CN117523573A - Water meter reading method and system based on multi-stage visual detection fusion fault early warning - Google Patents
Water meter reading method and system based on multi-stage visual detection fusion fault early warning Download PDFInfo
- Publication number
- CN117523573A CN117523573A CN202311637194.1A CN202311637194A CN117523573A CN 117523573 A CN117523573 A CN 117523573A CN 202311637194 A CN202311637194 A CN 202311637194A CN 117523573 A CN117523573 A CN 117523573A
- Authority
- CN
- China
- Prior art keywords
- water meter
- image
- character
- reading
- counter
- 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.)
- Pending
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 111
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000001514 detection method Methods 0.000 title claims abstract description 57
- 230000004927 fusion Effects 0.000 title claims abstract description 22
- 230000000007 visual effect Effects 0.000 title claims abstract description 22
- 230000005540 biological transmission Effects 0.000 claims description 10
- 238000003702 image correction Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000011179 visual inspection Methods 0.000 claims 1
- 238000005457 optimization Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000012937 correction Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/1444—Selective acquisition, locating or processing of specific regions, e.g. highlighted text, fiducial marks or predetermined fields
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1463—Orientation detection or correction, e.g. rotation of multiples of 90 degrees
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/147—Determination of region of interest
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1475—Inclination or skew detection or correction of characters or of image to be recognised
- G06V30/1478—Inclination or skew detection or correction of characters or of image to be recognised of characters or characters lines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/16—Image preprocessing
- G06V30/1607—Correcting image deformation, e.g. trapezoidal deformation caused by perspective
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/18—Extraction of features or characteristics of the image
- G06V30/1801—Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes or intersections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
- G06V30/19147—Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
- G06V30/1916—Validation; Performance evaluation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
- G06V30/19173—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- 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
- Y02A20/00—Water conservation; Efficient water supply; Efficient water use
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
The invention discloses a water meter reading method and system based on multi-stage visual detection fusion fault early warning, which comprises the following steps of S1: collecting a water meter image, wherein the water meter image is a character wheel type water meter image; s2: detecting a counter frame of the water meter image to obtain a counter frame image; s3: performing corner detection on the counter block image to obtain the corner position of the counter block image; s4: correcting the counter block image by using the angular point position; s5: and carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the water meter reading. The invention can improve the working efficiency of the water meter reading and reduce the labor cost; the rotary and distortion dial plate can be corrected to provide help for subsequent character recognition, so that more stable reading is realized, and the unreadable water meter can be screened out to avoid entering a character detection stage, thereby saving recognition time; the dynamic optimization strategy is adopted to improve the problem of half character recognition which is easy to occur in character recognition of the character wheel type water meter.
Description
Technical Field
The invention relates to the technical field of water meter reading, in particular to a water meter reading method and system based on multi-stage visual detection fusion fault early warning.
Background
The reading identification of the water meter is an important link in the urban water management process, along with the continuous expansion of the urban scale, the information quantity of urban management is also larger and larger, the intelligent degree of urban information management is required to be continuously improved at the moment, and the management and the utilization of water resources are not separated no matter the urban information management is carried out or the intelligent life of residents is carried out. Intelligent management of water resources is not separated from intelligent management of water meters. The water meter has an indispensable function in the aspect of water resource management; and plays an important role in the whole digital city development process. The water meter has wide application in the fields of industry, life and the like. The conventional water meter reading often depends on manual reading, and the manual meter reading mode has the defects of large workload, low efficiency, long time consumption and high labor cost, and workers are prone to making mistakes when processing a large amount of data.
Disclosure of Invention
(one) solving the technical problems
Aiming at the defects of the prior art, the invention provides a water meter reading method and system based on multi-stage visual detection fusion fault early warning, which can solve the technical problems.
(II) technical scheme
In order to solve the technical problems, the invention provides the following technical scheme: a water meter reading method based on multi-stage visual detection fusion fault early warning comprises the following steps:
s1: collecting a water meter image, wherein the water meter image is a character wheel type water meter image;
s2: detecting a counter frame of the water meter image to obtain a counter frame image;
s3: performing corner detection on the counter block image to obtain the corner position of the counter block image;
s4: correcting the counter block image by using the angular point position;
s5: and carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the water meter reading.
Preferably, before step S2, the method further comprises: and preprocessing the water meter image.
Preferably, in step S2, after the counter frame image is obtained, the counter frame image is further edge-filled.
Preferably, step S3 further includes: and classifying the block images of the counter in readable/unreadable mode so as to realize fault early warning of the water meter.
Preferably, the counter frame image comprises four corner points: upper left corner, lower left corner, upper right corner, and lower right corner.
Preferably, in step S5, when two characters appear in the same character wheel frame of the counter frame image, the areas of the two characters are calculated respectively, and when the area ratio of one character exceeds a preset threshold value, the area ratio of the one character is the character reading of the current character wheel frame.
Preferably, in step S5, when the area ratio of the two characters does not exceed the preset threshold, the adjustment is further performed according to the character reading of the character wheel frame of the last measuring range: if the character reading of the character wheel frame of the last measuring range is 9, the character reading of the current character wheel frame is the upper character of the two characters; if the character reading of the character wheel frame of the last measuring range is 0, the character reading of the current character wheel frame is the lower character of the two characters.
Preferably, after step S5, the method further comprises: and outputting the water meter reading and the counter frame image.
In order to solve the technical problems, the invention provides another technical scheme as follows: a water meter reading system based on multi-stage visual detection fusion fault pre-warning, comprising:
and an image acquisition module: the method is used for collecting water meter images, wherein the water meter images are character wheel type water meter images;
a counter frame detection module: the counter frame detection module is used for detecting the water meter image to obtain a counter frame image;
the corner detection module: the method comprises the steps of carrying out corner detection on a counter frame image to obtain the corner position of the counter frame image;
an image correction module: the method is used for correcting the counter block image by utilizing the angular point position;
and a character recognition module: the method is used for carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the water meter reading.
Preferably, the image acquisition module specifically comprises a controller, a remote transmission module and a camera shooting module, and the controller is respectively connected with the remote transmission module and the camera shooting module.
(III) beneficial effects
Compared with the prior art, the invention provides a water meter reading method and system based on multi-stage visual detection fusion fault early warning, which have the following beneficial effects: (1) According to the invention, the target detection network is utilized to perform character recognition on the counter block image, so that automatic reading of the water meter is realized, and compared with the traditional manual meter reading mode, the automatic meter reading method has the advantages that the working efficiency is improved, the labor cost is reduced, and errors in the reading of the water meter caused by human factors are better avoided; (2) The counter block image is obtained by detecting the water meter image through the counter block, namely, the acquired water meter image is subjected to feature extraction, so that the method can be better adapted to the changes of different shooting scenes, and meanwhile, the character recognition efficiency of the water meter image can be improved; (3) By correcting the counter block image, help can be provided for subsequent character recognition, and more stable reading is realized.
Drawings
FIG. 1 is a flow chart of the steps of a method for reading a water meter based on multi-stage visual detection fusion fault early warning of the invention;
FIG. 2 is a schematic block diagram of a water meter reading system based on multi-stage visual detection fusion fault early warning of the present invention;
FIG. 3 is a network structure diagram of the improved YOL0v7 model of the present invention;
FIG. 4 is a block diagram of a multi-tasking network for corner detection and fault pre-warning according to the present invention;
FIG. 5 is a schematic diagram of the present invention for transmission correction of predicted corner points;
fig. 6 is a schematic diagram of a counter frame image of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention provides a water meter reading method based on multi-stage visual detection fusion fault early warning, which comprises the following steps:
s1: the method comprises the steps of collecting water meter images, wherein the water meter images are character wheel type water meter images, one water meter image corresponds to one character wheel type water meter, a counter used for representing water meter reading is arranged on the water meter images, and one counter comprises a plurality of character wheel frames with different measuring ranges.
S2: the image of the meter is detected in a counter frame to obtain a counter frame image, and it is understood that the counter frame image includes a counter for indicating the reading of the meter.
Preferably, before step S2, the method further comprises: the water meter image is preprocessed, the screening image is good in quality, the counter is clearly visible, and different angles, illumination conditions and environment backgrounds are covered as much as possible. Furthermore, an image marking tool can be used for marking the acquired water meter image, and the position of each counter is marked by a frame, so that the frame is ensured to closely surround the counter. And then, carrying out data enhancement on the marked image so as to expand a data set, and adopting operations such as rotation, translation, scaling, overturning and the like to increase the diversity and the robustness of the data. And simultaneously, the operation of blurring, brightness adjustment and the like is used for simulating the change under different environmental conditions. Finally, the data set can be divided into a training set, a verification set and a test set according to the ratio of 7:2:1. The data set can be further used for training the counter frame detection model, so that the counter frame detection model obtained through training is used for detecting the counter frame of the water meter image.
Preferably, the counter frame detection model employed by the present invention is a modified YOLOv7 model, as shown in fig. 3, which uses E-ELAN (extended high-efficiency layer aggregation network), connection-based model scaling, and re-parameterization to achieve an advantageous balance between detection efficiency and accuracy. The improved YOLOv7 model described above employs a Mosaic and mixed data enhancement technique in the preprocessing portion of the input and uniformly scales the image to 640x640 size input backbone network. The backbone network consists of three main components of CBS, E-ELAN and MP1, and introduces a CBAM attention mechanism in the backbone network and the head network to capture the correlation of space and channels, improve the network feature extraction capability and avoid the interference of other background features; the regression prediction adopts IoU Loss, the classification prediction adopts BCE Loss to train the network, and a corresponding weight file is generated, wherein IoU is the intersection ratio of a prediction frame and a real frame, and the Loss function of BCE Loss (Binary Cross-entry Loss) is shown as follows:
BCE Loss＝-(y_true*log(y_pred)+(1-y_true)*log(1-y_Pred))
in the above equation, y_true represents the true label, y_pred represents the output of the model prediction, log represents a natural logarithmic function, and x represents element-by-element multiplication between elements.
It should be understood that after the water meter image goes through the step S2, the accurate position coordinates of the counter in the picture of the detected water meter can be identified, so as to cut out the image of the counter in the water meter image: a counter frame image. Preferably, in step S2, after obtaining the counter frame image, edge filling is further performed on the counter frame image: the frame of the counter frame image is enlarged by 10% in a zero filling mode, so that the subsequent corner detection is facilitated.
S3: and performing corner detection on the counter frame image to obtain the corner position of the counter frame image.
Specifically, the counter frame image includes four corner points: upper left corner, lower left corner, upper right corner, and lower right corner.
In addition, step S3 further includes: and classifying the block images of the counter in readable/unreadable mode so as to realize fault early warning of the water meter.
Specifically, in step S3, the counter frame image may be analyzed and 9 outputs predicted through a multi-tasking network: eight floating point numbers to represent four corner locations, and an array containing two floating point numbers to represent the probability that the counter frame is readable/unreadable, the multitasking network structure is shown in fig. 4, in fig. 4 conv represents convolution and max represents max pooling.
S4: and correcting the counter block image by using the angular point positions.
After the angular point position is predicted in the step S3, the counter frame image can be corrected by calculating coordinates and transmission transformation: firstly, after the coordinates of the corner points are determined, the four corner points before correction of the counter block image are called p 1 …p 4 (i.e., the corner positions obtained in the above substep S32), the four corners after correction of the counter frame image are called p' 1 …p′ 4 Each corner has corresponding x and y coordinates, i.e. corner positions, as shown in fig. 5, the predicted corner is mapped to the correctionThe latter coordinate space will assume that the corrected image has a fixed size, p' 1 And p' 2 The angular point positions respectively positioned at the left upper corner and the right upper corner of the corrected image can calculate p 'in advance' 1 And p' 2 Is the value of (1): for example, for an image of size 320x320, p 'is assumed to be the same as the corrected image' 1 And p' 2 The positions of (10, 10) and (310, 10) are reserved with a certain number of pixels so as to avoid that the information loss of the image edge during transformation affects the correction result; further, considering the ratio of width to height, p 'can be determined by the following formula' 3 Coordinates of (c):
x′ 3 ＝x′ 1
in the above, [ p ] m ，p n ] dist Is p m ，p n The Euclidean distance between them, wrct, represents the width of the corrected counter frame image, i.e. p' 1 And p' 2 Distance in the x-axis. Finally through x' 4 ＝x′ 2 And y' 4 ＝y′ 3 To determine p' 4 。
Furthermore, the counter block images are classified as readable/unreadable: an array consisting of two floating points can be predicted specifically through a softmax activation function, the relative probability of a readable or unreadable water meter is represented, and the sum of the two floating points is 1; by setting the threshold value for the probability, the fault early warning of the water meter is realized, for example, if the unreadable probability exceeds the preset threshold value, the counter frame image is correspondingly unreadable, so that the unreadable counter frame image is screened out, the entering of the subsequent character recognition stage is avoided, and the recognition time is saved. Specifically, in addition to the coordinates of the angular points, a readable and unreadable classification label is added in the data set image sample to distinguish a normal water meter image from an unreadable water meter image such as a blocked water meter image, and the multi-task network learns the characteristics and modes of different counter frame samples in the training process and can give out readable and unreadable probability values of each counter frame sample; in the multitasking network of fig. 4, there are two unshared full-connect layers for each output, in the second unshared full-connect layer, there are two units for predicting the readability of the counter frame, and normalize the output of the predicted result using the softmax activation function, converting it to a value between (0, 1) as a readable/unreadable probability distribution, and the sum of the two values is 1, ensuring that the sum of the probabilities is 1. For each counter frame image sample, if the readability unit output probability is high, it indicates that the multitasking network considers the counter frame as readable, otherwise, it indicates that the counter frame is unreadable, thereby implementing the classification task that the counter frame is readable/unreadable.
S5: and carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the water meter reading.
Preferably, step S5 may specifically use the improved YOLOv7 model similar to step S2 to perform character recognition, where the detected object is a character wheel type number character from 0 to 9.
In addition, in step S5, when two characters appear in the same character wheel frame of the counter frame image, that is, there is one character (two incomplete characters, that is, half character problem common to the character wheel type characters) respectively on the upper and lower sides in the same character wheel frame, the dynamic optimization strategy is adopted to adjust the reading of the recognition characters: at this time, the areas of the two characters (specifically, the areas of the respective boundary boxes of the two characters) are respectively calculated, and when the area ratio of one of the characters in the character wheel frame exceeds a preset threshold value, the area ratio is the character reading of the current character wheel frame: for example, as shown in fig. 6, two characters appear in the 5 th character wheel frame from left to right: the upper character 9 and the lower character 0, and at this time, assuming that the area ratio of the lower character 0 in the 5 th character wheel frame exceeds the preset threshold value of 70%, the character reading of the current character wheel frame (the 5 th character wheel frame) is corresponding to 0.
In addition, in step S5, when the area ratio of the two characters in the character wheel frame does not exceed the preset threshold (for example, the area ratio of the upper character in the character wheel frame is 40%, the area ratio of the lower character in the character wheel frame is 60%, and the preset threshold is 70%), the adjustment is further performed according to the character reading of the character wheel frame of the last range: if the character reading of the character wheel frame of the last measuring range is 9, the character reading of the current character wheel frame is the upper character of the two characters; if the character reading of the character wheel frame of the last measuring range is 0, the character reading of the current character wheel frame is the lower character of the two characters. For example, as shown in fig. 6, two characters appear in the 4 th character wheel frame from left to right: the character reading of the character wheel frame of the last measuring range (namely the 5 th character wheel frame) is 0, and the character reading of the current character wheel frame (the 4 th character wheel frame) is the lower character of the two characters: 6.
preferably, after step S5, the method further comprises: and outputting the water meter reading and the counter frame image.
The invention also provides a water meter reading system based on the multi-stage visual detection fusion fault early warning, which comprises the following steps:
the image acquisition module 11: the method is used for collecting the water meter image, wherein the water meter image is a character wheel type water meter image. Preferably, the image acquisition module 11 specifically includes a controller, a remote transmission module and a camera module, the controller is connected with the remote transmission module and the camera module respectively, it should be understood that the image acquisition module 11 is installed at the water meter end, and the controller controls the remote transmission module to upload the water meter image acquired by the camera module to the counter frame detection module; in order to balance the power consumption, the cost and the image quality, a low-resolution camera module can be used, an image compression algorithm is adopted, and a water meter image is acquired for compression and uploading.
Counter frame detection module 12: the counter frame detection module is used for detecting the water meter image to obtain a counter frame image;
corner detection module 13: the method comprises the steps of carrying out corner detection on a counter frame image to obtain the corner position of the counter frame image;
image correction module 14: the method is used for correcting the counter block image by utilizing the angular point position;
character recognition module 15: the method is used for carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the water meter reading.
Compared with the prior art, the invention provides a water meter reading method and system based on multi-stage visual detection fusion fault early warning, which have the following beneficial effects: (1) According to the invention, the target detection network is utilized to perform character recognition on the counter block image, so that automatic reading of the water meter is realized, and compared with the traditional manual meter reading mode, the automatic meter reading method has the advantages that the working efficiency is improved, the labor cost is reduced, and errors in the reading of the water meter caused by human factors are better avoided; (2) The counter block image is obtained by detecting the water meter image through the counter block, namely, the acquired water meter image is subjected to feature extraction, so that the method can be better adapted to the changes of different shooting scenes, and meanwhile, the character recognition efficiency of the water meter image can be improved; (3) The corner detection and fault early warning tasks are completed through the multi-task network, the rotating and distortion dial plates can be corrected, the subsequent character recognition can be assisted, more stable reading is realized, and the unreadable water meter can be screened out to avoid entering a character detection stage, so that the recognition time is saved; (4) Compared with the method for photographing and identifying the water meter reading by using a mobile phone and a meter reading machine, the method can finish timing photographing and uploading by using a simple controller, a camera module and a remote transmission module, does not need water interruption in installation, has low installation cost, and is more suitable for popularization and use in urban areas; (5) The invention can be deployed in a cloud server for use, more calculation force can be applied compared with the deployment in a surface end controller, and the power consumption of the surface end controller can be lower and more durable; (6) The invention adopts dynamic optimization strategy to improve the problem of character recognition of character wheel type water meter, thereby ensuring the accuracy of water meter reading.
It should be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (10)
1. A water meter reading method based on multi-stage visual detection fusion fault early warning is characterized by comprising the following steps:
s1: collecting a water meter image, wherein the water meter image is a character wheel type water meter image;
s2: detecting the water meter image by a counter frame to obtain a counter frame image;
s3: performing corner detection on the counter frame image to obtain a corner position of the counter frame image;
s4: correcting the counter block image by utilizing the angular point position;
s5: and carrying out character recognition on the counter block image by utilizing the target detection network so as to obtain the reading of the water meter.
2. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: before the step S2, the method further comprises: and preprocessing the water meter image.
3. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: in the step S2, after the counter frame image is obtained, edge filling is further performed on the counter frame image.
4. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: the step S3 further includes: and classifying the counter block images in readable/unreadable mode so as to realize fault early warning of the water meter.
5. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: the counter block image comprises four corner points: upper left corner, lower left corner, upper right corner, and lower right corner.
6. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: in the step S5, when two characters appear in the same character wheel frame of the counter frame image, the areas of the two characters are respectively calculated, and when the area ratio of one character exceeds a preset threshold value, the area ratio of the one character is the character reading of the current character wheel frame.
7. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 6, wherein the method comprises the following steps of: in the step S5, when the area ratio of the two characters does not exceed the preset threshold, the adjustment is further performed according to the character reading of the character wheel frame of the last measuring range: if the character reading of the character wheel frame of the last measuring range is 9, the character reading of the current character wheel frame is the upper character of the two characters; and if the character reading of the character wheel frame of the last measuring range is 0, the character reading of the current character wheel frame is the lower character of the two characters.
8. The method for reading the water meter based on the multi-stage visual detection fusion fault pre-warning according to claim 1, wherein the method comprises the following steps of: after the step S5, the method further comprises: and outputting the water meter reading and the counter frame image.
9. Water meter reading system based on multistage visual detection fuses trouble early warning, characterized by comprising:
and an image acquisition module: the method is used for collecting water meter images, wherein the water meter images are character wheel type water meter images;
a counter frame detection module: the counter frame detection module is used for detecting the water meter image to obtain a counter frame image;
the corner detection module: the method comprises the steps of carrying out corner detection on a counter frame image to obtain the corner position of the counter frame image;
an image correction module: the counter block image correction device is used for correcting the counter block image by utilizing the angular point position;
and a character recognition module: and the counter frame image is used for carrying out character recognition on the counter frame image by utilizing the target detection network so as to obtain the water meter reading.
10. The multi-stage visual inspection fusion fault pre-warning based water meter reading system of claim 9, wherein: the image acquisition module specifically comprises a controller, a remote transmission module and a camera shooting module, wherein the controller is respectively connected with the remote transmission module and the camera shooting module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311637194.1A CN117523573A (en) | 2023-12-01 | 2023-12-01 | Water meter reading method and system based on multi-stage visual detection fusion fault early warning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311637194.1A CN117523573A (en) | 2023-12-01 | 2023-12-01 | Water meter reading method and system based on multi-stage visual detection fusion fault early warning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117523573A true CN117523573A (en) | 2024-02-06 |
Family
ID=89760713
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311637194.1A Pending CN117523573A (en) | 2023-12-01 | 2023-12-01 | Water meter reading method and system based on multi-stage visual detection fusion fault early warning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117523573A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117576677A (en) * | 2023-12-01 | 2024-02-20 | 广州柏诚智能科技有限公司 | Water meter reading identification method |
-
2023
- 2023-12-01 CN CN202311637194.1A patent/CN117523573A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117576677A (en) * | 2023-12-01 | 2024-02-20 | 广州柏诚智能科技有限公司 | Water meter reading identification method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111950330B (en) | Pointer instrument indication detection method based on target detection | |
| CN109460753B (en) | Method for detecting floating object on water | |
| CN109145830B (en) | Intelligent water gauge identification method | |
| CN101751572A (en) | Pattern detection method, device, equipment and system | |
| CN117523573A (en) | Water meter reading method and system based on multi-stage visual detection fusion fault early warning | |
| CN111597941B (en) | Target detection method for dam defect image | |
| CN111914767B (en) | Scattered sewage enterprise detection method and system based on multi-source remote sensing data | |
| CN110852164A (en) | YOLOv 3-based method and system for automatically detecting illegal building | |
| CN112802016A (en) | Real-time cloth defect detection method and system based on deep learning | |
| CN115719338A (en) | PCB (printed circuit board) surface defect detection method based on improved YOLOv5 | |
| CN111598098A (en) | Water gauge water line detection and effectiveness identification method based on full convolution neural network | |
| CN116863274A (en) | Semi-supervised learning-based steel plate surface defect detection method and system | |
| CN113033315A (en) | Rare earth mining high-resolution image identification and positioning method | |
| CN114022774A (en) | Radar image-based marine mesoscale vortex monitoring method and device | |
| CN114373009A (en) | Building shadow height measurement intelligent calculation method based on high-resolution remote sensing image | |
| CN115775236A (en) | Surface tiny defect visual detection method and system based on multi-scale feature fusion | |
| CN114627461A (en) | Method and system for high-precision identification of water gauge data based on artificial intelligence | |
| CN112906689B (en) | Image detection method based on defect detection and segmentation depth convolutional neural network | |
| CN113628180A (en) | Semantic segmentation network-based remote sensing building detection method and system | |
| CN113284185A (en) | Rotating target detection method for remote sensing target detection | |
| CN117496426A (en) | Precast beam procedure identification method and device based on mutual learning | |
| CN112381794A (en) | Printing defect detection method based on deep convolution generation network | |
| CN111767826A (en) | Timing fixed-point scene abnormity detection method | |
| CN117576677A (en) | Water meter reading identification method | |
| CN113298755B (en) | Method and device for rapidly detecting ecological environment change patch based on time sequence image |
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 |