CN117994769A - Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge - Google Patents
Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge Download PDFInfo
- Publication number
- CN117994769A CN117994769A CN202410032515.3A CN202410032515A CN117994769A CN 117994769 A CN117994769 A CN 117994769A CN 202410032515 A CN202410032515 A CN 202410032515A CN 117994769 A CN117994769 A CN 117994769A
- Authority
- CN
- China
- Prior art keywords
- instrument
- image recognition
- image
- model
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000013523 data management Methods 0.000 claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims description 16
- 239000003086 colorant Substances 0.000 claims description 14
- 230000000977 initiatory effect Effects 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 10
- 238000002372 labelling Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 5
- 238000004519 manufacturing process Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 238000007726 management method Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000010191 image analysis Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007728 cost analysis Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013400 design of experiment Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- 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
- G06V30/162—Quantising the image signal
-
- 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
- G06V30/18105—Extraction of features or characteristics of the image related to colour
-
- 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
- G06V30/182—Extraction of features or characteristics of the image by coding the contour of the pattern
-
- 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
- G06V30/26—Techniques for post-processing, e.g. correcting the recognition result
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
The invention discloses an intelligent image recognition method and system for a lightning arrester counter and a gas pressure gauge, which relate to the technical field of image recognition and comprise the steps that a data management platform initiates a meter data acquisition request; collecting an instrument image by adopting a movable terminal; preprocessing the instrument image to enhance the identifiable characteristics of the image; building an instrument model identification model, and identifying the picture through the instrument model identification model to obtain an instrument model; according to the obtained instrument model, a corresponding instrument structure model is called, instrument readings in the image are identified through the structure model by using an image identification algorithm, and instrument data are obtained; and uploading the meter data to a data management platform and correcting. The invention adopts the color brightness cross entropy loss as a standard, so that the picture can accurately identify the color under different brightness. The invention also coordinates the instrument model by grids, and can accurately identify the data in the instrument display area, thereby realizing efficient and accurate instrument data acquisition.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to an intelligent image recognition method and system for a lightning arrester counter and a gas pressure gauge.
Background
The operation times and alternating current leakage current data of a transformer substation arrester discharge counter and SF6 equipment gas pressure meter data are recorded on site manually, and are recorded in a power grid management platform manually. The gas pressure data of all transformer substation lightning arresters and SF6 equipment are manually recorded in each month and are manually recorded into a power grid management platform, so that the problems of large workload, low efficiency and recording errors exist. The invention aims to realize efficient and accurate instrument data acquisition. The method provides powerful support for production management of enterprises, is beneficial to improving production efficiency and reducing cost, and creates greater value for the enterprises.
Disclosure of Invention
The present invention has been made in view of the above-described problems.
Accordingly, the present invention aims to solve the problems: the traditional method has the problems of low efficiency, recording errors and recording errors.
In order to solve the technical problems, the invention provides the following technical scheme: a lightning arrester counter and gas pressure gauge intelligent image recognition method comprises the steps that a data management platform initiates an instrument data acquisition request; collecting an instrument image by adopting a movable terminal; preprocessing the instrument image to enhance the identifiable characteristics of the image; building an instrument model identification model, and identifying the picture through the instrument model identification model to obtain an instrument model; according to the obtained instrument model, a corresponding instrument structure model is called, instrument readings in the image are identified through the structure model by using an image identification algorithm, and instrument data are obtained; and uploading the meter data to a data management platform and correcting.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: the data acquisition request comprises a data acquisition range and a data acquisition time period; the mobile terminal comprises a smart phone, a tablet computer or special data acquisition equipment.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: the preprocessing of the instrument image comprises gray level conversion of the image; reducing image noise by using Gaussian blur; and performing binarization processing on the image.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: selecting any kind of image recognition model, carrying out feature labeling on information reflecting instrument type colors in a picture, inputting the information into the image recognition model to be marked as a first image recognition model, carrying out feature labeling on brightness values corresponding to various types of colors in the picture subjected to gray scale processing, inputting the brightness values into the image recognition model to be marked as a second image recognition model, carrying out feature labeling on information reflecting instrument type structural shapes in the picture subjected to binarization processing, inputting the information into the image recognition model to be marked as a third image recognition model, and finally inputting an original image into the image recognition model to be marked as a fourth image recognition model; acquiring correct color, brightness and structure information which should be output in each picture, calculating the probability of outputting the correct color in a first image recognition model, calculating the probability of outputting the correct brightness in a second image recognition model, calculating the probability of outputting the correct structure in a third image recognition model, and calculating the probability of outputting the correct color and the correct structure in a fourth image recognition model; according to the obtained probability, calculating the color brightness cross entropy loss of the correct color probability of the first image recognition model and the correct brightness probability of the second image recognition model, adjusting the super-parameters of the first image recognition model according to the color brightness cross entropy loss until the color brightness cross entropy loss meets the preset requirement, ending the adjustment, obtaining a secondary first image recognition model, and calculating the probability of the secondary first image recognition model outputting correct colors; calculating the color cross entropy loss of the probability of outputting correct colors of the secondary first image recognition model and the probability of outputting correct colors of the fourth image recognition model, and adjusting the super parameters of the fourth image recognition model according to the color cross entropy loss until the color cross entropy loss meets the preset requirement; calculating the structure cross entropy loss of the correct structure probability of the third image recognition model and the correct structure probability of the fourth image recognition model, and adjusting the super-parameters of the fourth image recognition model according to the structure cross entropy loss until the structure cross entropy loss meets the preset requirement; and outputting a fourth image recognition model after the adjustment is finished, namely, obtaining the instrument model recognition model.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: the instrument reading in the identification image comprises the steps of identifying the specific position of the reading according to the structural model; acquiring readings in specific positions and matching the readings with the type of the instrument to obtain a numerical unit; the extracted readings are calibrated to correct for the identification errors.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: the specific position of the identification reading comprises gridding an extracted instrument structure model, determining the size of each grid according to the actual working condition, determining the intersection point between each grid as a grid point, giving coordinate values, enabling the coordinate of the centremost grid point to be (0, 0), enabling the length size of each grid to be 1, sequentially obtaining the coordinate of each grid point, splitting the structural characteristics of the instrument into a reading display frame characteristic and an instrument shell characteristic, and marking labels respectively; loading the reading display frame characteristics and the instrument shell characteristics into the meshed instrument structure model to obtain grid point coordinates with corresponding labels respectively, collecting grid point coordinates with two labels simultaneously, invalidating the grid point coordinates, and finally outputting a structural grid coordinate graph; binarizing the image by taking the structural characteristic of the instrument as a threshold condition, marking the pixel points larger than the threshold as a target object, marking the pixel points smaller than the threshold as a background, and finally outputting the image of the target object; embedding the structural grid coordinate graph into a target object image, acquiring pixel information of grid point coordinates with labels being reading display frame characteristics, and extracting the pixel information to obtain primary reading; the calibration comprises the steps of combining the primary reading with the matched numerical units to obtain a middle reading, checking the middle reading according to the common meter reading number of the database, if the deviation number is smaller than or equal to one, the middle reading is the meter data and is output, if the deviation number is larger than one, the invalid grid points in the structural grid coordinate graph are attached with reading display frame characteristic labels, reading is carried out on the target object image again, and the new middle reading is the meter data and is output.
As a preferable scheme of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge, the invention comprises the following steps: uploading the data management platform and correcting comprises uploading instrument data, verifying the instrument data based on historical data, and issuing a correction request when certain instrument data exceeds a preset range; the original image is found based on the correction request to calibrate the meter data.
It is another object of the present invention to provide an intelligent image recognition system for a lightning arrester counter and a gas pressure gauge, which can recognize the type of meter and meter data quickly and accurately.
In order to solve the technical problems, the invention provides the following technical scheme: a system for an intelligent image recognition method of a lightning arrester counter and a gas pressure gauge comprises the following components: the device comprises a request initiating module, an image acquisition module, a characteristic enhancement module, an instrument identification module, a data reading module and a data uploading module, wherein the request initiating module, the image acquisition module, the characteristic enhancement module, the instrument identification module, the data reading module and the data uploading module are connected in sequence; the request initiating module is used for initiating an instrument data acquisition request by the data management platform; the image acquisition module is used for acquiring an instrument image by adopting a movable terminal; the characteristic enhancement module is used for preprocessing the instrument image to enhance the identifiable characteristic of the image; the instrument identification module is used for constructing an instrument model identification model, and identifying pictures through the instrument model identification model to obtain an instrument model; the data reading module is used for calling a corresponding instrument structure model according to the obtained instrument model, and identifying instrument readings in the image by using an image identification algorithm through the structure model to obtain instrument data; and the data uploading module is used for uploading the instrument data to the data management platform and correcting the instrument data.
A computer device comprising a memory and a processor, said memory storing a computer program, characterized in that said processor, when executing said computer program, implements the steps of a lightning arrester counter and gas pressure gauge intelligent image recognition method as described above.
A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of a lightning arrester counter and gas pressure gauge intelligent image recognition method as described above.
The invention has the beneficial effects that: the method can solve the problem of low efficiency and inaccuracy of the traditional manual method, and can better match the relation between the color and the brightness by adopting the color brightness cross entropy loss as a standard, so that the picture can accurately identify the color under different brightness. The invention also coordinates the instrument model by grids, and can accurately identify the data in the instrument display area, thereby realizing efficient and accurate instrument data acquisition. The method provides powerful support for production management of enterprises, is beneficial to improving production efficiency and reducing cost, and creates greater value for the enterprises.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:
fig. 1 is a flowchart of a method for identifying an intelligent image of a lightning arrester counter and a gas pressure gauge in embodiment 1.
Fig. 2 is a block diagram of an intelligent image recognition system for a lightning arrester counter and a gas pressure gauge in embodiment 3.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Example 1
Referring to fig. 1, a first embodiment of the present invention provides a lightning arrester counter and gas pressure gauge intelligent image recognition method, which includes, as shown in fig. 1:
step 1: the data management platform initiates a meter data acquisition request.
The data acquisition request includes a data acquisition range and a data acquisition time period.
In this data acquisition request, the platform explicitly proposes the range and time period of data acquisition. The data collection range refers to the total number of meters in units of area of interest to the platform.
Meanwhile, the data acquisition request also defines the time period of data acquisition. This period of time is divided into two aspects: firstly, the starting time and the ending time of data acquisition ensure the consistency of the data acquisition process; and secondly, the period of data acquisition, such as daily, weekly or monthly, ensures the real-time update of the data. By setting a definite time period, the platform can ensure timeliness and accuracy of the collected data, and is beneficial to enterprises to discover problems and adjust strategies in time.
Step 2: collecting an instrument image by adopting a movable terminal;
The mobile terminal comprises a smart phone, a tablet computer or special data acquisition equipment. Smart phones and tablet computers become important tools in production sites because of their high popularity, portability, and rich functionality. The method can collect the instrument images in the production environment at any time and any place, and provide real-time and accurate data support for management staff.
Step 3: preprocessing the meter image enhances the image identifiable characteristics. The method comprises the following specific steps:
Performing gray level conversion on the image: gray-scale conversion is the process of converting a color image into a gray-scale image, which helps reduce redundant information in the image, making the image processing algorithm more efficient. In the gradation conversion process, the color information of each pixel point is replaced with its corresponding luminance value. In this way, the color information in the image is simplified into one-dimensional data, and subsequent processing is facilitated.
Image noise reduction using gaussian blur: noise is an adverse effect of the image during transmission, capture and processing, which can reduce image quality and affect the accuracy of image analysis. Gaussian blur is a classical noise reduction method that achieves noise reduction by weighted averaging of images in the spatial domain. The Gaussian blur can effectively eliminate random noise in the image, and meanwhile, the edge information of the image is reserved, so that a clean and clear image foundation is provided for subsequent processing.
Binarizing the image: binarization is the setting of the pixel value in an image to 0 or 255, dividing the image into two regions: background and target. The binarization processing is helpful to highlight the target object in the image, so that the subsequent target recognition and segmentation are facilitated. In the binarization process, a threshold value needs to be set, an area with a pixel value larger than the threshold value is set as a target, and an area with a pixel value smaller than the threshold value is set as a background. The degree of distinguishing the target from the background in the binarized image can be flexibly controlled by adjusting the threshold value.
In summary, by performing gray level conversion, gaussian blur and binarization processing on the image, a clean, clear and easy-to-analyze image can be obtained. The three steps are basic means of image preprocessing, and lay a foundation for image recognition and analysis. In practical application, the preprocessed image can be further processed, such as edge detection, feature extraction and the like, according to requirements of different tasks, so that more accurate image analysis and recognition can be realized.
Step 4: and constructing an instrument model identification model, and identifying the picture through the instrument model identification model to obtain the instrument model.
Collecting image datasets containing different instrument models; this data set should be highly diverse and representative so that the model can learn the characteristics of various meters. Such data may be obtained from the internet, an in-enterprise database, or a professional data provider. Care is taken to ensure that the image quality is high and the definition is sufficient for subsequent image processing and recognition operations when collecting the data.
Selecting any kind of image recognition models, such as a convolutional neural network model, a depth residual network model and the like, carrying out feature labeling on information reflecting instrument type colors in a picture, inputting the information into the image recognition models to be marked as a first image recognition model, carrying out feature labeling on brightness values corresponding to various types of colors in the picture subjected to gray processing, inputting the brightness values into the image recognition models to be marked as a second image recognition model, carrying out feature labeling on information reflecting instrument type structural shapes in the picture subjected to binarization processing, inputting the information into the image recognition models to be marked as a third image recognition model, and finally inputting an original picture into the image recognition models to be marked as a fourth image recognition model.
And acquiring correct color, brightness and structure information which should be output in each picture, calculating the probability of outputting the correct color in the first image recognition model, calculating the probability of outputting the correct brightness in the second image recognition model, calculating the probability of outputting the correct structure in the third image recognition model, and calculating the probability of outputting the correct color and the correct structure in the fourth image recognition model.
According to the obtained probability, calculating the color brightness cross entropy loss of the correct color probability of the first image recognition model and the correct brightness probability of the second image recognition model, adjusting the super-parameters of the first image recognition model according to the color brightness cross entropy loss until the color brightness cross entropy loss meets the preset requirement, ending the adjustment, obtaining a secondary first image recognition model, and calculating the probability of the correct color output by the secondary first image recognition model.
Because in the actual working condition, the brightness of the collected pictures is different due to the difference of the shooting angle and the sunlight, and the invention adopts the color brightness cross entropy loss as a judgment standard to adjust, so that the relation between the colors and the brightness can be better matched, the pictures can accurately identify the correct colors under different brightness, and the accuracy is improved. The super parameters comprise learning rate, batch size, learning attenuation rate and the like, and are determined by the actual working conditions and the selected model.
And calculating the color cross entropy loss of the correct color probability output by the secondary first image recognition model and the correct color probability of the fourth image recognition model, and adjusting the super-parameters of the fourth image recognition model according to the color cross entropy loss until the color cross entropy loss meets the preset requirement, and ending the adjustment.
And calculating the structural cross entropy loss of the correct structural probability of the third image recognition model and the correct structural probability of the fourth image recognition model, and adjusting the super-parameters of the fourth image recognition model according to the structural cross entropy loss until the structural cross entropy loss meets the preset requirement, and ending the adjustment.
The cross entropy calculation formula is expressed as,
Where N is represented as the number of samples, y i is represented as the correct feature for the ith sample, and p i is represented as the probability that the model predicts the correct feature class for the ith sample.
And outputting a fourth image recognition model after the adjustment is finished, namely, obtaining the instrument model recognition model.
During the training process, test sets are continuously used to evaluate and optimize the model performance. The test set is a set of data independent of the training set for verifying the performance of the model on unknown data. By comparing the prediction result on the test set with the actual label, the performance index of the model, such as accuracy, recall rate and the like, can be obtained. According to the indexes, parameters such as a network structure, a learning rate and the like can be adjusted so as to improve the generalization capability of the model.
And identifying the instrument image acquired in real time by using the trained model, and outputting the instrument model.
And inputting the real-time image into a model, and outputting a recognition result, namely the instrument model by the model. This step can be applied to actual production scenarios, providing real-time, accurate meter identification services for enterprises.
Therefore, a high-performance instrument model identification model can be constructed, and accurate identification of different instrument models is realized. In practical application, the model structure and the training strategy are continuously optimized, so that the recognition accuracy and the real-time performance can be further improved.
Step 5: and calling a corresponding instrument structure model according to the obtained instrument model, and identifying instrument readings in the image by using an image identification algorithm through the structure model to obtain instrument data.
Identifying the meter reading in the image comprises the steps of identifying the specific position of the reading according to the structural model, gridding the extracted meter structural model, determining the size of each grid according to the actual working condition, determining the intersection point between each grid as a grid point, giving coordinate values, enabling the coordinate of the centremost grid point to be (0, 0), enabling the length size of each grid to be 1, sequentially obtaining the coordinate of each grid point, splitting the structural characteristics of the meter into the characteristics of a reading display frame and the characteristics of a meter shell, and labeling labels respectively.
And loading the reading display frame characteristics and the instrument shell characteristics into the meshed instrument structure model to obtain grid point coordinates with corresponding labels respectively, collecting grid point coordinates with two labels simultaneously, invalidating the grid point coordinates, and finally outputting a structural grid coordinate graph.
And binarizing the image by taking the structural characteristic of the instrument as a threshold condition, marking the pixel points larger than the threshold as a target object, marking the pixel points smaller than the threshold as a background, and finally outputting the target object image.
And embedding the structural grid coordinate graph into the target object image, acquiring pixel information at grid point coordinates with the tag being the reading display frame characteristic, and extracting the pixel information to obtain a primary reading.
And acquiring readings in specific positions and matching the readings with the type of the instrument to obtain a numerical unit.
And calibrating the extracted readings to correct the identification errors, combining the primary readings with the matched numerical units to obtain intermediate readings, checking the intermediate readings according to the common meter reading digits of the database, if the deviation digits are less than or equal to one digit, obtaining the intermediate readings as meter data and outputting the meter data, if the deviation digits are greater than one digit, attaching the invalid lattice points in the structural grid coordinate graph with reading display frame characteristic labels, and reading the target object image again to obtain new intermediate readings as meter data and outputting the new intermediate readings.
In the actual working condition, if a high-precision complex algorithm is adopted to identify the data in the reading display frame, although the precision is high, the iteration is troublesome and time-consuming, and in the working condition of the invention, the extremely high-precision data acquisition is not needed.
Step 6: and uploading the meter data to a data management platform and correcting.
The method comprises the following specific steps:
Uploading instrument data: first, the meter data needs to be uploaded in real time. The uploaded data includes, but is not limited to: temperature, pressure, flow and other various instrument parameters.
The meter data is verified based on the history data, and a correction request is issued when a certain meter data exceeds a preset range.
The method comprises the following specific steps: statistically analyzing the collected historical data using a statistical method to determine a typical reading range for the meter; setting a reasonable threshold according to the analysis result of the historical data; and comparing each item of instrument data with a set threshold value, and if the current data exceeds a preset range, issuing a correction request by the system. The collected historical data is specifically required to be arranged, so that the accuracy and the integrity of the data are ensured. This step is the basis for subsequent analysis, and then the consolidated data is subjected to descriptive analysis, including calculation of statistics of mean, standard deviation, maximum, minimum, etc. These statistics can help to understand the distribution characteristics of the data, and then from the results of the descriptive analysis, a typical reading range for the meter can be determined. A reasonable threshold can be calculated from the empirical formula of the normal distribution. Finally, each item of meter data is compared with a set threshold value. If the current data is outside the preset range, the system will issue a correction request.
The original image is found based on the correction request to calibrate the meter data.
The method comprises the following specific steps: retrieving the associated raw image data based on the information provided in the correction request; analyzing the retrieved image to determine an actual reading of the meter; calibrating the instrument according to the image analysis result; and updating the calibration record of the instrument, including information such as calibration date, calibration value, calibration personnel and the like.
Upon receiving the correction request, the system will retrieve the relevant raw image data based on the information provided in the request (e.g., timestamp, meter identifier, etc.). Once the image is retrieved, the system will use more accurate image processing and analysis algorithms to determine the actual reading of the meter. The actual reading results obtained are then used to calibrate the uploaded data, ensuring that the reading is accurate. After calibration is completed, the system updates the calibration record of the instrument, including calibration date, calibration value, and information such as the calibrator. These records are critical to tracking meter performance and maintenance history and also facilitate future fault diagnosis and preventative maintenance.
Example 2
A second embodiment of the present invention, which is different from the first embodiment, is: the intelligent image recognition method for the lightning arrester counter and the gas pressure gauge further comprises the step of comparing test results by means of scientific demonstration by adopting the traditional technical scheme with the method of the invention in order to verify and explain the technical effects adopted in the method, so that the actual effects of the method are verified.
Design of experiment
Efficiency comparison experiment: the time required for both methods (conventional manual recording vs. the method of the present invention) to complete the same number of data recording tasks was tested.
Accuracy test: the performance of the two methods in terms of data recording accuracy was compared.
Stability and adaptability test: the stability and adaptability of the automatic acquisition method is tested under different environmental conditions (e.g. different brightness and color conditions).
Cost analysis: the implementation costs of the two methods are compared, including equipment costs, maintenance costs, and labor costs.
User friendliness and ease of operation assessment: the skill and convenience required to implement and operate both methods was assessed. Finally, the data are summarized in table 1:
table 1: experimental data comparison table
Average completion time: the method of the invention obviously reduces the time for completing the same task and improves the efficiency.
Accuracy rate: the method of the invention utilizes a model algorithm, reduces human input errors and improves the accuracy of data.
Performance at different brightness: the method can keep high identification precision under various illumination conditions through the use of color brightness cross entropy loss.
Cost analysis: although the initial equipment cost of the method is higher, the operation and maintenance cost is lower in the long term due to the reduction of labor cost.
User friendliness: both approaches are similar in terms of user friendliness, requiring a degree of learning and adaptation by the operator.
Example 3
Referring to fig. 2, a third embodiment of the present invention is shown, which is different from the first two embodiments: a system of an intelligent image recognition method of a lightning arrester counter and a gas pressure gauge is characterized in that: the system comprises a request initiating module, an image acquisition module, a characteristic enhancement module, an instrument identification module, a data reading module and a data uploading module; the device comprises a request initiating module, an image acquisition module, a characteristic enhancement module, an instrument identification module, a data reading module and a data uploading module which are sequentially connected.
The request initiating module is used for initiating an instrument data acquisition request by the data management platform.
And the image acquisition module is used for acquiring the instrument image by adopting the movable terminal.
And the characteristic enhancement module is used for preprocessing the instrument image to enhance the identifiable characteristic of the image.
The instrument identification module is used for constructing an instrument model identification model, and identifying pictures through the instrument model identification model to obtain an instrument model.
And the data reading module is used for calling a corresponding instrument structure model according to the obtained instrument model, and identifying instrument readings in the image by using an image identification algorithm through the structure model to obtain instrument data.
And the data uploading module is used for uploading the instrument data to the data management platform and correcting the instrument data.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (10)
1. An intelligent image recognition method for a lightning arrester counter and a gas pressure gauge is characterized in that: comprising the steps of (a) a step of,
The data management platform initiates an instrument data acquisition request;
collecting an instrument image by adopting a movable terminal;
Preprocessing the instrument image to enhance the identifiable characteristics of the image;
building an instrument model identification model, and identifying the picture through the instrument model identification model to obtain an instrument model;
according to the obtained instrument model, a corresponding instrument structure model is called, instrument readings in the image are identified through the structure model by using an image identification algorithm, and instrument data are obtained;
and uploading the meter data to a data management platform and correcting.
2. The intelligent image recognition method for the lightning arrester counter and the gas pressure gauge according to claim 1, wherein the intelligent image recognition method comprises the following steps of: the data acquisition request comprises a data acquisition range and a data acquisition time period;
the mobile terminal comprises a smart phone, a tablet computer or special data acquisition equipment.
3. The intelligent image recognition method for the lightning arrester counter and the gas pressure gauge according to claim 2, wherein the intelligent image recognition method comprises the following steps of: the preprocessing of the instrument image comprises gray level conversion of the image; reducing image noise by using Gaussian blur; and performing binarization processing on the image.
4. A lightning arrester counter and gas pressure gauge intelligent image recognition method as defined in claim 3, wherein: selecting any kind of image recognition model, carrying out feature labeling on information reflecting instrument type colors in a picture, inputting the information into the image recognition model to be marked as a first image recognition model, carrying out feature labeling on brightness values corresponding to various types of colors in the picture subjected to gray scale processing, inputting the brightness values into the image recognition model to be marked as a second image recognition model, carrying out feature labeling on information reflecting instrument type structural shapes in the picture subjected to binarization processing, inputting the information into the image recognition model to be marked as a third image recognition model, and finally inputting an original image into the image recognition model to be marked as a fourth image recognition model;
acquiring correct color, brightness and structure information which should be output in each picture, calculating the probability of outputting the correct color in a first image recognition model, calculating the probability of outputting the correct brightness in a second image recognition model, calculating the probability of outputting the correct structure in a third image recognition model, and calculating the probability of outputting the correct color and the correct structure in a fourth image recognition model;
According to the obtained probability, calculating the color brightness cross entropy loss of the correct color probability of the first image recognition model and the correct brightness probability of the second image recognition model, adjusting the super-parameters of the first image recognition model according to the color brightness cross entropy loss until the color brightness cross entropy loss meets the preset requirement, ending the adjustment, obtaining a secondary first image recognition model, and calculating the probability of the secondary first image recognition model outputting correct colors;
calculating the color cross entropy loss of the probability of outputting correct colors of the secondary first image recognition model and the probability of outputting correct colors of the fourth image recognition model, and adjusting the super parameters of the fourth image recognition model according to the color cross entropy loss until the color cross entropy loss meets the preset requirement;
Calculating the structure cross entropy loss of the correct structure probability of the third image recognition model and the correct structure probability of the fourth image recognition model, and adjusting the super-parameters of the fourth image recognition model according to the structure cross entropy loss until the structure cross entropy loss meets the preset requirement;
And outputting a fourth image recognition model after the adjustment is finished, namely, obtaining the instrument model recognition model.
5. The intelligent image recognition method for the lightning arrester counter and the gas pressure gauge according to claim 4, wherein the intelligent image recognition method comprises the following steps: the instrument reading in the identification image comprises the steps of identifying the specific position of the reading according to the structural model;
acquiring readings in specific positions and matching the readings with the type of the instrument to obtain a numerical unit;
The extracted readings are calibrated to correct for the identification errors.
6. The intelligent image recognition method for the lightning arrester counter and the gas pressure gauge according to claim 5, wherein the intelligent image recognition method comprises the following steps of: the specific position of the identification reading comprises gridding an extracted instrument structure model, determining the size of each grid according to the actual working condition, determining the intersection point between each grid as a grid point, giving coordinate values, enabling the coordinate of the centremost grid point to be (0, 0), enabling the length size of each grid to be 1, sequentially obtaining the coordinate of each grid point, splitting the structural characteristics of the instrument into a reading display frame characteristic and an instrument shell characteristic, and marking labels respectively;
Loading the reading display frame characteristics and the instrument shell characteristics into the meshed instrument structure model to obtain grid point coordinates with corresponding labels respectively, collecting grid point coordinates with two labels simultaneously, invalidating the grid point coordinates, and finally outputting a structural grid coordinate graph;
binarizing the image by taking the structural characteristic of the instrument as a threshold condition, marking the pixel points larger than the threshold as a target object, marking the pixel points smaller than the threshold as a background, and finally outputting the image of the target object;
Embedding the structural grid coordinate graph into a target object image, acquiring pixel information of grid point coordinates with labels being reading display frame characteristics, and extracting the pixel information to obtain primary reading;
The calibration comprises the steps of combining the primary reading with the matched numerical units to obtain a middle reading, checking the middle reading according to the common meter reading number of the database, if the deviation number is smaller than or equal to one, the middle reading is the meter data and is output, if the deviation number is larger than one, the invalid grid points in the structural grid coordinate graph are attached with reading display frame characteristic labels, reading is carried out on the target object image again, and the new middle reading is the meter data and is output.
7. The intelligent image recognition method for the lightning arrester counter and the gas pressure gauge according to claim 6, wherein the intelligent image recognition method comprises the following steps: uploading the data management platform and correcting comprises uploading instrument data, verifying the instrument data based on historical data, and issuing a correction request when certain instrument data exceeds a preset range;
the original image is found based on the correction request to calibrate the meter data.
8. A system employing a lightning arrester counter and gas pressure gauge intelligent image recognition method according to any one of claims 1 to 7, characterized in that: the device comprises a request initiating module, an image acquisition module, a characteristic enhancement module, an instrument identification module, a data reading module and a data uploading module, wherein the request initiating module, the image acquisition module, the characteristic enhancement module, the instrument identification module, the data reading module and the data uploading module are connected in sequence;
the request initiating module is used for initiating an instrument data acquisition request by the data management platform;
the image acquisition module is used for acquiring an instrument image by adopting a movable terminal;
the characteristic enhancement module is used for preprocessing the instrument image to enhance the identifiable characteristic of the image;
The instrument identification module is used for constructing an instrument model identification model, and identifying pictures through the instrument model identification model to obtain an instrument model;
The data reading module is used for calling a corresponding instrument structure model according to the obtained instrument model, and identifying instrument readings in the image by using an image identification algorithm through the structure model to obtain instrument data;
and the data uploading module is used for uploading the instrument data to the data management platform and correcting the instrument data.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: the steps of the intelligent image recognition method of the lightning arrester counter and the gas pressure gauge according to any one of claims 1 to 7 are realized when the processor executes the computer program.
10. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program, when executed by a processor, implements the steps of a lightning arrester counter and gas pressure gauge intelligent image recognition method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410032515.3A CN117994769A (en) | 2024-01-09 | 2024-01-09 | Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410032515.3A CN117994769A (en) | 2024-01-09 | 2024-01-09 | Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117994769A true CN117994769A (en) | 2024-05-07 |
Family
ID=90900033
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410032515.3A Pending CN117994769A (en) | 2024-01-09 | 2024-01-09 | Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117994769A (en) |
-
2024
- 2024-01-09 CN CN202410032515.3A patent/CN117994769A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109635806B (en) | Ammeter value identification method based on residual error network | |
| CN116434266B (en) | Automatic extraction and analysis method for data information of medical examination list | |
| CN115713691A (en) | Pixel-level electric power popularity estimation method and device based on noctilucent remote sensing | |
| CN118211882B (en) | Product quality management system and method based on big data | |
| CN114324336B (en) | Nondestructive measurement method for biomass of soybean in whole growth period | |
| CN116091786A (en) | Holographic body ruler self-coding method, system, equipment and storage medium for pig weight estimation | |
| CN110849821A (en) | Black and odorous water body remote sensing identification method based on Bayesian theorem | |
| CN112529003A (en) | Instrument panel digital identification method based on fast-RCNN | |
| CN116863341A (en) | Crop classification and identification method and system based on time sequence satellite remote sensing image | |
| CN117994769A (en) | Intelligent image recognition method and system for lightning arrester counter and gas pressure gauge | |
| CN116824437A (en) | Measurement method and system for seat body forward bend | |
| CN115588196A (en) | Pointer type instrument reading method and device based on machine vision | |
| CN110866440B (en) | Pointer instrument deep learning network construction method, pointer instrument deep learning network detection method and pointer instrument deep learning network construction system | |
| CN113989632A (en) | Bridge detection method and device for remote sensing image, electronic equipment and storage medium | |
| CN114047471A (en) | Electric energy meter calibration method and device, electronic equipment and storage medium | |
| CN112906528A (en) | Urban building material classification method fusing multi-source satellite remote sensing data | |
| CN110276755A (en) | A kind of knub position positioning system and relevant apparatus | |
| CN115690585B (en) | Method and system for extracting wheat tillering number based on digital photo | |
| CN118097093B (en) | System for searching images on digital pathological section data set based on pathological large model | |
| CN117789037B (en) | Crop growth period prediction method and device | |
| CN116778563B (en) | Image processing method based on teeth and attached gingiva and related equipment | |
| CN118278822B (en) | Working data acquisition method and system based on image data analysis and electronic equipment | |
| CN117765540A (en) | Liquid crystal water meter reading identification method and system based on deep learning | |
| CN117252819A (en) | Image sensing and transmitting method and system based on light CNN (computer numerical network) | |
| CN118115883A (en) | Method, device and equipment for identifying crop planting areas along power transmission line |
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 |