Pointer meter reading recognition method for intelligent inspection robot

Abstract

The present invention relates to the technical field of image acquisition and recognition, and specifically provides a pointer meter reading recognition method for an intelligent inspection robot. The method comprises: importing a current working state and environmental monitoring data of a connection device corresponding to a target meter; importing related data of a target inspection device; determining a target photographing angle of the target inspection device, and carrying out image acquisition and video acquisition on the target meter multiple times; determining whether the reading of the target meter needs reading compensation, if not, recognizing the reading of the target meter, and if yes, confirming a display compensation value for the target meter; and setting an environmental compensation value for the target meter, and calculating confirmed display reading of the target meter. The present invention effectively solves the problem of lack of the consideration of the error and state of the meter and a specific meter environment, greatly improves the accuracy of pointer reading, and ensures the effectiveness, reliability, and rationality of the pointer reading.

Description

A method for identifying and reading pointer instruments of intelligent inspection robots Technical Field

[0001] The invention belongs to the technical field of image acquisition and recognition, and relates to a method for identifying and reading pointer-type instruments of an intelligent inspection robot. Background Art

[0002] With the development of visual recognition technology and the application of deep learning algorithms, machines can capture images through cameras and use algorithms to learn and identify the readings of pointer instruments. The emergence of intelligent inspection robots meets the needs of industrial automation, improves inspection efficiency and data accuracy, and promotes the development of industrial intelligence.

[0003] Currently, intelligent inspection robots primarily identify and read pointer instruments by capturing images of the instruments and then reading them based on the captured images. Prior art, such as the Chinese invention patent application publication number CN112488030A, discloses a method for reading pointer instruments based on machine vision. This method employs machine vision to calibrate the instrument, capture a frontal image, and perform grayscale processing, Gaussian noise reduction, and edge detection to determine the pointer coordinates. Ultimately, the pointer's deflection angle is calculated and digitally presented, resulting in a more scientific and accurate reading algorithm.

[0004] Another example of the prior art is a method for reading a pointer-type pressure instrument based on machine vision, as disclosed in Chinese invention patent application publication number CN111797909A. This method uses a standard template image of an instrument and an image of an instrument to be read. A matching algorithm is used to match the image of the instrument to be read with the standard template image of the instrument. The image of the instrument to be read is corrected and processed by denoising, threshold segmentation, and dilation operations, thereby confirming the instrument reading and making the instrument reading more accurate.

[0005] Regarding the above two technical solutions, it is obvious that they mainly improve the accuracy of readings by performing a series of processing on the collected images. Obviously, there are still the following deficiencies: 1. They do not take into account the errors and states of the instrument itself, such as the jitter of the pointer, which makes the readings insufficiently effective, and thus leads to certain deficiencies in the reliability and rationality of the instrument readings collected by the inspection machine. At the same time, it may not be possible for the robot to accurately capture these errors when identifying the readings, and thus misjudgment or inaccuracy may occur in the reading identification process, thereby affecting the accuracy of the readings.

[0006] 2. The impact of the specific instrument environment was not taken into account. For example, the smoothness of the robot's moving path will lead to deviations in the camera angle and changes in temperature and humidity in the environment, which in turn increases the probability of errors in the inspection robot's readings, resulting in certain deviations in the effectiveness and accuracy of the recognition results. It is impossible to ensure the improvement of the inspection robot's reading and recognition accuracy.

[0007] Summary of the Invention

[0008] In view of this, in order to solve the problems raised in the above background technology, a method for identifying and reading pointer-type instruments of an intelligent inspection robot is proposed.

[0009] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a method for identifying and reading pointer-type instruments of an intelligent inspection robot, the method comprising: step one, importing reading environmental monitoring data: recording the current target inspection robot as the target inspection device, recording the current instrument to be identified as the target instrument, and importing the current working status and environmental monitoring data of the corresponding connected device of the target instrument.

[0010] Step 2: Import relevant data of inspection equipment: Import relevant data of the target inspection equipment, including the currently set camera angle and the corresponding monitored vibration frequency at each driving time point.

[0011] Step 3: Instrument image acquisition: Determine the target camera angle of the target inspection equipment, and perform multiple image acquisitions on the target instrument to obtain each acquired image, record the acquisition time point of each acquired image, and start the auxiliary camera of the target inspection equipment to acquire the target instrument video.

[0012] Step 4: Preliminary assessment of instrument images: Based on the collected images and videos of the target instrument, determine whether the target instrument reading requires reading compensation. If the judgment result is no, start step 5; if the judgment result is yes, start step 6.

[0013] Step 5: Instrument image reading recognition: take the last captured image as the target recognition image, recognize the target recognition image, and output the reading Z of the target recognition image. m .

[0014] Step 6. Confirm the instrument display compensation: Confirm the display compensation value Z of the target instrument η .

[0015] Step 7: Confirmation of instrument image readings: used to set the target instrument environmental compensation value Z according to the environmental monitoring data. ε , and calculate the confirmed display reading of the target meter.

[0016] Preferably, the determining of the target camera angle of the target inspection device includes: recording the camera angle currently set by the target inspection device as

[0017] With the driving time point as the horizontal coordinate and the monitored vibration frequency as the vertical coordinate, the driving vibration change curve of the target inspection equipment is constructed, and the total length l of the curve segment above the set interference driving vibration frequency is selected from it. g and the length of each curve segment, and at the same time, the maximum value is selected from the length of each curve segment and recorded as l x .

[0018] The length of the driving vibration change curve is recorded as l b , calculate the vibration interference trend index γ of the target inspection equipment,

[0019] If γ is less than or equal to the set reference vibration interference tendency index γ′, As the target camera angle of the target inspection equipment, As the target camera angle of the target inspection equipment, The compensation camera angle corresponds to the set unit vibration interference trend index difference.

[0020] Preferably, the determining whether the target instrument reading requires reading compensation includes: extracting the calibration tracking log of the target instrument and the set accuracy and calibration interval days from a device information library.

[0021] The calibration date and calibration deviation value of each cumulative calibration are located from the calibration tracking log, and the current numerical error trend degree δ of the target instrument is calculated based on the calibration date and calibration deviation value.

[0022] Identify each collected image, obtain the indication area position of the pointer corresponding to the target instrument in each collected image, and calculate the rotation error trend β of the target instrument according to the collection time point of each collected image.

[0023] According to the video of the target instrument, the consistency λ of the current indication state of the target instrument is analyzed.

[0024] δ>δ′ is used as judgment condition A, β>β′ is used as judgment condition B, and λ<λ′ is used as judgment condition C. δ′, β′, and λ′ are respectively the numerical error trend, rotation error trend, and indication state consistency of the set reference.

[0025] When the judgment condition A, judgment condition B and judgment condition C are all not established, "no" will be used as the judgment result, otherwise "yes" will be used as the judgment result.

[0026] Preferably, the statistical target instrument's current numerical error trend includes: taking the cumulative calibration order as the horizontal coordinate and the calibration deviation value as the vertical coordinate, constructing a cumulative calibration deviation change curve, and extracting the slope from it, which is recorded as k pAt the same time, the total length of the curve segment above the target instrument setting accuracy J′ is selected and recorded as l p .

[0027] If l p =0 and k p ≤k′ p , taking μ0 as the current numerical error trend of the target instrument, k′ p is the set reference deviation growth rate.

[0028] If l p =0 and k p >k′ p , calculate the current expected deviation value of the target instrument, recorded as J a ,Will The current numerical error trend of the target instrument is recorded as μ1.

[0029] If l p >0, the current numerical error trend of the target instrument is evaluated by the instrument display deviation evaluation rule, recorded as μ2, and the current numerical error trend of the target instrument δ is obtained, and the value of δ is μ0, μ1 or μ2, and μ2>μ1>μ0.

[0030] Preferably, the counting of the rotation error trend of the target instrument includes: if the current working state of the target instrument corresponding to the connected device is working, extracting the reference pointer deflection value X0 of the target instrument within a corresponding unit time point in the working state of the connected device from the device information database;

[0031] Based on the indication area position of the corresponding pointer of the target instrument in each collected image, the pointer deflection value of the target instrument at each collection time point is obtained, which is recorded as X t , t represents the acquisition time point number, t=1,2,......u.

[0032] Will The rotation error trend of the target instrument is recorded as φ0, X t+1 It represents the pointer deflection value of the target instrument at the t+1th acquisition time point, u represents the number of acquisition time points, ΔX is the set allowable pointer deflection deviation value, is the floor symbol.

[0033] If the current working state of the connected device corresponding to the target instrument is idle, the collected images of the target instrument are sorted in chronological order according to the collection time, the first-ranked image is used as the reference image, and the other collected images are used as comparison images.

[0034] If the position of the pointer area in a comparison image is inconsistent with that in the reference image, the comparison image is recorded as a deviation image, and the number of deviation images M0 is counted, and the number of collected images M′ is counted at the same time.

[0035] Will As the rotation error trend of the target instrument, it is recorded as φ1. is the rounding sign, ΔX′ is the set allowable fluctuation deflection value, and the rotation error trend β of the target instrument is obtained, and the value of β is φ0 or φ1.

[0036] Preferably, the analysis of the consistency of the current indication state of the target instrument includes: detecting and tracking the target instrument in its video through a target detection algorithm and a tracking algorithm, obtaining the motion trajectory of the corresponding pointer of the target instrument between consecutive frames, analyzing the number of jitters of the target instrument based on this, and extracting the motion trajectory and motion trajectory length of each jitter.

[0037] The average length of each jitter trajectory is calculated and the result is recorded as The number of jitters of the target instrument and the duration of the corresponding video are recorded as C and T respectively, and the consistency of the current indication state of the target instrument is calculated as λ. l h , C′, and k0 are the jitter trajectory length, number of jitters, and jitter frequency set for warning respectively.

[0038] Preferably, the determination of the display compensation value of the target instrument includes: if the judgment condition A is met, taking (δ-δ′)*Z0 as the display compensation value of the target instrument, recorded as Z a , Z0 is the compensation value corresponding to the unit numerical error trend deviation of the set target instrument.

[0039] If the judgment condition B is met, (β-β′)*Z1 is used as the display compensation value of the target instrument, recorded as Z b , Z1 is the compensation value corresponding to the unit rotation error trend deviation of the set target instrument.

[0040] If the judgment condition C is established, analyze the display compensation value of the target instrument under the judgment condition C, and record it as Z c .

[0041] If the judgment condition A and the judgment condition B are met, Z a +Z b As the display compensation value of the target instrument, it is recorded as Z d .

[0042] If the judgment conditions A and C are met, Z a +Z c As the display compensation value of the target instrument, it is recorded as Z e.

[0043] If the judgment conditions B and C are met, Z b +Z c As the display compensation value of the target instrument, it is recorded as Z f .

[0044] If the judgment conditions A, B and C are all met, Z a +Z b +Z c As the display compensation value of the target instrument, it is recorded as Z g , and thus obtain the display compensation value Z of the target instrument η , Z η The value is Z a or Z b or Z c or Z d or Z e or Z f or Z g .

[0045] Preferably, the analysis and judgment of the display compensation value of the target instrument under condition C includes: comparing the motion trajectories of each jitter with each other, and if the motion trajectories of each jitter are the same, As a display compensation weight.

[0046] If the motion trajectory of a certain jitter is inconsistent with the motion trajectory of another jitter, the jitter trajectory deviation ψ is calculated and the As the display compensation weight, it is recorded as In this way, the display compensation weight τ is obtained, and τ is taken as or ψ′ is the set allowable jitter trajectory deviation.

[0047] Take (λ′-λ)*Z2+τ*Z′ as the display compensation value Z of the target instrument under the analysis and judgment condition C c , Z2 is the compensation value corresponding to the set target instrument corresponding to the unit indication state consistency deviation, and Z′ is the increased compensation value of the set target instrument corresponding to the unit display compensation weight.

[0048] Preferably, the setting of the target instrument environment compensation value includes: extracting the temperature difference, maximum humidity value and average vibration frequency value of each monitoring day from the environmental monitoring data, and determining the number of temperature interference monitoring days R according to the environmental interference judgment rule. wd , humidity interference monitoring days R sd and the number of vibration interference monitoring days R zd .

[0049] The vibration frequency of each monitoring day is averaged to obtain the average vibration frequency f, and the number of monitoring days R0 is counted.

[0050] Will As the target instrument environment compensation value Z ε , f′ is the set interference vibration frequency, Z υ It is the environmental compensation value of the unit environmental interference factor corresponding to the set target instrument.

[0051] Preferably, the calculation of the confirmation display reading of the target instrument includes: when the judgment result of step 4 is yes, Z m +Z η +Z ε Serves as confirmation display reading of the target meter.

[0052] When the judgment result of step 4 is no, Z m +Z ε Serves as confirmation display reading of the target meter.

[0053] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention determines whether reading compensation is required based on each captured image of the target instrument, and when the judgment result is yes, confirms and displays the compensation value, and sets the environmental compensation value according to the environmental monitoring data of the target instrument. This effectively solves the current problem of not taking into account the instrument's own errors and status and the specific instrument environment, greatly improves the accuracy of pointer readings, and at the same time ensures the validity, reliability and rationality of pointer readings, and reduces errors in the measurement process as much as possible, thereby improving the representativeness of the measured values ​​and subsequent actual reading values.

[0054] (2) The present invention improves the stability and clarity of image acquisition by confirming the target camera angle according to the vibration frequency monitored at each target driving time point, avoiding the need to spend additional time and effort on corrections due to poor image quality in the subsequent data processing and analysis stages, thereby ensuring the availability and efficiency of image acquisition. It is also beneficial to accurately capture the pointer position and reading of the instrument, which can reduce errors caused by angle problems and improve the accuracy of data acquisition.

[0055] (3) The present invention judges whether the target instrument reading requires reading compensation from three dimensions: numerical error during calibration, pointer rotation error, and indication state matching. This achieves a comprehensive judgment on the target instrument reading compensation requirement, which can reduce the probability of misjudgment caused by a single factor, and thus more objectively evaluate whether reading compensation is needed, avoiding unnecessary adjustments and interventions. At the same time, it can also provide a more comprehensive understanding of the accuracy of the target instrument reading. On another level, it can also provide maintenance personnel with clearer guidance and formulate targeted instrument maintenance plans.

[0056] (4) The present invention compensates for the deficiency that the current robot may not be able to accurately capture subtle errors when identifying readings by confirming the target instrument display compensation value and the environmental compensation value, and reduces the probability of misjudgment or inaccuracy in the reading identification process as much as possible, thereby ensuring the accuracy of the readings. It also reduces the error rate in the automatic reading process of the inspection equipment, ensures the consistency between the recognition result and the actual value, and thus effectively guarantees the improvement of the inspection equipment reading recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0058] FIG1 is a flow chart of the implementation steps of the method of the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Please refer to Figure 1. The present invention provides a method for identifying and reading pointer-type instruments of an intelligent inspection robot. The method includes: Step 1, importing reading environmental monitoring data: recording the current target inspection robot as the target inspection device, recording the current instrument to be identified as the target instrument, and importing the current working status and environmental monitoring data of the connected device corresponding to the target instrument.

[0061] Specifically, the current working state is one of idle and working.

[0062] In another specific embodiment, the environmental monitoring data includes but is not limited to the temperature difference, the maximum humidity value and the average vibration frequency value of each monitoring day.

[0063] Step 2: Import relevant data of inspection equipment: Import relevant data of the target inspection equipment, including the currently set camera angle and the corresponding monitored vibration frequency at each driving time point.

[0064] Step 3: Instrument image acquisition: Determine the target camera angle of the target inspection equipment, and perform multiple image acquisitions on the target instrument to obtain each acquired image, record the acquisition time point of each acquired image, and start the auxiliary camera of the target inspection equipment to acquire the target instrument video.

[0065] For example, the method of determining the target camera angle of the target inspection device includes: F1, recording the currently set camera angle of the target inspection device as

[0066] F2. Using the driving time point as the horizontal coordinate and the monitored vibration frequency as the vertical coordinate, construct the driving vibration change curve of the target inspection equipment, and select the total length l of the curve segment above the set interference driving vibration frequency. g and the length of each curve segment, and at the same time, the maximum value is selected from the length of each curve segment and recorded as l x .

[0067] F3. The length of the driving vibration change curve is recorded as l b , calculate the vibration interference trend index γ of the target inspection equipment,

[0068] F4, if γ is less than or equal to the set reference vibration interference trend index γ′, As the target camera angle of the target inspection equipment, As the target camera angle of the target inspection equipment, The compensation camera angle corresponds to the set unit vibration interference trend index difference.

[0069] It should be added that the vibration frequency of the inspection equipment is monitored to evaluate the impact of the flatness of the inspection path where the inspection equipment is located on the shooting angle of the inspection equipment. That is, the vibration frequency monitored at each driving time point of the inspection equipment often refers to the vibration frequency of the chassis of the inspection equipment. When the inspection equipment is vibrated, the shooting angle of view will often be biased toward the chassis direction due to the influence of the vibration in the chassis direction, and the angle of view will be biased downward, resulting in a decrease in the camera angle. Therefore, the method of increasing the camera angle is adopted to adjust the shooting angle affected by the vibration.

[0070] The embodiment of the present invention improves the stability and clarity of image acquisition by confirming the target camera angle based on the vibration frequency monitored at each target driving time point, avoiding the need to spend additional time and effort on corrections due to poor image quality in the subsequent data processing and analysis stages, thereby ensuring the availability and efficiency of image acquisition. It is also beneficial to accurately capture the pointer position and reading of the instrument, which can reduce errors caused by angle problems and improve the accuracy of data acquisition.

[0071] Step 4: Preliminary assessment of instrument images: Based on the collected images and videos of the target instrument, determine whether the target instrument reading requires reading compensation. If the judgment result is no, start step 5; if the judgment result is yes, start step 6.

[0072] Illustratively, determining whether a target instrument reading requires reading compensation includes: H1, extracting the calibration tracking log of the target instrument and the set accuracy and calibration interval days from a device information library.

[0073] H2. Locate the calibration date and calibration deviation value of each cumulative calibration from the calibration tracking log, and calculate the current numerical error trend δ of the target instrument based on the calibration date and calibration deviation value.

[0074] H3. Identify each collected image, obtain the indication area position of the corresponding pointer of the target instrument in each collected image, and calculate the rotation error trend β of the target instrument according to the collection time point of each collected image.

[0075] It should be noted that image recognition technology is a relatively mature technology currently available, and its specific recognition process will not be described in detail here.

[0076] H4. Analyze the consistency λ of the target instrument's current indication state based on the target instrument's video.

[0077] H5. Take δ>δ′ as judgment condition A, β>β′ as judgment condition B, and λ<λ′ as judgment condition C. δ′, β′, and λ′ are the numerical error trend, rotation error trend, and indication state consistency of the set reference respectively.

[0078] H6. When the judgment conditions A, B and C are all not met, "No" will be used as the judgment result, otherwise "Yes" will be used as the judgment result.

[0079] The embodiment of the present invention determines whether the target instrument reading requires reading compensation based on three dimensions: numerical error during calibration, pointer rotation error, and indication state consistency. This achieves a comprehensive judgment on the target instrument reading compensation requirement, which can reduce the probability of misjudgment caused by a single factor, and thus more objectively evaluate whether reading compensation is needed, avoiding unnecessary adjustments and interventions. At the same time, it can also provide a more comprehensive understanding of the accuracy of the target instrument's reading. On another level, it can also provide maintenance personnel with clearer guidance and formulate targeted instrument maintenance plans.

[0080] Furthermore, in step H2, the current numerical error trend of the target instrument is calculated, including: H21, with the cumulative calibration order as the horizontal axis and the calibration deviation value as the vertical axis, a cumulative calibration deviation change curve is constructed, and the slope is extracted from it, which is recorded as k pAt the same time, the total length of the curve segment above the target instrument setting accuracy J′ is selected and recorded as l p .

[0081] It should be added that the slope refers to the slope of the regression line corresponding to the curve.

[0082] H22, if l p =0 and k p ≤k′ p , taking μ0 as the current numerical error trend of the target instrument, k′ p is the set reference deviation growth rate.

[0083] H23, if l p =0 and k p >k′ p , calculate the current expected deviation value of the target instrument, recorded as J a ,Will The current numerical error trend of the target instrument is recorded as μ1.

[0084] It should be added that the specific statistical process of calculating the current expected deviation value of the target instrument is as follows: the calibration date of the last cumulative calibration is selected from the calibration dates of each cumulative calibration and used as the analysis calibration date.

[0085] The number of days between the current date and the analysis calibration date is taken as the remaining days for calibration D jz , the interval between the calibration date of the first cumulative calibration and the last cumulative calibration date is taken as

[0086] The calibration deviation value of the last cumulative calibration is selected from the calibration deviation values ​​of each cumulative calibration, and recorded as J θ ,Will As the current expected deviation value J of the target instrument a .

[0087] H24, if l p >0, the current numerical error trend of the target instrument is evaluated by the instrument display deviation evaluation rule, recorded as μ2, and the current numerical error trend of the target instrument δ is obtained, and the value of δ is μ0, μ1 or μ2, and μ2>μ1>μ0.

[0088] In a specific embodiment, μ0 is equal to 0.

[0089] It should be added that the current display error trend of the target instrument is obtained by evaluating the instrument display deviation evaluation rule. The specific evaluation process is as follows: U1, when l p >0 and k p ≤k′ p, the total length exceeding the allowable deviation and the length of the cumulative calibration deviation change curve are recorded as l p and l w ,Will The current display error trend of the target instrument is recorded as σ0.

[0090] U2, when l p >0 and k p >k′ p ,Will As the current display error trend of the target instrument, record it as σ1, and get the current display error trend of the target instrument μ2, μ2 takes the value of σ0 or σ1, σ1>σ0, It is the compensation error trend under the set deviation growth trend.

[0091] Furthermore, in step H3, the rotation error trend of the target instrument is calculated, including: H31, if the current working state of the target instrument corresponding to the connected device is working, extracting the reference pointer deflection value X0 of the target instrument within a corresponding unit time point when the connected device is in the working state from the device information database;

[0092] H32. Based on the position of the pointer of the target instrument in each captured image, the pointer deflection value of the target instrument at each capture time point is obtained, which is recorded as X t , t represents the acquisition time point number, t=1,2,......u.

[0093] It should be added that the pointer deflection value of the target instrument at each acquisition time point is obtained by comparing the indicated area positions of the corresponding pointers of the target instrument in each acquired image with each other, obtaining the corresponding pointer deflection angle of the target instrument at each acquisition time point, and using it as the pointer deflection value.

[0094] H33, will The rotation error trend of the target instrument is recorded as φ0, X t+1 It represents the pointer deflection value of the target instrument at the t+1th acquisition time point, u represents the number of acquisition time points, ΔX is the set allowable pointer deflection deviation value, is the floor symbol.

[0095] H34. If the current working state of the connected device corresponding to the target instrument is idle, the acquired images of the target instrument are sorted in chronological order according to the acquisition time, the image ranked first is used as the reference image, and the other acquired images are used as comparison images.

[0096] H35. If the position of the pointer area in a comparison image is inconsistent with that in the reference image, the comparison image is recorded as a deviation image, and the number of deviation images M0 is counted, and the number of collected images M′ is counted at the same time.

[0097] H36, will As the rotation error trend of the target instrument, it is recorded as φ1. is the rounding sign, ΔX′ is the set allowable fluctuation deflection value, and the rotation error trend β of the target instrument is obtained, and the value of β is φ0 or φ1.

[0098] Furthermore, step H4 analyzes the degree of consistency of the current indication state of the target instrument, including: H41, detecting and tracking the target instrument in its video through the target detection algorithm and tracking algorithm, obtaining the motion trajectory of the corresponding pointer of the target instrument between consecutive frames, and analyzing the number of jitters of the target instrument based on this, and extracting the motion trajectory and motion trajectory length of each jitter.

[0099] In a specific embodiment, target detection algorithms include, but are not limited to, the YOLO algorithm and the SSD algorithm, which are used to detect targets in preprocessed video frames to determine the position and bounding box of the target instrument in each frame. Target tracking algorithms include, but are not limited to, Kalman filtering and correlation filtering. The target tracking algorithm tracks the target to obtain the target's motion trajectory between consecutive frames. The target detection and tracking algorithms are relatively mature existing algorithms, and their specific detection and tracking processes are not detailed here.

[0100] It should be added that the number of jitters of the target instrument is determined by the jitter determination rule, wherein the specific determination process of the jitter determination rule is as follows: based on the motion trajectory of the pointer corresponding to the target instrument between consecutive frames, the speed of the pointer corresponding to the target instrument in each video frame is calculated and compared with the set jitter limit speed.

[0101] If the pointer speed in a certain video frame is greater than the set jitter limit speed, it is determined to be jittering, and the number of jitters of the target instrument is counted.

[0102] It should also be added that, in a specific embodiment, the speed can be calculated by mathematical methods such as the difference method and the integration method, and the specific calculation process will not be repeated.

[0103] H42. Calculate the average length of each jitter trajectory and record the result as The number of jitters of the target instrument and the duration of the corresponding video are recorded as C and T respectively, and the consistency of the current indication state of the target instrument is calculated as λ. l h , C′, and k0 are the jitter trajectory length, number of jitters, and jitter frequency set for warning respectively.

[0104] Step 5: Instrument image reading recognition: take the last captured image as the target recognition image, recognize the target recognition image, and output the reading Z of the target recognition image. m .

[0105] In a specific embodiment, the output of the target recognition image reading is mainly achieved through image recognition technology, that is, the indication area position of the pointer corresponding to the target instrument and the overall indication value of the target instrument are identified through image recognition technology, and the angle between the pointer corresponding to the target instrument and the starting counting end corresponding to the target instrument is obtained as the target deflection angle.

[0106] Extract the maximum range of the target instrument from the device information library and As a readout of the target recognition image.

[0107] It should also be added that before target recognition image recognition is performed, operations such as image denoising, image enhancement, and image segmentation are also included. The specific implementation process is a relatively mature method available and will not be described in detail here.

[0108] Step 6. Confirm the instrument display compensation: Confirm the display compensation value Z of the target instrument η .

[0109] Exemplarily, confirming the display compensation value of the target instrument includes: Q1, if the judgment condition A is met, (δ-δ′)*Z0 is used as the display compensation value of the target instrument, recorded as Z a , Z0 is the compensation value corresponding to the unit numerical error trend deviation of the set target instrument.

[0110] Q2. If the judgment condition B is met, (β-β′)*Z1 is used as the display compensation value of the target instrument, recorded as Z b , Z1 is the compensation value corresponding to the unit rotation error trend deviation of the set target instrument.

[0111] Q3. If the judgment condition C is met, analyze the display compensation value of the target instrument under the judgment condition C, and record it as Z c .

[0112] Understandably, the analysis and judgment of the display compensation value of the target instrument under condition C includes: Q31, comparing the motion trajectories of each jitter with each other. If the motion trajectories of each jitter are the same, As a display compensation weight.

[0113] Q32. If the motion trajectory of a certain jitter is inconsistent with the motion trajectory of another jitter, calculate the jitter trajectory deviation ψ. As the display compensation weight, it is recorded as In this way, the display compensation weight τ is obtained, and τ is taken as or ψ′ is the set allowable jitter trajectory deviation.

[0114] In a specific embodiment, The value is 0.

[0115] It should be added that the specific statistical process of calculating the jitter trajectory deviation is as follows: N1. Calculate the variance of the motion trajectory length of each jitter, and use the calculation result as the jitter amplitude unevenness, recorded as χ.

[0116] N2. Extract the jitter trajectory of the first jitter from the motion trajectories of each jitter as a reference trajectory, and use the motion trajectories of the other jitters as analysis trajectories.

[0117] N3. Compare and overlap each analysis trajectory with the reference trajectory to obtain the overlap length of each analysis trajectory with the reference trajectory, and select the maximum overlap trajectory length and the minimum overlap trajectory length from them. Subtract the two to obtain the overlap trajectory length difference, which is recorded as Δl.

[0118] N4, statistical jitter trajectory deviation ψ, χ′ and Δl′ are the jitter amplitude unevenness and the overlap track length difference of the set reference, respectively.

[0119] Q33. Take (λ′-λ)*Z2+τ*Z′ as the display compensation value Z of the target instrument under the analysis and judgment condition C. c , Z2 is the compensation value corresponding to the set target instrument corresponding to the unit indication state consistency deviation, and Z′ is the increased compensation value of the set target instrument corresponding to the unit display compensation weight.

[0120] In a specific actual scenario, the display compensation value of the target instrument under judgment condition C can also be set by flat filtering, lag filtering, etc. The specific setting method is a relatively mature means available and will not be described in detail here.

[0121] Q4. If the judgment conditions A and B are met, Z a +Z b As the display compensation value of the target instrument, it is recorded as Z d .

[0122] Q5. If the judgment conditions A and C are met, Z a +Z c As the display compensation value of the target instrument, it is recorded as Z e .

[0123] Q6. If the judgment conditions B and C are met, Z b +Z cAs the display compensation value of the target instrument, it is recorded as Z f .

[0124] Q7. If the judgment conditions A, B and C are all met, set Z a +Z b +Z c As the display compensation value of the target instrument, it is recorded as Z g , and thus obtain the display compensation value Z of the target instrument η , Z η The value is Z a or Z b or Z c or Z d or Z e or Z f or Z g .

[0125] Step 7: Confirmation of instrument image readings: used to set the target instrument environmental compensation value Z according to the environmental monitoring data. ε , and calculate the confirmed display reading of the target meter.

[0126] For example, setting the target instrument environment compensation value includes: L1, extracting the temperature difference, maximum humidity value and average vibration frequency value of each monitoring day from the environmental monitoring data, and obtaining the number of temperature interference monitoring days R according to the environmental interference judgment rule. wd , humidity interference monitoring days R sd and the number of vibration interference monitoring days R zd .

[0127] It should be noted that the specific judgment process of the environmental interference judgment rule is as follows: if the temperature difference on a monitoring day is greater than the set instrument operation interference temperature difference, the monitoring day will be recorded as a temperature interference monitoring day.

[0128] If the maximum humidity value on a monitoring day is greater than the set instrument operation interference humidity value, the monitoring day will be recorded as a humidity interference monitoring day.

[0129] If the vibration frequency of a monitoring day is greater than the set instrument operation interference vibration frequency, the monitoring day will be recorded as a vibration interference monitoring day, and the number of temperature interference monitoring days, humidity interference monitoring days and vibration interference monitoring days will be evaluated.

[0130] L2. Calculate the mean of the vibration frequency of each monitoring day to obtain the average vibration frequency f, and count the number of monitoring days R0.

[0131] L3, will As the target instrument environment compensation value Z ε , f′ is the set interference vibration frequency, Z υIt is the environmental compensation value of the unit environmental interference factor corresponding to the set target instrument.

[0132] In a specific embodiment, a high temperature environment may cause the pointer to deviate from the true value, while a low temperature environment may make the pointer insensitive. A high humidity environment may cause the internal parts of the instrument to rust or get damp, thereby affecting the movement flexibility and accuracy of the pointer. Continuous vibration may cause the pointer to shake or deviate from the correct position, affecting the indication accuracy. Therefore, environmental compensation settings are made based on the three factors of temperature, humidity, and vibration frequency.

[0133] The embodiment of the present invention compensates for the deficiency that the current robot may not be able to accurately capture subtle errors when identifying readings by confirming the target instrument display compensation value and the environmental compensation value, thereby minimizing the probability of misjudgment or inaccuracy in the reading recognition process, thereby ensuring the accuracy of the readings, and also reducing the error rate in the automatic reading process of the inspection equipment, ensuring the consistency between the recognition result and the actual value, thereby effectively ensuring the improvement of the accuracy of the inspection equipment reading recognition.

[0134] In another exemplary embodiment, the calculation of the confirmation display reading of the target instrument includes: when the judgment result of step 4 is yes, m +Z η +Z ε Serves as confirmation display reading of the target meter.

[0135] When the judgment result of step 4 is no, Z m +Z ε Serves as confirmation display reading of the target meter.

[0136] The embodiment of the present invention determines whether reading compensation is required based on each collected image of the target instrument, and when the judgment result is yes, confirms and displays the compensation value, and sets the environmental compensation value according to the environmental monitoring data of the target instrument. This effectively solves the current problem of not taking into account the instrument's own errors and status and the specific instrument environment, greatly improves the accuracy of pointer readings, and at the same time ensures the validity, reliability and rationality of pointer readings, and reduces errors in the measurement process as much as possible, thereby improving the representativeness of the measured values ​​and subsequent actual reading values.

[0137] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all fall within the scope of protection of the present invention.

Claims

1. A method for identifying and reading pointer-type instruments of an intelligent inspection robot, characterized by: The method includes: Step 1: Import reading environmental monitoring data: record the current target inspection robot as the target inspection device, record the current instrument to be identified as the target instrument, and import the current working status and environmental monitoring data of the connected device corresponding to the target instrument; Step 2: Import relevant data of inspection equipment: Import relevant data of the target inspection equipment, including the currently set camera angle and the corresponding monitored vibration frequency at each driving time point; Step 3: Instrument image acquisition: Determine the target camera angle of the target inspection device, and perform multiple image acquisitions on the target instrument to obtain each acquired image, record the acquisition time point of each acquired image, and start the auxiliary camera of the target inspection device to acquire the target instrument video; Step 4: Preliminary evaluation of instrument images: Based on the collected images and videos of the target instrument, determine whether the target instrument reading requires reading compensation. If the judgment result is no, start step 5; if the judgment result is yes, start step 6; Step 5: Instrument image reading recognition: take the last captured image as the target recognition image, recognize the target recognition image, and output the reading Z of the target recognition image. m ; Step 6. Confirm the instrument display compensation: Confirm the display compensation value Z of the target instrument η ; Step 7: Confirmation of instrument image readings: used to set the target instrument environmental compensation value Z according to the environmental monitoring data. ε , and calculate the confirmed display reading of the target meter.

2. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 1, characterized in that: Determining the target camera angle of the target inspection equipment includes: The current camera angle of the target inspection device is recorded as θ0; With the driving time point as the horizontal coordinate and the monitored vibration frequency as the vertical coordinate, the driving vibration change curve of the target inspection equipment is constructed, and the total length l of the curve segment above the set interference driving vibration frequency is selected from it. g and the length of each curve segment, and at the same time, the maximum value is selected from the length of each curve segment and recorded as l x ; The length of the driving vibration change curve is recorded as l b , calculate the vibration interference trend index γ of the target inspection equipment, If γ is less than or equal to the set reference vibration interference trend index γ′, θ0 is used as the target camera angle of the target inspection equipment. Otherwise, θ0+(γ-γ′)*θ′ is used as the target camera angle of the target inspection equipment, and θ′ is the compensation camera angle corresponding to the set unit vibration interference trend index difference.

3. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 2, characterized in that: The determining whether the target instrument reading requires reading compensation includes: Extract the calibration tracking log of the target instrument and the set accuracy and calibration interval days from the device information library; Locate the calibration date and calibration deviation value of each cumulative calibration from the calibration tracking log, and calculate the current numerical error trend δ of the target instrument based on the calibration date and calibration deviation value; Identify each collected image, obtain the indication area position of the corresponding pointer of the target instrument in each collected image, and calculate the rotation error trend β of the target instrument according to the collection time point of each collected image; Analyze the consistency λ of the target instrument's current indication state based on the target instrument's video; δ>δ′ is used as judgment condition A, β>β′ is used as judgment condition B, and λ<λ′ is used as judgment condition C. δ′, β′, and λ′ are respectively the numerical error trend, rotation error trend, and indication state consistency of the set reference; When the judgment condition A, judgment condition B and judgment condition C are all not established, "no" will be used as the judgment result, otherwise "yes" will be used as the judgment result.

4. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 3, wherein: The current numerical error trend of the statistical target instrument includes: With the cumulative calibration order as the horizontal axis and the calibration deviation value as the vertical axis, the cumulative calibration deviation change curve is constructed, and the slope is extracted from it, which is recorded as k p At the same time, the total length of the curve segment above the target instrument setting accuracy J′ is selected and recorded as l p ; If l p =0 and k p ≤k′ p , taking μ0 as the current numerical error trend of the target instrument, k′ p is the set reference deviation growth rate; If l p =0 and k p >k′ p , calculate the current expected deviation value of the target instrument, recorded as J a ,Will The current numerical error trend of the target instrument is recorded as μ1; If l p >0, the current value of the target instrument is obtained by evaluating the instrument display deviation evaluation rule. The numerical error trend of the target instrument is recorded as μ2, and the current numerical error trend δ of the target instrument is obtained. The value of δ is μ0, μ1 or μ2, and μ2>μ1>μ0.

5. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 4, characterized in that: The statistical rotation error trend of the target instrument includes: If the current working state of the target instrument corresponding to the connected device is working, extract the reference pointer deflection value X0 of the target instrument corresponding to the unit time point in the working state of the connected device from the device information database; Based on the indication area position of the corresponding pointer of the target instrument in each collected image, the pointer deflection value of the target instrument at each collection time point is obtained, which is recorded as X t , t represents the acquisition time point number, t=1,2,......u; Will The rotation error trend of the target instrument is recorded as φ0, X t+1 It represents the pointer deflection value of the target instrument at the t+1th acquisition time point, u represents the number of acquisition time points, ΔX is the set allowable pointer deflection deviation value, is the floor rounding symbol; If the current working state of the connected device corresponding to the target instrument is idle, the collected images of the target instrument are sorted in order of collection time, the image ranked first is used as the reference image, and the other collected images are used as comparison images; If the position of the pointer area in a certain comparison image is inconsistent with that in the reference image, the comparison image is recorded as a deviation image, and the number of deviation images M0 is counted, and the number of collected images M′ is counted at the same time; Will As the rotation error trend of the target instrument, it is recorded as φ1. is the rounding sign, ΔX′ is the set allowable fluctuation deflection value, and the rotation error trend β of the target instrument is obtained, and the value of β is φ0 or φ1.

6. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 4, characterized in that: The analysis of the consistency of the current indication status of the target instrument includes: The target instrument is detected and tracked in the video using target detection and tracking algorithms. The motion trajectory of the target instrument's pointer between consecutive frames is obtained. Based on this, the number of jitters of the target instrument is analyzed, and the motion trajectory and length of each jitter are extracted. The average length of each jitter trajectory is calculated and the result is recorded as The number of jitters of the target instrument and the duration of the corresponding video are recorded as C and T respectively, and the consistency of the current indication state of the target instrument is calculated as λ. l h , C′, and k0 are the jitter trajectory length, number of jitters, and jitter frequency set for warning respectively.

7. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 6, characterized in that: The step of confirming the display compensation value of the target instrument includes: If the judgment condition A is met, (δ-δ′)*Z0 is used as the display compensation value of the target instrument, recorded as Z a , Z0 is the compensation value corresponding to the unit numerical error trend deviation of the set target instrument; If the judgment condition B is met, (β-β′)*Z1 is used as the display compensation value of the target instrument, recorded as Z b , Z1 is the compensation value corresponding to the unit rotation error trend deviation of the set target instrument; If the judgment condition C is established, analyze the display compensation value of the target instrument under the judgment condition C, and record it as Z c ; If the judgment condition A and the judgment condition B are met, Z a +Z b As the display compensation value of the target instrument, it is recorded as Z d ; If the judgment conditions A and C are met, Z a +Z c As the display compensation value of the target instrument, it is recorded as Z e ; If the judgment conditions B and C are met, Z b +Z c As the display compensation value of the target instrument, it is recorded as Z f ; If the judgment conditions A, B and C are all met, Z a +Z b +Z c As the display compensation value of the target instrument, it is recorded as Z g , and thus obtain the display compensation value Z of the target instrument η , Z η The value is Z a or Z b or Z c or Z d or Z e or Z f or Z g .

8. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 7, characterized in that: The display compensation value of the target instrument under the analysis and judgment condition C includes: Compare the motion trajectories of each jitter with each other. If the motion trajectories of each jitter are the same, As a display compensation weight; If the motion trajectory of a certain jitter is inconsistent with the motion trajectory of another jitter, the jitter trajectory deviation degree ψ is calculated. As the display compensation weight, it is recorded as In this way, the display compensation weight τ is obtained, and τ is taken as or ψ′ is the set allowable jitter trajectory deviation; Take (λ′-λ)*Z2+τ*Z′ as the display compensation value Z of the target instrument under the analysis and judgment condition C c , Z2 is the compensation value corresponding to the set target instrument corresponding to the unit indication state consistency deviation, and Z′ is the increased compensation value of the set target instrument corresponding to the unit display compensation weight.

9. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 1, wherein: The setting of the target instrument environment compensation value includes: The temperature difference, maximum humidity value and average vibration frequency value of each monitoring day are extracted from the environmental monitoring data, and the number of temperature interference monitoring days R is obtained by judging the environmental interference judgment rule. wd , humidity interference monitoring days R sd and the number of vibration interference monitoring days R zd ; The vibration frequency of each monitoring day is averaged to obtain the average vibration frequency f, and the number of monitoring days R0 is counted; Will As the target instrument environment compensation value Z ε , f′ is the set interference vibration frequency, Z υ It is the environmental compensation value of the unit environmental interference factor corresponding to the set target instrument.

10. The method for identifying and reading pointer-type instruments of an intelligent inspection robot according to claim 1, characterized in that: The calculation target meter's confirmed display reading includes: When the judgment result of step 4 is yes, Z m +Z η +Z ε Serves as confirmation display reading of target instrument; When the judgment result of step 4 is no, Z m +Z ε Serves as confirmation display reading of the target meter.

Images