level detection method, device and storage medium |
| CN113223073B (en) * | 2021-04-16 | 2022-04-19 | 北京科技大学 | River channel water level assessment method |
| CN114593792A (en) * | 2022-03-29 | 2022-06-07 | 中国水利水电科学研究院 | Underground water level monitoring method and device and storage medium |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105760886A (en) * | 2016-02-23 | 2016-07-13 | 北京联合大学 | Image scene multi-object segmentation method based on target identification and saliency detection |
| CN106504233A (en) * | 2016-10-18 | 2017-03-15 | 国网山东省电力公司电力科学研究院 | Image electric power widget recognition methodss and system are patrolled and examined based on the unmanned plane of Faster R CNN |
| CN107367310A (en) * | 2017-07-11 | 2017-11-21 | 华南理工大学 | A kind of river level remote monitoring method based on computer vision |
| CN107563303A (en) * | 2017-08-09 | 2018-01-09 | 中国科学院大学 | A kind of robustness Ship Target Detection method based on deep learning |
| CN108073905A (en) * | 2017-12-21 | 2018-05-25 | 北京邮电大学 | A kind of method, system and the equipment of intelligence water gauge reading |
| CN108509919A (en) * | 2018-04-03 | 2018-09-07 | 哈尔滨哈船智控科技有限责任公司 | A kind of detection and recognition methods based on deep learning to waterline in video or picture |
| CN108596221A (en) * | 2018-04-10 | 2018-09-28 | 江河瑞通（北京）技术有限公司 | The image-recognizing method and equipment of rod reading |
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
| CN108985322A (en) * | 2018-06-01 | 2018-12-11 | 广东电网有限责任公司 | A kind of cable tunnel ponding positioning identifying method based on ZF-Faster RCNNs |
| CN109063586A (en) * | 2018-07-11 | 2018-12-21 | 东南大学 | A kind of Faster R-CNN driver's detection method based on candidate's optimization |
| CN109145830A (en) * | 2018-08-24 | 2019-01-04 | 浙江大学 | A kind of intelligence water gauge recognition methods |
| KR101970442B1 (en) * | 2018-12-04 | 2019-04-19 | 주식회사 넥스파시스템 | Illegal parking enforcement system Using Fast R-CNN based on Vehicle detection |
| KR20190063729A (en) * | 2017-11-30 | 2019-06-10 | 주식회사 넥토마이닝 | Life protection system for social disaster using convergence technology like camera, sensor network, and directional speaker system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10453200B2 (en) * | 2016-11-02 | 2019-10-22 | General Electric Company | Automated segmentation using deep learned priors |
-
2019
- 2019-06-14 CN CN201910517818.3A patent/CN110223341B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105760886A (en) * | 2016-02-23 | 2016-07-13 | 北京联合大学 | Image scene multi-object segmentation method based on target identification and saliency detection |
| CN106504233A (en) * | 2016-10-18 | 2017-03-15 | 国网山东省电力公司电力科学研究院 | Image electric power widget recognition methodss and system are patrolled and examined based on the unmanned plane of Faster R CNN |
| CN107367310A (en) * | 2017-07-11 | 2017-11-21 | 华南理工大学 | A kind of river level remote monitoring method based on computer vision |
| CN107563303A (en) * | 2017-08-09 | 2018-01-09 | 中国科学院大学 | A kind of robustness Ship Target Detection method based on deep learning |
| KR20190063729A (en) * | 2017-11-30 | 2019-06-10 | 주식회사 넥토마이닝 | Life protection system for social disaster using convergence technology like camera, sensor network, and directional speaker system |
| CN108073905A (en) * | 2017-12-21 | 2018-05-25 | 北京邮电大学 | A kind of method, system and the equipment of intelligence water gauge reading |
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
| CN108509919A (en) * | 2018-04-03 | 2018-09-07 | 哈尔滨哈船智控科技有限责任公司 | A kind of detection and recognition methods based on deep learning to waterline in video or picture |
| CN108596221A (en) * | 2018-04-10 | 2018-09-28 | 江河瑞通（北京）技术有限公司 | The image-recognizing method and equipment of rod reading |
| CN108985322A (en) * | 2018-06-01 | 2018-12-11 | 广东电网有限责任公司 | A kind of cable tunnel ponding positioning identifying method based on ZF-Faster RCNNs |
| CN109063586A (en) * | 2018-07-11 | 2018-12-21 | 东南大学 | A kind of Faster R-CNN driver's detection method based on candidate's optimization |
| CN109145830A (en) * | 2018-08-24 | 2019-01-04 | 浙江大学 | A kind of intelligence water gauge recognition methods |
| KR101970442B1 (en) * | 2018-12-04 | 2019-04-19 | 주식회사 넥스파시스템 | Illegal parking enforcement system Using Fast R-CNN based on Vehicle detection |
Non-Patent Citations (2)
| Title |
|---|
| 基于快速区域建议网络的图像多目标分割算法;黄劲潮;;山东大学学报(工学版);20180525(第04期);全文 * |
| 基于自适应叠合分割与深度神经网络的人数统计方法;郭文生;包灵;钱智成;曹万里;;计算机科学(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110223341A (en) | 2019-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110223341B (en) | Intelligent water level monitoring method based on image recognition | |
| Hashemi-Beni et al. | Flood extent mapping: An integrated method using deep learning and region growing using UAV optical data | |
| CN106127204B (en) | A kind of multi-direction meter reading Region detection algorithms of full convolutional neural networks | |
| CN106228125B (en) | Method for detecting lane lines based on integrated study cascade classifier | |
| Jafari et al. | Real-time water level monitoring using live cameras and computer vision techniques | |
| CN107423760A (en) | Based on pre-segmentation and the deep learning object detection method returned | |
| CN109635875A (en) | A kind of end-to-end network interface detection method based on deep learning | |
| Kalke et al. | Support vector machine learning applied to digital images of river ice conditions | |
| CN111598098B (en) | Water gauge water line detection and effectiveness identification method based on full convolution neural network | |
| CN109492665A (en) | Detection method, device and the electronic equipment of growth period duration of rice | |
| CN109186706A (en) | A method of for the early warning of Urban Storm Flood flooding area | |
| CN111507375A (en) | Urban waterlogging risk rapid assessment method and system | |
| CN108492298A (en) | Based on the multispectral image change detecting method for generating confrontation network | |
| CN116258608B (en) | Water conservancy real-time monitoring information management system integrating GIS and BIM three-dimensional technology | |
| Xiao et al. | Building segmentation and modeling from airborne LiDAR data | |
| CN109242019A (en) | A kind of water surface optics Small object quickly detects and tracking | |
| CN113469097A (en) | SSD (solid State disk) network-based real-time detection method for water surface floating object multiple cameras | |
| CN110636248A (en) | Target tracking method and device | |
| KR102602439B1 (en) | Method for detecting rip current using CCTV image based on artificial intelligence and apparatus thereof | |
| CN107886049B (en) | Visibility recognition early warning method based on camera probe | |
| Tang et al. | A new method for extracting trees and buildings from sparse LiDAR data in urban areas | |
| CN115497036A (en) | Waterlogging depth calculation method and system and readable storage medium | |
| Mali et al. | Experimental and numerical study of flood in a river-network-floodplain set-up | |
| CN109523509A (en) | Detection method, device and the electronic equipment of wheat heading stage | |
| CN113870224A (en) | Flood monitoring method, system, equipment and medium |
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 |