Search

CN-122023766-A - Instrument identification method, industrial equipment identification method and system thereof

CN122023766ACN 122023766 ACN122023766 ACN 122023766ACN-122023766-A

Abstract

The invention provides an instrument identification method, which is a method for determining a dynamic anchor point through prior information and target information of a patrol area, wherein the prior information comprises at least 2 pieces of prior information of instrument frames, the prior information of the patrol area comprises position information, a center point or at least two groups of reference coordinates of each piece of prior information of the instrument frames, and the target information of the patrol area comprises the position information, the center point or at least two groups of reference coordinates of each piece of target information of the instrument frames. The dynamic anchor point is used for selecting to replace global offset estimation, so that the integral matching is prevented from being influenced by single errors, 100% of accuracy is still maintained under the condition that 1/3 of the instrument is missing, the environmental robustness is enhanced, the global offset, random offset of one instrument and the newly added instrument scene are supported, the identification accuracy of a plurality of instruments of the same type or different types is improved, and the time difference and the efficiency between the inspection data of the instruments are greatly increased.

Inventors

  • CAI YONGFENG
  • WANG SHIXING
  • QIU YAO
  • HUANG YULONG
  • XIONG WENBO
  • LIU TAO

Assignees

  • 杭州安森智能信息技术有限公司
  • 西安安森智能仪器股份有限公司

Dates

Publication Date
20260512
Application Date
20251231

Claims (17)

  1. 1. The instrument identification method is characterized by comprising a method for determining a dynamic anchor point according to prior information of a patrol area and target information thereof, wherein the prior information comprises prior information of at least 2 instrument frames; the patrol area prior information comprises: position information, a central point or at least two groups of reference coordinates of each prior information instrument frame; the patrol area target information comprises: Position information, center point, or at least two sets of reference coordinates for each target information instrument box.
  2. 2. The meter identification method of claim 1, wherein the method of determining a dynamic anchor point comprises: selecting a reference candidate combination with the minimum comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points; or alternatively; Selecting a reference candidate combination with the minimum comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points, wherein the comprehensive errors meet a preset comprehensive error threshold; or alternatively; Selecting a reference candidate combination which has the smallest comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points and meets the preset screening condition; or alternatively; Selecting a reference candidate combination which has the smallest comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points and meets the preset screening condition, wherein the comprehensive error meets the preset comprehensive error threshold; The selected reference candidate combination is the dynamic anchor point.
  3. 3. The meter identification method according to claim 2, wherein the calculation method of the integrated error of the center point or at least two sets of reference coordinate points of all the reference candidate combination meter frames is as follows: Calculating self errors of the central points of the reference candidate combination instrument frames or at least two groups of reference coordinate points in all the reference candidate combinations respectively, and calculating matching average errors of the central points of the other matching candidate combination instrument frames or at least two groups of reference coordinate points respectively; and the comprehensive error of each reference candidate combination instrument frame center point or at least two groups of reference coordinate points is the sum of the self errors of the reference candidate combination instrument frame center points or at least two groups of reference coordinate points and the matching average errors of the other matching candidate combination instrument frame center points or at least two groups of reference coordinate points by adopting a preset weighting coefficient.
  4. 4. The meter identification method according to claim 3, wherein the calculation method of the self error of the center point of the reference candidate combination meter frame or at least two sets of reference coordinate points is as follows: Firstly, calculating a relative position vector difference of a central point of the reference candidate combination instrument frame, and then calculating the self error of the central point of the reference candidate combination instrument frame according to the square sum of two coordinate values of the relative position vector difference and a root number; Or alternatively Firstly, respectively calculating the relative position vector difference of at least two groups of reference coordinate points of the reference candidate combination instrument frame, then respectively calculating the single-point self error of the center point of the reference candidate combination instrument frame by the square sum and the root number of two coordinate values of the relative position vector difference, and then calculating the sum of the single-point self errors of the at least two groups of reference coordinate points by adopting a preset weight coefficient, namely the self errors of the at least two groups of reference coordinate points of the reference candidate combination instrument frame.
  5. 5. The meter identification method according to claim 4, wherein the calculation method of the average error of the matching of the center point or at least two sets of reference coordinate points of the other matching candidate combination meter frame is as follows: Based on the reference candidate combination, predicting theoretical positions of the reference candidate combination instrument frame according to relative position vector differences between the center points of other prior information instrument frames and the center points of the reference candidate combination instrument frame one by one, respectively matching the reference candidate combination instrument frame with the nearest instrument frame in the target information instrument frame to respectively form a plurality of other matching candidate combinations, and respectively calculating matching errors of the center points of the plurality of other matching candidate combination instrument frames; Or alternatively Based on the reference candidate combination, predicting theoretical positions of the reference candidate combination instrument frames one by one according to relative position vector differences between center points of other prior information instrument frames and center points of the reference candidate combination instrument frames, respectively matching the reference candidate combination instrument frames with nearest instrument frames in a target information instrument frame to respectively form a plurality of other matching candidate combinations, respectively calculating matching errors between at least two groups of reference coordinate points of the plurality of other matching candidate combination instrument frames, and carrying out weighted average on the matching errors between at least two groups of reference coordinate points of all the other matching candidate combination instrument frames to obtain matching average errors based on at least two groups of reference coordinate points of the other matching candidate combination instrument frames.
  6. 6. The instrument identification method according to claim 2,3, 4 or 5, wherein the inspection area prior information further comprises category information and/or outline dimensions of each prior information instrument frame, the inspection area target information further comprises category information and/or outline dimensions of each target information instrument frame, and the preset screening conditions comprise size verification and/or category verification.
  7. 7. The meter identification method of claim 6, wherein the dimension verification is that a difference between the two meter frame physical dimensions of the reference candidate combination is no greater than 20% of its a priori information meter frame physical dimensions.
  8. 8. The meter identification method of claim 6 wherein the category verification is that the categories of the two meter frames of the reference candidate combination are the same.
  9. 9. The meter identification method of claim 3 or 4 or 5 or 7 or 8, further comprising: Judging whether the matching error values of other matching candidate combinations corresponding to the dynamic anchor points meet a preset matching error threshold value or not, if not, canceling the other matching candidate combinations, and outputting an instrument matching abnormal signal; or/and the combination of the two, Judging whether the difference value of the appearance sizes of other matching candidate combinations corresponding to the dynamic anchor points is not more than 20% of the appearance sizes of the prior information instrument frames, if not, canceling the other matching candidate combinations, and outputting an instrument size abnormal signal; or/and the combination of the two, Judging whether the categories of other matching candidate combinations corresponding to the dynamic anchor points are consistent or not, if not, canceling the other matching candidate combinations, and outputting instrument category abnormal signals.
  10. 10. The meter identification method of claim 6, further comprising: Judging whether the matching error values of other matching candidate combinations corresponding to the dynamic anchor points meet a preset matching error threshold value or not, if not, canceling the other matching candidate combinations, and outputting an instrument matching abnormal signal; or/and the combination of the two, Judging whether the difference value of the appearance sizes of other matching candidate combinations corresponding to the dynamic anchor points is not more than 20% of the appearance sizes of the prior information instrument frames, if not, canceling the other matching candidate combinations, and outputting an instrument size abnormal signal; or/and the combination of the two, Judging whether the categories of other matching candidate combinations corresponding to the dynamic anchor points are consistent or not, if not, canceling the other matching candidate combinations, and outputting instrument category abnormal signals.
  11. 11. The meter identification method of claim 1 or 2 or 3 or 4 or 5 or 7 or 8 or 10, wherein the inspection area prior information further comprises a meter value of each prior information meter box, and the inspection area target information further comprises a meter value of each target information meter box.
  12. 12. The instrument identification method according to claim 6, wherein the inspection area prior information further comprises an instrument value of each prior information instrument box, and the inspection area target information further comprises an instrument value of each target information instrument box.
  13. 13. The meter identification method of claim 9 wherein the inspection area prior information further comprises a meter value for each prior information meter box and the inspection area target information further comprises a meter value for each target information meter box.
  14. 14. A meter identification method for an industrial device identification method with a visually identifiable status identification or parameter display structure based on any of claims 1-13, the meter being replaced with an industrial device with a visually identifiable status identification or parameter display structure, characterized in that the prior information is prior information comprising at least 2 pieces of industrial devices of the same or different kind with a visually identifiable status identification or parameter display structure; the patrol area prior information comprises: position information, a center point or at least two sets of reference coordinates of each piece of prior information industrial equipment; the patrol area target information comprises: location information, a center point, or at least two sets of reference coordinates for each piece of target information industry equipment.
  15. 15. The method for identifying industrial equipment according to claim 14, wherein the inspection area prior information further comprises a state identifier or a parameter display value of each piece of prior information industrial equipment, and the inspection area target information further comprises a state identifier or a parameter display value of each piece of target information industrial equipment.
  16. 16. The system based on the instrument identification method of any one of claims 1-13, which is characterized by comprising a motion positioning navigation unit, a camera acquisition control unit, a data storage unit, a main control calculation unit and a matching identification result output unit, wherein: the motion positioning navigation unit is used for controlling the movement of the inspection equipment according to a preset inspection path, and sending an image acquisition trigger control signal to the camera acquisition control unit after the inspection equipment accurately reaches a preset instrument inspection point; The camera shooting acquisition control unit is used for triggering the main control calculation unit to judge whether prior information of the inspection area exists in the data storage unit when receiving an image acquisition triggering control signal of the motion positioning navigation unit, if not, carrying out image acquisition of the inspection area according to preset image acquisition parameters, and if so, carrying out image acquisition of the inspection area according to first image acquisition parameters, transmitting the acquired image of the inspection area to the main control calculation unit, and triggering the main control calculation unit to carry out validity check on the image of the inspection area: When the validity check result of the target detection module of the main control computing unit is an invalid image, the image re-acquisition signal fed back by the main control computing unit triggers the camera acquisition control unit to re-adjust the focal length and multiplying power of the camera, and re-acquire the patrol area image; the main control computing unit comprises a target detection module and a multi-table matching module, wherein, The target detection module is used for analyzing and processing the image transmitted by the camera acquisition control unit to obtain prior information or target information of the inspection area, determining whether an instrument exists in the image of the inspection area according to whether the prior information or the target information contains the position coordinates of the instrument, and checking the validity of the image of the inspection area; The multi-table matching module is used for simultaneously reading the prior information of the first inspection area corresponding to the data storage unit according to the target information output by the target detection module, and determining a dynamic anchor point and other corresponding matching candidate combinations, or verifying the dynamic anchor point and other corresponding matching candidate combinations and other matching candidate combinations corresponding to the dynamic anchor point candidate combinations, so as to obtain a matching identification result and storing the matching identification result to the data storage unit; The data storage unit is used for storing the following data, namely an effective inspection area image output by the camera shooting acquisition control unit, an ineffective inspection area image with an abnormal mark output by the main control calculation unit, first inspection area prior information, non-first inspection area target information and a matching identification result output by the main control calculation unit; and the matching recognition result output unit is used for pushing the matching recognition result in the data storage unit to the remote system platform.
  17. 17. A system based on the industrial equipment identification method with the visual identification state identification or parameter display structure in claim 14 or 15 is characterized by comprising a motion positioning navigation unit, a camera acquisition control unit, a data storage unit, a main control calculation unit and a matching identification result output unit, wherein: The motion positioning navigation unit is used for controlling the movement of the inspection equipment according to a preset inspection path, and sending an image acquisition trigger control signal to the camera acquisition control unit after the inspection equipment accurately reaches an inspection point of the preset industrial equipment; The camera shooting acquisition control unit is used for triggering the main control calculation unit to judge whether prior information of the inspection area exists in the data storage unit when receiving an image acquisition triggering control signal of the motion positioning navigation unit, if not, carrying out image acquisition of the inspection area according to preset image acquisition parameters, and if so, carrying out image acquisition of the inspection area according to first image acquisition parameters, transmitting the acquired image of the inspection area to the main control calculation unit, and triggering the main control calculation unit to carry out validity check on the image of the inspection area: When the validity check result of the target detection module of the main control computing unit is an invalid image, the image re-acquisition signal fed back by the main control computing unit triggers the camera acquisition control unit to re-adjust the focal length and multiplying power of the camera, and re-acquire the patrol area image; the main control computing unit comprises a target detection module and a multi-industry equipment matching module, wherein, The target detection module is used for analyzing and processing the inspection area image transmitted by the camera acquisition control unit to obtain priori information or target information of the inspection area, determining whether industrial equipment exists in the inspection area image according to whether the priori information or the target information contains the position coordinates of the industrial equipment, and checking the validity of the inspection area image, wherein if the priori information or the target information exists, the priori information or the target information of the inspection area is respectively output to the data storage unit and the multi-industrial equipment matching module; The multi-industrial equipment matching module is used for simultaneously reading the prior information of the first inspection area corresponding to the data storage unit according to the target information output by the target detection module, and determining a dynamic anchor point and other corresponding matching candidate combinations, or verifying the dynamic anchor point and the other corresponding matching candidate combinations and the other matching candidate combinations corresponding to the dynamic anchor point candidate combinations to obtain a matching identification result and storing the matching identification result to the data storage unit; The data storage unit is used for storing the following data, namely an effective inspection area image output by the camera shooting acquisition control unit, an ineffective inspection area image with an abnormal mark output by the main control calculation unit, first inspection area prior information, non-first inspection area target information and a matching identification result output by the main control calculation unit; and the matching recognition result output unit is used for pushing the matching recognition result in the data storage unit to the remote system platform.

Description

Instrument identification method, industrial equipment identification method and system thereof Technical Field The invention belongs to the technical field of computer vision, and particularly relates to an instrument identification method, an industrial equipment identification method and a system thereof. Background In the prior art, the related meters are required to be shot for a plurality of times one by one and respectively identified, then result storage is carried out, so that the time difference between meter inspection data is greatly increased, accurate meter inspection is not facilitated, and in addition, in the prior meter matching technology, only the position of the meter or the surface characteristic of the meter is matched, so that mismatching among different types of meters is caused. Disclosure of Invention In order to solve the problems in the prior art, the invention provides an instrument identification method, an industrial equipment identification method and a system thereof. The instrument identification method comprises a method for determining a dynamic anchor point according to prior information of a patrol area and target information thereof, wherein the prior information comprises prior information of at least 2 instrument frames; the patrol area prior information comprises: position information, a central point or at least two groups of reference coordinates of each prior information instrument frame; the patrol area target information comprises: Position information, center point, or at least two sets of reference coordinates for each target information instrument box. Preferably, the method for determining the dynamic anchor point includes: selecting a reference candidate combination with the minimum comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points; or alternatively; Selecting a reference candidate combination with the minimum comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points, wherein the comprehensive errors meet a preset comprehensive error threshold; or alternatively; Selecting a reference candidate combination which has the smallest comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points and meets the preset screening condition; or alternatively; Selecting a reference candidate combination which has the smallest comprehensive error in the comprehensive errors of all reference candidate combination instrument frame center points or at least two groups of reference coordinate points and meets the preset screening condition, wherein the comprehensive error meets the preset comprehensive error threshold; The selected reference candidate combination is the dynamic anchor point. Preferably, the calculation method of the integrated error of the center point of the frame of all the reference candidate combination instrument or at least two groups of reference coordinate points comprises the following steps: Calculating self errors of the central points of the reference candidate combination instrument frames or at least two groups of reference coordinate points in all the reference candidate combinations respectively, and calculating matching average errors of the central points of the other matching candidate combination instrument frames or at least two groups of reference coordinate points respectively; and the comprehensive error of each reference candidate combination instrument frame center point or at least two groups of reference coordinate points is the sum of the self errors of the reference candidate combination instrument frame center points or at least two groups of reference coordinate points and the matching average errors of the other matching candidate combination instrument frame center points or at least two groups of reference coordinate points by adopting a preset weighting coefficient. Preferably, the calculation method of the self error of the center point of the reference candidate combination instrument frame or at least two groups of reference coordinate points comprises the following steps: Firstly, calculating a relative position vector difference of a central point of the reference candidate combination instrument frame, and then calculating the self error of the central point of the reference candidate combination instrument frame according to the square sum of two coordinate values of the relative position vector difference and a root number; Or alternatively Firstly, respectively calculating the relative position vector difference of at least two groups of reference coordinate points of the reference candidate combination instrument frame, then respectively calculating the single-point self e