CN113888741A - Method for correcting rotating image of instrument in power distribution room - Google Patents
Method for correcting rotating image of instrument in power distribution room Download PDFInfo
- Publication number
- CN113888741A CN113888741A CN202111471700.5A CN202111471700A CN113888741A CN 113888741 A CN113888741 A CN 113888741A CN 202111471700 A CN202111471700 A CN 202111471700A CN 113888741 A CN113888741 A CN 113888741A
- Authority
- CN
- China
- Prior art keywords
- image
- correction
- instrument
- rotated
- target detection
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012937 correction Methods 0.000 claims abstract description 52
- 238000001514 detection method Methods 0.000 claims abstract description 39
- 238000011156 evaluation Methods 0.000 claims abstract description 18
- 238000002372 labelling Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 8
- 230000000694 effects Effects 0.000 abstract description 4
- 238000012423 maintenance Methods 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 abstract description 2
- 238000003702 image correction Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/60—Rotation of whole images or parts thereof
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of intelligent operation and maintenance of power transformation, in particular to a method for correcting a rotating image of an instrument in a power distribution room. The method realizes the rotation correction of the whole image by the rotation correction of the target object in the image, generates multi-angle copies based on the rotation of the whole image, adopts a target detection network to complete the multi-target structured information prediction in the copies in batch, adopts weighted average to calculate the evaluation value, takes the copy corresponding to the minimum value as a correction source, judges the direction of the correction source through the length-width ratio, and realizes the rotation correction of the image by reversely rotating 90 degrees if the directions are inconsistent. The correction method is suitable for the targets which are shot images and rotate at a small angle (plus or minus 45 degrees), have no obvious distortion and have clear rectangular outlines in scene images, and can reduce the difficulty of algorithm processing and improve the algorithm processing effect.
Description
Technical Field
The invention relates to the technical field of intelligent operation and maintenance of power transformation, in particular to a method for correcting a rotating image of an instrument in a power distribution room.
Background
Targets such as various meters, switches and indicator lamps in the power distribution room need uninterrupted monitoring and state recognition, so that various devices in the power distribution room can operate stably. At present, a common mode is to continuously acquire images of a monitored object by means of an image acquisition device, and then realize detection and identification of the monitored object by the aid of a computer vision field technology.
Due to environmental changes such as wall vibration, artificial touch, turntable errors and the like, the deflection of the acquisition equipment is caused, the deflection of the acquired image is further caused, the interference is formed on image processing and target identification, and the accuracy of the target identification result is influenced. At present, aiming at the rotation correction of images, the rotation angle of the whole image is firstly detected, then the reverse angle rotation correction is carried out, the learning difficulty of a related algorithm is higher, the method is mainly applied to the field of remote sensing image detection, the existing mature target detection network cannot be directly used due to the increase of angle parameters, and meanwhile, a mature use case is not found in the field provided by the method.
Disclosure of Invention
The invention aims to provide a method for correcting a rotating image of an instrument in a power distribution room, which realizes the rotating correction of the whole image by means of the rotating correction of a target object in the image and is suitable for a target which is shot in the image, rotates at a small angle (plus or minus 45 degrees), has no obvious distortion and has a clear rectangular outline in a scene image.
In order to achieve the purpose, the invention adopts the following technical scheme:
a method for correcting a rotating image of an instrument in a power distribution room comprises the following steps:
and S1, collecting the instrument image in the power distribution room, and labeling the part of the instrument image with the rectangular outer contour to obtain an instrument image labeling data set. In consideration of the fact that the targets with rectangular inner and outer outlines in the power distribution room scene are more and are easy to find, the instrument image marking method marks the parts with rectangular outer outlines in the instrument image. And the rectangular target is detected, and when the rotation angle is 0, the detection result is the actual contour of the target, so that the calculation is more convenient.
S2, constructing a target detection network model, and training the target detection network model by using the instrument image labeling data set to obtain the trained target detection network model.
S3, setting the rotation step length as d, and sequentially rotating the image to be detected by d degrees in the anticlockwise or clockwise direction until the image to be detected is rotated by 90 degrees to obtain 90/d rotated image copies; wherein the value of d is 1-90 degrees. And setting an image rotation interval angle in the range of 0-90 degrees, namely a rotation step length d, sequentially rotating the image to be detected by d degrees in the range of 0-90 degrees according to the rotation step length d until the image to be detected is rotated by 90 degrees, and further obtaining 90/d rotation image copies.
And S4, inputting the 90/d rotating image copies into the trained target detection network model, and outputting the structural information of each rotating image copy by the target detection network model. The structured information, i.e. the detection result, mainly includes: rectangular areas (x, y, width, height), categories, similarities, etc.
S5, calculating an evaluation value of the structured information of each rotated image copy using the following formula:
where Z is the weighted average of the structured information of the rotated image copy, N is the number of objects with rectangular outlines in the rotated image copy, ConfiConfidence of the ith object with a rectangular outer contour, ConffixAs confidence threshold, ConffixTaking decimal fraction between 0 and 1, PCPredetermined coefficient, P, for the class to which the ith object whose outer contour is rectangular belongsCTaking decimal fraction between 0 and 1, SiThe target outer section rectangular area with the ith outer contour being rectangular.
And S6, selecting the rotating image copy with the minimum evaluation value as a correction source. The target detection algorithm adopted by the invention comprises the following detection results: and a rectangular area which is an outer truncated rectangle, wherein the outer truncated rectangle is matched with the actual target contour only when the target rotation angle is 0, and the evaluation value is the minimum, so that the corresponding image is selected as a correction source.
And S7, judging the direction of the correction source, determining a correction image, and correcting the image to be detected according to the correction image. Specifically, the direction of a correction source is judged, an additional filling background of the correction source in the process of rotating the image to be detected to the current position is removed, an image part is reserved, an outer section rectangle of the image part is extracted, and the direction of the image is judged according to the length-width ratio of the outer section rectangle; if the image direction of the correction source is consistent with the direction of the original image, taking the correction source as an optimal copy, and correcting the image to be detected according to the optimal copy; if the directions are not consistent, the opposite rotation direction in step S3 is selected, the correction source is rotated by 90 degrees, the rotated correction source is the correction image, and the image to be detected is corrected according to the correction image.
Furthermore, the target detection network model adopts one or two-order target detection networks of YoloV3\ V4\ V5, CornerNet \ CenterNet, Fast-Cascade, Fast-RcNN \ Fast-RCNN. YoloV5, Cascade-RCNN and the like are algorithms with higher maturity in the industry at present, and the algorithms have good overall effect and are in continuous promotion.
Further, the structured information includes regions, categories, and confidence levels.
Compared with the prior art, the invention has the advantages that:
(1) the method realizes the rotation correction of the whole image by the rotation correction of the target object in the image, generates multi-angle copies based on the rotation of the whole image, adopts a target detection network to complete the multi-target structured information prediction in the copies in batch, adopts weighted average to calculate the evaluation value, takes the copy corresponding to the minimum value as a correction source, judges the direction of the correction source through the length-width ratio, and realizes the rotation correction of the image by reversely rotating 90 degrees if the directions are inconsistent. The correction method is suitable for the targets which are shot images and rotate at a small angle (plus or minus 45 degrees), have no obvious distortion and have clear rectangular outlines in scene images, and can reduce the difficulty of algorithm processing and improve the algorithm processing effect.
(2) The angle of the whole image is judged by detecting the rotation angle of the rectangular target in the image, so that the target can be found in the instrument image in the power distribution room more easily, and the detection process is simpler and more convenient. The method for detecting the rotation angle of the rectangular target in the image adopts the traversal detection of the structured information set of the rectangular target set in the multi-angle copy, and selects the corresponding rotation angle of the rotation angle copy as the required rotation angle through the evaluation value.
(3) The method and the device can complete target detection without adding additional parameters, and the image rotation is conventional operation of image processing, so that the technology is mature. The method is mainly applied to the field of remote sensing image detection, and a mature use case is not found for rotating images of instruments in a power distribution room. And the whole image is subjected to inclination identification and correction, 4 vertex coordinates are generally required to be marked, 4 points of connecting lines after marking are required to form a rectangle, and the marking difficulty is improved. The method is suitable for the rotation correction of the target with small angle, no obvious distortion and clear rectangular outline.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings:
example one
A certain distribution room acquires instrument images at regular time, and identification of instrument readings, indicator lights, on-off states and the like is further realized by instrument image correction and instrument target detection. As shown in fig. 1, the method for correcting the rotating image of the instrument in the power distribution room according to the present invention includes the following steps:
and S11, collecting the instrument image, selecting and marking the parts containing rectangular targets such as signboards, rectangular meters and the like in the instrument image, and obtaining an instrument image marking data set.
And S12, constructing a target detection network model, wherein the target detection network model in the embodiment is a CornerNet network. And (3) training the CornerNet network by using the instrument image labeling data set, and outputting a trained target detection model.
And S13, selecting the distribution room instrument image with the length of 2560 pixels and the width of 1440 pixels as an original image.
And S14, selecting 1 degree as a rotation step, and sequentially rotating the original image by 1 degree in the counterclockwise direction until the original image is rotated by 90 degrees to obtain 90 rotated image copies.
And S15, importing 90 image rotating copies in batch to the trained CornerNet network for reasoning, and outputting the structured information in batch by the CornerNet network. The structured information comprises the category, confidence and position area of a plurality of rectangular targets (signs, rectangular meters and the like), the evaluation value of each rectangular target is calculated one by adopting an evaluation formula, and Conf is takenfix=0.5, PC is taken as 1, and the evaluation formula is:
s16, the rotated image copy of the rotation angle corresponding to the minimum evaluation value is selected as the correction source, where the rotation angle corresponding to the minimum evaluation value is 4 degrees.
And S17, judging the directions of the correction source image and the original image according to the aspect ratio of the correction source image and the aspect ratio of the original image. In the present embodiment, the source image is corrected for a length 2654 pixels and a width 1615 pixels, with a pixel ratio R1=1.64 greater than 1. The length of the original image is 2560 pixels, the width of the original image is 1440 pixels, the pixel ratio R2=1.78 is larger than 1, the direction of the corrected source image is consistent with that of the original image, the corrected source image is the corrected image, the requirement is met, and the image to be detected is corrected according to the corrected image.
Example two
A certain distribution room acquires instrument images at regular time, and identification of instrument readings, indicator lights, on-off states and the like is further realized by instrument image correction and instrument target detection. The method for correcting the rotating image of the instrument in the power distribution room comprises the following steps:
and S21, collecting instrument images, selecting images containing rectangular targets such as signboards and rectangular meters from the instrument images, and labeling to obtain a meter image labeling data set.
And S22, constructing a target detection network model, wherein the target detection network model in the embodiment is a CornerNet detection network. And training a CornerNet detection network by using an instrument image annotation data set, and outputting a trained target detection model.
And S23, selecting the distribution room instrument image with the length of 2560 pixels and the width of 1440 pixels as an original input image.
And S24, selecting 1 degree as a rotation step, and rotating the original image clockwise to generate 90 rotation copy images in total by 1-90 degrees.
S25, importing images in batch into a trained CornerNet detection network, which infers and outputs structured information including types, confidences, and position areas of a plurality of targets, calculating an evaluation value of each rectangular target one by one using an evaluation formula, taking Conffix =0.5, taking PC as 1, and the evaluation formula being:
and S26, selecting the copy with the rotation angle of 90 degrees corresponding to the minimum evaluation value as a correction source.
S27, acquiring a correction source image with the length 1615 and the width 2654, wherein the pixel ratio R1=0.61 is less than 1, calculating that the pixel ratio R2=1.78 of the original image with the length 2560 and the width 1440 is greater than 1.78, and the two directions are not consistent, wherein the correction source image is an image rotated by 90 degrees, the image is corrected by rotating 90 degrees anticlockwise to meet the requirement, and the image to be detected is corrected according to the corrected image.
The method realizes the rotation correction of the whole image by the rotation correction of the target object in the image, generates multi-angle copies based on the rotation of the whole image, adopts a target detection network to complete the multi-target structured information prediction in the copies in batch, adopts weighted average to calculate the evaluation value, takes the copy corresponding to the minimum value as a correction source, judges the direction of the correction source through the length-width ratio, and realizes the rotation correction of the image by reversely rotating 90 degrees if the directions are inconsistent. The correction method is suitable for the targets which are shot images and rotate at a small angle (plus or minus 45 degrees), have no obvious distortion and have clear rectangular outlines in scene images, and can reduce the difficulty of algorithm processing and improve the algorithm processing effect.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (3)
1. A method for correcting a rotating image of an instrument in a power distribution room is characterized by comprising the following steps:
s1, collecting instrument images in the power distribution room, and labeling the parts of the instrument images with rectangular outer contours to obtain instrument image labeling data sets;
s2, constructing a target detection network model, and training the target detection network model by using the instrument image annotation data set to obtain a trained target detection network model;
s3, setting the rotation step length as d, and sequentially rotating the image to be detected by d degrees in the anticlockwise or clockwise direction until the image to be detected is rotated by 90 degrees to obtain 90/d rotated image copies; wherein the value of d is 1-90 degrees;
s4, inputting 90/d of the rotated image copies into a trained target detection network model, and outputting structural information of each rotated image copy by the target detection network model;
s5, calculating an evaluation value of the structured information of each rotated image copy using the following formula:
where Z is the weighted average of the structured information of the rotated image copy, N is the number of objects with rectangular outlines in the rotated image copy, ConfiConfidence of the ith object with a rectangular outer contour, ConffixAs confidence threshold, ConffixTaking decimal fraction between 0 and 1, PCPredetermined coefficient, P, for the class to which the ith object whose outer contour is rectangular belongsCTaking decimal fraction between 0 and 1, SiThe area of the target outer section rectangle with the ith outer contour as a rectangle;
s6, selecting the rotating image copy with the minimum evaluation value as a correction source;
s7, judging the direction of the correction source, removing the additional filling background of the correction source in the process of rotating the image to be detected to the current position, reserving the image part, extracting the outer truncated rectangle of the image part, and judging the direction of the image according to the length-width ratio of the outer truncated rectangle; if the image direction of the correction source is consistent with the direction of the original image, taking the correction source as an optimal copy, and correcting the image to be detected according to the optimal copy; if the directions are not consistent, the opposite rotation direction in step S3 is selected, the correction source is rotated by 90 degrees, the rotated correction source is the correction image, and the image to be detected is corrected according to the correction image.
2. The method for correcting the rotating image of the instrument in the power distribution room as claimed in claim 1, wherein the target detection network model adopts one or two-order target detection networks selected from the group consisting of YoloV3\ V4\ V5, CornerNet \ centrnet, Fast-Cascade, Fast-RcNN \ Fast-RcNN.
3. The method according to claim 1, wherein the structured information includes region, category and confidence level.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111471700.5A CN113888741B (en) | 2021-12-06 | 2021-12-06 | Method for correcting rotating image of instrument in power distribution room |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111471700.5A CN113888741B (en) | 2021-12-06 | 2021-12-06 | Method for correcting rotating image of instrument in power distribution room |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113888741A true CN113888741A (en) | 2022-01-04 |
CN113888741B CN113888741B (en) | 2022-03-18 |
Family
ID=79016344
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111471700.5A Active CN113888741B (en) | 2021-12-06 | 2021-12-06 | Method for correcting rotating image of instrument in power distribution room |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113888741B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107066998A (en) * | 2016-12-30 | 2017-08-18 | 山东鲁能软件技术有限公司 | A kind of pointer-type circular single instrument board real-time identification method of utilization mobile device |
CN107103587A (en) * | 2017-06-05 | 2017-08-29 | 新疆大学 | A kind of inclined bearing calibration of biochip image and device |
CN110443220A (en) * | 2019-08-13 | 2019-11-12 | 树根互联技术有限公司 | Digital table image-recognizing method, device, electronic equipment and storage medium |
US20200013371A1 (en) * | 2019-08-08 | 2020-01-09 | Lg Electronics Inc. | Method, controller, and system for adjusting screen through inference of image quality or screen content on display |
CN110852954A (en) * | 2019-11-19 | 2020-02-28 | 随锐科技集团股份有限公司 | Image tilt correction method and system for pointer instrument |
CN111681173A (en) * | 2020-05-25 | 2020-09-18 | 浙江大学 | Inclined image correction method based on combination of minimum area bounding box and projection method |
CN111814827A (en) * | 2020-06-08 | 2020-10-23 | 三峡大学 | Key point target detection method based on YOLO |
CN112464785A (en) * | 2020-11-25 | 2021-03-09 | 浙江大华技术股份有限公司 | Target detection method and device, computer equipment and storage medium |
-
2021
- 2021-12-06 CN CN202111471700.5A patent/CN113888741B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107066998A (en) * | 2016-12-30 | 2017-08-18 | 山东鲁能软件技术有限公司 | A kind of pointer-type circular single instrument board real-time identification method of utilization mobile device |
CN107103587A (en) * | 2017-06-05 | 2017-08-29 | 新疆大学 | A kind of inclined bearing calibration of biochip image and device |
US20200013371A1 (en) * | 2019-08-08 | 2020-01-09 | Lg Electronics Inc. | Method, controller, and system for adjusting screen through inference of image quality or screen content on display |
CN110443220A (en) * | 2019-08-13 | 2019-11-12 | 树根互联技术有限公司 | Digital table image-recognizing method, device, electronic equipment and storage medium |
CN110852954A (en) * | 2019-11-19 | 2020-02-28 | 随锐科技集团股份有限公司 | Image tilt correction method and system for pointer instrument |
CN111681173A (en) * | 2020-05-25 | 2020-09-18 | 浙江大学 | Inclined image correction method based on combination of minimum area bounding box and projection method |
CN111814827A (en) * | 2020-06-08 | 2020-10-23 | 三峡大学 | Key point target detection method based on YOLO |
CN112464785A (en) * | 2020-11-25 | 2021-03-09 | 浙江大华技术股份有限公司 | Target detection method and device, computer equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
邱金红 等: ""一种圆形指针式仪表倾斜检测方法"", 《自动化与信息工程》 * |
Also Published As
Publication number | Publication date |
---|---|
CN113888741B (en) | 2022-03-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112906694B (en) | Reading correction system and method for transformer substation inclined pointer instrument image | |
CN111950330B (en) | Pointer instrument indication detection method based on target detection | |
CN109948469A (en) | The automatic detection recognition method of crusing robot instrument based on deep learning | |
CN104899609B (en) | A kind of mechanical meter recognition methods based on image registration | |
CN111368906B (en) | Pointer type oil level meter reading identification method based on deep learning | |
CN104794484B (en) | The time series data arest neighbors sorting technique decomposed based on segmentation orthogonal polynomial | |
CN111369495B (en) | Panoramic image change detection method based on video | |
CN105303168A (en) | Multi-view pointer type instrument identification method and device | |
CN109508709B (en) | Single pointer instrument reading method based on machine vision | |
CN110659637A (en) | Electric energy meter number and label automatic identification method combining deep neural network and SIFT features | |
CN110909738A (en) | Automatic reading method of pointer instrument based on key point detection | |
CN114241469A (en) | Information identification method and device for electricity meter rotation process | |
CN114066887B (en) | Rice chalkiness area detection method, device, equipment and storage medium | |
CN116824726A (en) | Campus environment intelligent inspection method and system | |
CN113888741B (en) | Method for correcting rotating image of instrument in power distribution room | |
CN117746025A (en) | Indication recognition method and system for pointer instrument of substation equipment | |
CN111652244A (en) | Pointer type meter identification method based on unsupervised feature extraction and matching | |
CN115713750B (en) | Lane line detection method and device, electronic equipment and storage medium | |
CN102521874B (en) | Normal sampling recalculation method based on image reconstruction three-dimensional data | |
CN110232660B (en) | Novel infrared image recognition preprocessing gray stretching method | |
CN109784257B (en) | Transformer thermometer detection and identification method | |
CN116958983A (en) | Pointer type pressure gauge automatic calibrator indication reading method based on machine vision | |
CN116188755A (en) | Instrument angle correction and reading recognition device based on deep learning | |
CN115115889A (en) | Instrument image analysis method and device | |
CN112417975A (en) | Method and device for obtaining reading of pointer instrument |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |