CN113536895A - Disc pointer meter identification method - Google Patents
Disc pointer meter identification method Download PDFInfo
- Publication number
- CN113536895A CN113536895A CN202110592226.5A CN202110592226A CN113536895A CN 113536895 A CN113536895 A CN 113536895A CN 202110592226 A CN202110592226 A CN 202110592226A CN 113536895 A CN113536895 A CN 113536895A
- Authority
- CN
- China
- Prior art keywords
- image
- template
- pointer
- matching
- disc
- 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 28
- 239000011159 matrix material Substances 0.000 claims abstract description 17
- 230000009466 transformation Effects 0.000 claims abstract description 15
- 238000007689 inspection Methods 0.000 claims abstract description 8
- 230000001131 transforming effect Effects 0.000 claims abstract description 4
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention relates to a disc pointer meter identification method. The method comprises the following steps: step 1, moving the inspection robot to a prefabricated point position, and collecting the current meter image in real time. Step 2, performing feature matching on the collected images and corresponding images in the template, and calculating pixel offset difference of the two images; and calculating the horizontal and longitudinal adjusting angles according to the relation between the pixel difference and the adjusting angle under the current focal length, and feeding back to the robot to perform corresponding control and adjustment on the holder. And 3, matching according to the characteristic points between the collected meter image and the corresponding template, then obtaining and calculating a transformation matrix through the corresponding matching points, transforming the collected image, and fitting a disc in the image into a circle through Hough transformation. And 4, unfolding the disc image into a long rectangular image, inputting the long rectangular image into a pointer identification model, and calculating a pointer value. The invention can be applied to various disc pointer meters and has universality.
Description
Technical Field
The invention belongs to the technical field of meter identification based on images, and particularly relates to a disc pointer meter identification method.
Background
Through decades of development, the field of computer vision is rapidly improved, and the computer vision system has superiority in many application scenes. The reading of the pointer instrument equipment of the transformer substation is an important index related to the safety maintenance of the transformer substation equipment, and at present, the image acquisition of the power instrument can be remotely carried out through the inspection robot, and then the reading of the instrument is identified through a related image identification algorithm.
The conventional method for identifying the meter reading based on the inspection robot at present comprises the steps of detecting a disc and a pointer in an image through Hough transformation after a camera on the inspection robot captures a meter image, calculating the degree of an included angle between the pointer and a starting point, and finally converting the reading pointed by the current pointer.
According to the method, Hough transformation is required to be carried out on an original collected image to detect a disc and a pointer, but in an actual situation, due to the limitations of the pose of a cloud deck of the inspection robot and the like, the collected meter image is not a perfect circle; meanwhile, different starting scale marks may exist for different types of disc meters, and the traditional algorithm cannot be used universally.
Disclosure of Invention
The technical problem of the invention is mainly solved by the following technical scheme:
a disc pointer meter identification method is characterized by comprising
Step 1, moving the inspection robot to a prefabricated point position, and collecting the current meter image in real time.
Step 2, performing feature matching on the collected images and corresponding images in the template, and calculating pixel offset difference of the two images; and calculating the horizontal and longitudinal adjusting angles according to the relation between the pixel difference and the adjusting angle under the current focal length, and feeding back to the robot to perform corresponding control and adjustment on the holder.
And 3, matching according to the characteristic points between the collected meter image and the corresponding template, then obtaining and calculating a transformation matrix through the corresponding matching points, transforming the collected image, and fitting a disc in the image into a circle through Hough transformation.
And 4, unfolding the disc image into a long rectangular image.
And 5, inputting the long rectangular image into the pointer identification model, and calculating a pointer value.
In the above method for identifying a disc pointer meter, in step 2, the feature matching is performed based on a template matching algorithm, specifically: the template slides from left to right and from top to bottom on the image to be detected, and the template refers to the image which is collected under the current focal distance and is just opposite to the dial plate;
and when a pixel point is reached, capturing an image with the same size as the template from the source image by taking the pixel point as the top left corner vertex, carrying out pixel comparison operation on the image and the template, and calculating the pixel offset difference R (x, y) of the two images. Let the resulting matrix be R (x, y), the template image matrix be T (x ', y'), and the source image matrix be I (x, y).
In the above method for identifying a disc pointer meter, in step 2, the specific moving mode of the template is as follows: starting from the upper left corner of the source image, moving from left to right and from top to bottom by taking the upper left corner pixel point of the template as a unit each time, and acquiring a pixel offset difference R (x, y) based on the following formula:
in the above identification method for the disc pointer meter, in step 2, the first matching is rough matching, and multiple times of matching are required to ensure the final matching accuracy. And when the pixel offset difference between the two images is smaller than the threshold value, entering the next focal length, acquiring the images again, calculating the pixel offset between the acquired images and the current focal length template based on the template matching algorithm, and feeding back to the robot to adjust the cradle head until obtaining the high-definition image containing the dial plate at the final focal length.
In the above method for identifying a disc pointer meter, in step 3, matching is performed according to feature points between a meter image acquired at a final focal length and a corresponding template, then a transformation matrix is obtained and calculated through the corresponding matching points, the corresponding template refers to a dial image acquired at the final focal length, the acquired dial image is photographed in a forward direction, and after the acquired image is transformed, a disc in the image is fitted into a circle through hough transformation.
In the disk pointer meter identification method, in step 5, the long rectangular image is input into a pointer identification model, the position of the pointer is identified based on a pointer identification algorithm, and the pointer value is calculated according to the proportion of the abscissa of the pointer point in the length of the whole rectangular image.
In the above identification method for a disc pointer meter, the specific calculation method in step 5 is: defining the meter measuring range as range, and finally calculating the pointer numerical digit (tx/width) range according to the proportion of the abscissa (tx) of the pointer point in the length (width) of the whole rectangular image.
Therefore, the invention has the following advantages: 1. the universal disc pointer meter can be suitable for various disc pointer meters and has universality; 2. the pointer is identified by introducing a deep learning method, so that the accuracy and the robustness can be improved to a greater extent. 3. Through disc fitting, the accuracy of the unfolded image is ensured, and the accuracy of the final detection result can be improved; 4. compared with the traditional algorithm, the method not only increases the identification accuracy, but also improves the detection efficiency.
Drawings
FIG. 1 is a schematic flow chart of the algorithm of the present invention;
FIG. 2 is a collected disk pointer meter image;
FIG. 3 is an image after a disk fit;
FIG. 4 is a disk unfolded to a rectangular bar image;
fig. 5 shows the position of the recognition pointer, wherein the red arrow is the recognition result.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
Example (b):
the traditional identification algorithm for the dial gauges is basically identified by determining the included angle between a pointer and an initial position, different types of dial gauges cannot be used universally, different parameters need to be manually input to achieve the identification algorithm, and inconvenience is increased.
The specific implementation mode of the method provided by the invention is as follows:
the first step is as follows: and moving the inspection robot to a prefabricated point position, and acquiring the current meter image in real time.
The second step is that: the method comprises the steps of performing feature matching on an acquired image and a corresponding image in a template, extracting matching points in the acquired image and the corresponding image in the template by an AKAZE feature matching method, eliminating invalid matching points by a RANSAC algorithm, and calculating the pixel offset difference of the two images based on the template matching algorithm: the template (at the current focal length, the collected image just facing the dial) slides from left to right and from top to bottom on the image to be detected, the template moves from left to right and from top to bottom by taking a pixel point at the upper left corner of the template as a unit every time from the upper left corner of the source image, and when one pixel point is reached, the pixel point is taken as the top left corner vertex to intercept the image with the same size as the template from the source image and carry out pixel comparison operation with the template, and the pixel offset difference R (x, y) of the two images is calculated. Let the resulting matrix be R (x, y), the template image matrix be T (x ', y'), the source image matrix be I (x, y)
And calculating the horizontal and longitudinal adjusting angles according to the relation between the pixel difference and the adjusting angle under the current focal length, and feeding back to the robot to perform corresponding control and adjustment on the holder. The first matching is rough matching, and multiple times of matching are needed to ensure the final matching accuracy. Therefore, when the pixel offset difference between the two images is smaller than the threshold value, the next focal distance is entered, the images are collected again, matching points in the collected images and the corresponding images in the template are extracted through an AKAZE feature matching method, invalid matching points are eliminated through a RANSAC algorithm, the pixel offset between the collected images and the current focal distance template is calculated based on the template matching algorithm, and the pixel offset is fed back to the robot adjusting holder until a high-definition image containing the dial plate under the final focal distance ((three different focal distances in total) is obtained.
The third step: matching according to characteristic points (such as sift and other characteristics) between a meter image acquired to a final focal length and a corresponding template (namely, under the final focal length, the acquired dial image is shot positively), then obtaining and calculating a transformation matrix (E (essential matrix) through the corresponding matching points, wherein the transformation matrix reflects the relation between the representation of image points of a space point P in a camera coordinate system under cameras with different visual angles), transforming the acquired image, and fitting a disc in the image into a circle through Hough transformation.
The fourth step: the disk image is expanded to a long rectangular image.
The fifth step: inputting the long strip rectangular image into a pointer identification model, identifying the position of a pointer based on a pointer identification algorithm, and calculating the pointer value according to the proportion of the abscissa of the pointer point in the length of the whole rectangular image. (assuming that the measuring range of the meter is range, and finally, calculating the numerical value (tx/width) range of the pointer according to the proportion of the abscissa (tx) of the pointer point in the length (width) of the whole rectangular image.)
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (7)
1. A disc pointer meter identification method is characterized by comprising
Step 1, moving the inspection robot to a prefabricated point position, and collecting a current meter image in real time;
step 2, performing feature matching on the collected image and the corresponding image in the template, extracting matching points in the collected image and the corresponding image in the template by an AKAZE feature matching method, eliminating invalid matching points by a RANSAC algorithm, and calculating a pixel offset difference of the two images; calculating the horizontal and longitudinal adjusting angles according to the relation between the pixel difference and the adjusting angle under the current focal length, and feeding back to the robot to perform corresponding control adjustment on the holder;
step 3, matching according to the collected characteristic points between the meter image and the corresponding template, then obtaining and calculating a transformation matrix through the corresponding matching points, transforming the collected image, and fitting a disc in the image into a circle through Hough transformation;
step 4, unfolding the disc image into a long rectangular image;
and 5, inputting the long rectangular image into the pointer identification model, and calculating a pointer value.
2. The disc pointer meter identification method according to claim 1, wherein in step 2, feature matching is performed based on a template matching algorithm, specifically: sliding a template from left to right and from top to bottom on an image to be detected, wherein the template refers to an image which is collected under the current focal length and is just opposite to a dial plate;
when a pixel point is reached, an image with the same size as the template is intercepted from a source image by taking the pixel point as the top left corner, and the image and the template are subjected to pixel comparison operation, and the pixel offset difference R (x, y) of the two images is calculated; let the resulting matrix be R (x, y), the template image matrix be T (x ', y'), and the source image matrix be I (x, y).
3. The method for identifying a disc pointer meter according to claim 1, wherein in the step 2, the specific moving mode of the template is as follows: starting from the upper left corner of the source image, moving from left to right and from top to bottom by taking the upper left corner pixel point of the template as a unit each time, and acquiring a pixel offset difference R (x, y) based on the following formula:
4. the disc pointer meter identification method according to claim 1, wherein in step 2, the first matching is a rough matching, and multiple matching is required to ensure the final matching accuracy; and therefore, when the pixel offset difference between the two images is smaller than the threshold value, entering the next focal length, acquiring the images again, extracting matching points in the acquired images and the corresponding images in the template by using an AKAZE characteristic matching method, eliminating invalid matching points by using a RANSAC algorithm, calculating the pixel offset between the acquired images and the current focal length template based on the template matching algorithm, and feeding back to the robot adjusting holder until obtaining a high-definition image containing the dial plate at the final focal length.
5. The method for identifying a disc pointer meter according to claim 1, wherein in the step 3, matching is performed according to feature points between the meter image acquired at the final focal length and a corresponding template, then a transformation matrix is obtained and calculated through the corresponding matching points, the corresponding template refers to a captured dial image which is photographed in the positive direction under the final focal length, and after the captured image is transformed, a disc in the image is fitted into a circle through Hough transformation.
6. The method of claim 1, wherein in step 5, the long rectangular image is input into a pointer recognition model (as shown in the following figure), the position of the pointer is recognized based on a pointer recognition algorithm, and the pointer value is calculated according to the proportion of the abscissa of the pointer point to the length of the whole rectangular image.
7. The disc pointer meter identification method according to claim 1, wherein the specific calculation method in step 5 is: defining the meter measuring range as range, and finally calculating the pointer numerical digit (tx/width) range according to the proportion of the abscissa (tx) of the pointer point in the length (width) of the whole rectangular image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110592226.5A CN113536895A (en) | 2021-05-28 | 2021-05-28 | Disc pointer meter identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110592226.5A CN113536895A (en) | 2021-05-28 | 2021-05-28 | Disc pointer meter identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113536895A true CN113536895A (en) | 2021-10-22 |
Family
ID=78124470
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110592226.5A Pending CN113536895A (en) | 2021-05-28 | 2021-05-28 | Disc pointer meter identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113536895A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115171091A (en) * | 2022-09-06 | 2022-10-11 | 浩鲸云计算科技股份有限公司 | Meter identification method for substation inspection |
| CN117705058A (en) * | 2023-12-13 | 2024-03-15 | 重庆市测绘科学技术研究院(重庆市地图编制中心) | Inclination monitoring method based on bubble image of level pipe |
| CN117953248A (en) * | 2024-03-27 | 2024-04-30 | 昭工表面制品（深圳）有限公司 | Watch dial precision detection method based on template matching |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107042511A (en) * | 2017-03-27 | 2017-08-15 | 国机智能科技有限公司 | The inspecting robot head method of adjustment of view-based access control model feedback |
| CN107609557A (en) * | 2017-08-24 | 2018-01-19 | 华中科技大学 | A kind of readings of pointer type meters recognition methods |
| CN109711400A (en) * | 2018-11-06 | 2019-05-03 | 华中科技大学 | A kind of electric inspection process method and apparatus identifying simulated pointer formula meter reading |
| CN109993166A (en) * | 2019-04-03 | 2019-07-09 | 同济大学 | The readings of pointer type meters automatic identifying method searched based on scale |
| CN110245654A (en) * | 2018-11-20 | 2019-09-17 | 浙江大华技术股份有限公司 | Total indicator reading recognition methods and device, computer installation and readable storage medium storing program for executing |
| CN111310753A (en) * | 2019-12-27 | 2020-06-19 | 广东电网有限责任公司电力科学研究院 | Meter alignment method and device |
| CN111814919A (en) * | 2020-08-31 | 2020-10-23 | 江西小马机器人有限公司 | Instrument positioning and identifying system based on deep learning |
| CN111950330A (en) * | 2019-05-16 | 2020-11-17 | 杭州测质成科技有限公司 | Pointer instrument indicating number detection method based on target detection |
-
2021
- 2021-05-28 CN CN202110592226.5A patent/CN113536895A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107042511A (en) * | 2017-03-27 | 2017-08-15 | 国机智能科技有限公司 | The inspecting robot head method of adjustment of view-based access control model feedback |
| CN107609557A (en) * | 2017-08-24 | 2018-01-19 | 华中科技大学 | A kind of readings of pointer type meters recognition methods |
| CN109711400A (en) * | 2018-11-06 | 2019-05-03 | 华中科技大学 | A kind of electric inspection process method and apparatus identifying simulated pointer formula meter reading |
| CN110245654A (en) * | 2018-11-20 | 2019-09-17 | 浙江大华技术股份有限公司 | Total indicator reading recognition methods and device, computer installation and readable storage medium storing program for executing |
| CN109993166A (en) * | 2019-04-03 | 2019-07-09 | 同济大学 | The readings of pointer type meters automatic identifying method searched based on scale |
| CN111950330A (en) * | 2019-05-16 | 2020-11-17 | 杭州测质成科技有限公司 | Pointer instrument indicating number detection method based on target detection |
| CN111310753A (en) * | 2019-12-27 | 2020-06-19 | 广东电网有限责任公司电力科学研究院 | Meter alignment method and device |
| CN111814919A (en) * | 2020-08-31 | 2020-10-23 | 江西小马机器人有限公司 | Instrument positioning and identifying system based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 罗伊•希尔克罗特 等: "《深入理解OpenCV:实用计算机视觉项目解析（原书第3版）》" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115171091A (en) * | 2022-09-06 | 2022-10-11 | 浩鲸云计算科技股份有限公司 | Meter identification method for substation inspection |
| CN117705058A (en) * | 2023-12-13 | 2024-03-15 | 重庆市测绘科学技术研究院(重庆市地图编制中心) | Inclination monitoring method based on bubble image of level pipe |
| CN117953248A (en) * | 2024-03-27 | 2024-04-30 | 昭工表面制品（深圳）有限公司 | Watch dial precision detection method based on template matching |
| CN117953248B (en) * | 2024-03-27 | 2024-06-11 | 昭工表面制品（深圳）有限公司 | Watch dial precision detection method based on template matching |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112906694B (en) | Reading correction system and method for transformer substation inclined pointer instrument image | |
| CN105740856B (en) | A kind of pointer instrument registration read method based on machine vision | |
| CN113536895A (en) | Disc pointer meter identification method | |
| CN107229930B (en) | Intelligent identification method for numerical value of pointer instrument | |
| CN102521560B (en) | Instrument pointer image identification method of high-robustness rod | |
| CN107993258B (en) | Image registration method and device | |
| CN106886216B (en) | Robot automatic tracking method and system based on RGBD face detection | |
| US20110025834A1 (en) | Method and apparatus of identifying human body posture | |
| CN103149087B (en) | Follow-up window and digital image-based non-contact real-time strain measurement method | |
| CN111624203B (en) | Relay contact point alignment non-contact measurement method based on machine vision | |
| Mi et al. | A vision-based displacement measurement system for foundation pit | |
| CN110879080A (en) | High-precision intelligent measuring instrument and measuring method for high-temperature forge piece | |
| CN104899609A (en) | Image registration-based mechanical meter identification method | |
| CN103729631A (en) | Vision-based connector surface feature automatically-identifying method | |
| CN113688817A (en) | Instrument identification method and system for automatic inspection | |
| CN110991360A (en) | Robot inspection point location intelligent configuration method based on visual algorithm | |
| CN111563896A (en) | Image processing method for catenary anomaly detection | |
| CN109540918B (en) | Hot-rolled coil edge defect detection device and method | |
| CN109086763A (en) | A kind of pointer instrument read method and device | |
| CN111832760B (en) | Automatic inspection method for well lid based on visual algorithm | |
| CN113607058B (en) | Straight blade size detection method and system based on machine vision | |
| CN109410272B (en) | Transformer nut recognition and positioning device and method | |
| CN109359637B (en) | Pointer instrument value reading method based on machine vision | |
| CN116818233A (en) | High-precision bridge state monitoring method and system adopting laser and machine vision | |
| CN115861407A (en) | Safe distance detection method and system based on deep learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211022 |
|
| RJ01 | Rejection of invention patent application after publication |