CN110060299A - Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology - Google Patents
Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology Download PDFInfo
- Publication number
- CN110060299A CN110060299A CN201910313628.XA CN201910313628A CN110060299A CN 110060299 A CN110060299 A CN 110060299A CN 201910313628 A CN201910313628 A CN 201910313628A CN 110060299 A CN110060299 A CN 110060299A
- Authority
- CN
- China
- Prior art keywords
- danger source
- binocular
- target
- camera
- danger
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
- G06T7/85—Stereo camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses danger sources in a kind of passway for transmitting electricity based on binocular vision technology to identify and position method, including the following steps: a, collects transmission line of electricity danger source picture, marks out position coordinates of the danger source in picture and dangerous Source Type by hand;B, enhance algorithm by data and increase sample size;C, using depth convolutional neural networks training danger source target detection model;D, binocular camera is demarcated, binocular camera parameter is obtained;E, the target detection model obtained using c detects the target position in binocular image;F, the danger source target detected in binocular image is subjected to Stereo matching, obtains depth information, to obtain danger source target range, realize detection and positioning to danger source.The present invention realizes the continuous monitoring to information such as dangerous source position, speed.It realizes and Fast Classification detection and effectively identification is carried out to danger source target below transmission line of electricity, realize the Distance positioning to target.
Description
Technical field
The present invention relates to a kind of computer vision fields, and in particular in a kind of passway for transmitting electricity based on binocular vision technology
Danger source identifies and positions method.
Background technique
Building strong smart grid, making safely controllable electric system of new generation is that state's net development of company consolidates it
It wants, by research binocular vision technology and deep learning image recognition technology, the classification of passway for transmitting electricity danger source is carried out primary
Classification, and the direction of motion of line channel atural object, movement velocity are obtained, realize dangerous identifing source, positioning, tracking in passway for transmitting electricity
Detection;Early warning can be carried out to passway for transmitting electricity danger source in real time, to promote the management level of passway for transmitting electricity.
Summary of the invention
The purpose of the present invention is to provide in a kind of method simplicity, the good passway for transmitting electricity based on binocular vision technology of effect
Danger source identifies and positions method.
The technical solution of the invention is as follows:
Danger source identifies and positions method in a kind of passway for transmitting electricity based on binocular vision technology, it is characterized in that: under including
Column step:
A, transmission line of electricity danger source picture is collected, marks out position coordinates and danger source class of the danger source in picture by hand
Type;
B, enhance algorithm by data and increase sample size;
C, using depth convolutional neural networks training danger source target detection model;
D, binocular camera is demarcated, binocular camera parameter is obtained;
E, the target detection model obtained using c detects the target position in binocular image;
F, the danger source target detected in binocular image is subjected to Stereo matching, depth information is obtained, to obtain
Danger source target range realizes detection and positioning to danger source.
Step b the following steps are included:
Step b1: bi-directional scaling is carried out to transmission line of electricity danger source samples pictures, obtains different size of sample image;
Step b2: transmission line of electricity danger source samples pictures are carried out respectively within clockwise and counterclockwise 30 °
Angle rotation, obtain the sample image under different rotary angle;
Step b3: inputting confrontation for transmission line of electricity danger source samples pictures and generate network, and sample is added in network output picture
This.
Step c the following steps are included:
C1: building facilities network network layers are made of 6 layers of convolutional layer, pond layer and ReLU activation primitive respectively, image are sent into
Basic network obtains output o1;
C2: building Area generation network, for distinguishing foreground and background;Sliding window is applied on the output o1 of c1, often
A sliding window generates 9 candidate regions, and candidate region ratio is selected according to [1:2,1:1,2:1] Three models;By all times
Favored area is connected to Area generation network, returns out all candidate regions using softmax, selects preceding 500 regions as defeated
O2 out;
C3: the ResNet101 network that c2 output o2 access has been trained is subjected to transfer learning, obtains danger source detection mould
Type.
Step d the following steps are included:
D1: calibration for cameras internal reference;
D2: extracting scale invariant feature in binocular image, according to this feature, using stochastical sampling consistency method with
5 pairs of match points are randomly selected to machine every time, and obtain the initial attitude of camera by 5 algorithms;
D3: according to initial camera posture, the error matching points of the scale invariant feature matching centering overwhelming majority are filtered out;
D4: come using global light-stream adjustment while optimizing camera internal reference and Camera extrinsic, while reducing all matchings
Point and corresponding three-dimensional space point between the smallest back projection's error and;Camera parameter after global bundle adjustment optimization
As final camera parameter obtains the parallax D of the distance between camera focus F, binocular camera B and binocular camera.
Step f the following steps are included:
F1: the respective barycentric coodinates for calculating separately danger source target in binocular image are P and P ';
F2: the distance between P and P ' is obtained are as follows: B-D;
F3: according to image-forming principle:
Then target range are as follows:
Obtain final output: the distance of the position of target.
The present invention realizes the continuous monitoring to information such as dangerous source position, speed.Real-time video is acquired by sensor, is passed
It is defeated to arrive back-end processing system.Back-end processing system establishes study mould by sample collection using deep learning and mode identification method
Type is realized and carries out Fast Classification detection and effectively to danger sources targets such as crane, cement pump truck, wheeled digging machines below transmission line of electricity
Identification.Then Stereo Matching Technology is carried out using the binocular camera on shaft tower and obtain three-dimensional information, form point off density cloud, realize
To the Distance positioning of target.
Below with reference to embodiment, the invention will be further described.
Specific embodiment
Danger source identifies and positions method in a kind of passway for transmitting electricity based on binocular vision technology, including the following steps:
A, transmission line of electricity danger source picture is collected, marks out position coordinates and danger source class of the danger source in picture by hand
Type;
B, enhance algorithm by data and increase sample size;
C, using depth convolutional neural networks training danger source target detection model;
D, binocular camera is demarcated, binocular camera parameter is obtained;
E, the target detection model obtained using c detects the target position in binocular image;
F, the danger source target detected in binocular image is subjected to Stereo matching, depth information is obtained, to obtain
Danger source target range realizes detection and positioning to danger source.
Step b the following steps are included:
Step b1: bi-directional scaling is carried out to transmission line of electricity danger source samples pictures, obtains different size of sample image;
Step b2: transmission line of electricity danger source samples pictures are carried out respectively within clockwise and counterclockwise 30 °
Angle rotation, obtain the sample image under different rotary angle;
Step b3: inputting confrontation for transmission line of electricity danger source samples pictures and generate network, and sample is added in network output picture
This.
Step c the following steps are included:
C1: building facilities network network layers are made of 6 layers of convolutional layer, pond layer and ReLU activation primitive respectively, image are sent into
Basic network obtains output o1;
C2: building Area generation network, for distinguishing foreground and background;Sliding window is applied on the output o1 of c1, often
A sliding window generates 9 candidate regions, and candidate region ratio is selected according to [1:2,1:1,2:1] Three models;By all times
Favored area is connected to Area generation network, returns out all candidate regions using softmax, selects preceding 500 regions as defeated
O2 out;
C3: the ResNet101 network that c2 output o2 access has been trained is subjected to transfer learning, obtains danger source detection mould
Type.
Step d the following steps are included:
D1: calibration for cameras internal reference;
D2: extracting scale invariant feature in binocular image, according to this feature, using stochastical sampling consistency method with
5 pairs of match points are randomly selected to machine every time, and obtain the initial attitude of camera by 5 algorithms;
D3: according to initial camera posture, the error matching points of the scale invariant feature matching centering overwhelming majority are filtered out;
D4: come using global light-stream adjustment while optimizing camera internal reference and Camera extrinsic, while reducing all matchings
Point and corresponding three-dimensional space point between the smallest back projection's error and;Camera parameter after global bundle adjustment optimization
As final camera parameter obtains the parallax D of the distance between camera focus F, binocular camera B and binocular camera.
Step f the following steps are included:
F1: the respective barycentric coodinates for calculating separately danger source target in binocular image are P and P ';
F2: the distance between P and P ' is obtained are as follows: B-D;
F3: according to image-forming principle:
Then target range are as follows:
Obtain final output: the distance of the position of target.
Claims (5)
1. danger source identifies and positions method in a kind of passway for transmitting electricity based on binocular vision technology, it is characterized in that: including following
Step:
A, transmission line of electricity danger source picture is collected, marks out position coordinates of the danger source in picture and dangerous Source Type by hand;
B, enhance algorithm by data and increase sample size;
C, using depth convolutional neural networks training danger source target detection model;
D, binocular camera is demarcated, binocular camera parameter is obtained;
E, the target detection model obtained using c detects the target position in binocular image;
F, the danger source target detected in binocular image is subjected to Stereo matching, depth information is obtained, to obtain danger
Source target range realizes detection and positioning to danger source.
2. danger source identifies and positions method in the passway for transmitting electricity according to claim 1 based on binocular vision technology,
Be characterized in: step b the following steps are included:
Step b1: bi-directional scaling is carried out to transmission line of electricity danger source samples pictures, obtains different size of sample image;
Step b2: transmission line of electricity danger source samples pictures are subjected to the angle within clockwise and counterclockwise 30 ° respectively
Degree rotation, obtains the sample image under different rotary angle;
Step b3: inputting confrontation for transmission line of electricity danger source samples pictures and generate network, and sample is added in network output picture.
3. danger source identifies and positions method in the passway for transmitting electricity according to claim 1 based on binocular vision technology,
Be characterized in: step c the following steps are included:
C1: building facilities network network layers are made of 6 layers of convolutional layer, pond layer and ReLU activation primitive respectively, image are sent into basis
Network obtains output o1;
C2: building Area generation network, for distinguishing foreground and background;Sliding window, each cunning are applied on the output o1 of c1
Dynamic window generates 9 candidate regions, and candidate region ratio is selected according to [1:2,1:1,2:1] Three models;By all candidate regions
Domain is connected to Area generation network, returns out all candidate regions using softmax, selects preceding 500 regions as output o2;
C3: the ResNet101 network that c2 output o2 access has been trained is subjected to transfer learning, obtains danger source detection model.
4. danger source identifies and positions method in the passway for transmitting electricity according to claim 1 based on binocular vision technology,
Be characterized in: step d the following steps are included:
D1: calibration for cameras internal reference;
D2: extracting scale invariant feature in binocular image, according to this feature, using stochastical sampling consistency method randomly
5 pairs of match points are randomly selected every time, and obtain the initial attitude of camera by 5 algorithms;
D3: according to initial camera posture, the error matching points of the scale invariant feature matching centering overwhelming majority are filtered out;
D4: using global light-stream adjustment come and meanwhile optimize camera internal reference and Camera extrinsic, while reduce all match points with
And the smallest back projection's error between corresponding three-dimensional space point and;Camera parameter after the optimization of global bundle adjustment is
Final camera parameter obtains the parallax D of the distance between camera focus F, binocular camera B and binocular camera.
5. danger source identifies and positions method in the passway for transmitting electricity according to claim 1 based on binocular vision technology,
Be characterized in: step f the following steps are included:
F1: the respective barycentric coodinates for calculating separately danger source target in binocular image are P and P ';
F2: the distance between P and P ' is obtained are as follows: B-D;
F3: according to image-forming principle:
Then target range are as follows:
Obtain final output: the distance of the position of target.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910313628.XA CN110060299A (en) | 2019-04-18 | 2019-04-18 | Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910313628.XA CN110060299A (en) | 2019-04-18 | 2019-04-18 | Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110060299A true CN110060299A (en) | 2019-07-26 |
Family
ID=67319416
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910313628.XA Pending CN110060299A (en) | 2019-04-18 | 2019-04-18 | Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110060299A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110543824A (en) * | 2019-08-01 | 2019-12-06 | 江苏濠汉信息技术有限公司 | construction safety judgment method based on binocular vision |
CN110599489A (en) * | 2019-08-26 | 2019-12-20 | 华中科技大学 | Target space positioning method |
CN110675449A (en) * | 2019-09-02 | 2020-01-10 | 山东科技大学 | Binocular camera-based offshore flow detection method |
CN112539704A (en) * | 2020-12-24 | 2021-03-23 | 国网山东省电力公司检修公司 | Method for measuring distance between hidden danger in transmission line channel and lead |
CN113179389A (en) * | 2021-04-15 | 2021-07-27 | 江苏濠汉信息技术有限公司 | System and method for identifying crane jib of power transmission line dangerous vehicle |
CN113283496A (en) * | 2021-05-21 | 2021-08-20 | 国网宁夏电力有限公司中卫供电公司 | Method applied to intelligent identification of dangerous actions of distribution network uninterrupted operation |
CN113456033A (en) * | 2021-06-24 | 2021-10-01 | 江西科莱富健康科技有限公司 | Physiological index characteristic value data processing method and system and computer equipment |
CN113822249A (en) * | 2021-11-23 | 2021-12-21 | 山东信通电子股份有限公司 | Method and equipment for monitoring position of hidden danger of overhead line |
CN114639220A (en) * | 2022-03-16 | 2022-06-17 | 国能榆林能源有限责任公司 | Coal mining area alarm method, system and storage medium |
CN114894091A (en) * | 2022-05-09 | 2022-08-12 | 上海倍肯智能科技有限公司 | Circuit monitoring device and system with binocular vision ranging function |
CN116189100A (en) * | 2023-04-27 | 2023-05-30 | 江苏三棱智慧物联发展股份有限公司 | Gas hazard source detection and identification method and system based on spectral image |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106372662A (en) * | 2016-08-30 | 2017-02-01 | 腾讯科技(深圳)有限公司 | Helmet wearing detection method and device, camera, and server |
CN108665484A (en) * | 2018-05-22 | 2018-10-16 | 国网山东省电力公司电力科学研究院 | A kind of dangerous source discrimination and system based on deep learning |
US20180300590A1 (en) * | 2017-04-13 | 2018-10-18 | Facebook, Inc. | Panoramic camera systems |
CN108876855A (en) * | 2018-05-28 | 2018-11-23 | 哈尔滨工程大学 | A kind of sea cucumber detection and binocular visual positioning method based on deep learning |
CN109543505A (en) * | 2018-09-29 | 2019-03-29 | 江苏濠汉智能设备有限公司 | A kind of object detection system and method based on video image |
-
2019
- 2019-04-18 CN CN201910313628.XA patent/CN110060299A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106372662A (en) * | 2016-08-30 | 2017-02-01 | 腾讯科技(深圳)有限公司 | Helmet wearing detection method and device, camera, and server |
US20180300590A1 (en) * | 2017-04-13 | 2018-10-18 | Facebook, Inc. | Panoramic camera systems |
CN108665484A (en) * | 2018-05-22 | 2018-10-16 | 国网山东省电力公司电力科学研究院 | A kind of dangerous source discrimination and system based on deep learning |
CN108876855A (en) * | 2018-05-28 | 2018-11-23 | 哈尔滨工程大学 | A kind of sea cucumber detection and binocular visual positioning method based on deep learning |
CN109543505A (en) * | 2018-09-29 | 2019-03-29 | 江苏濠汉智能设备有限公司 | A kind of object detection system and method based on video image |
Non-Patent Citations (3)
Title |
---|
丁莹等: "《复杂环境运动目标检测技术及应用》", 31 January 2014 * |
李程启等: "基于深度学习的输电线路危险源智能监控***", 《南通大学学报(自然科学版)》 * |
王涛等: "自然场景中近景摄影测量相机的自标定方法", 《工具技术》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110543824A (en) * | 2019-08-01 | 2019-12-06 | 江苏濠汉信息技术有限公司 | construction safety judgment method based on binocular vision |
CN110599489A (en) * | 2019-08-26 | 2019-12-20 | 华中科技大学 | Target space positioning method |
CN110675449A (en) * | 2019-09-02 | 2020-01-10 | 山东科技大学 | Binocular camera-based offshore flow detection method |
CN112539704B (en) * | 2020-12-24 | 2022-03-01 | 国网山东省电力公司检修公司 | Method for measuring distance between hidden danger in transmission line channel and lead |
CN112539704A (en) * | 2020-12-24 | 2021-03-23 | 国网山东省电力公司检修公司 | Method for measuring distance between hidden danger in transmission line channel and lead |
CN113179389A (en) * | 2021-04-15 | 2021-07-27 | 江苏濠汉信息技术有限公司 | System and method for identifying crane jib of power transmission line dangerous vehicle |
CN113283496A (en) * | 2021-05-21 | 2021-08-20 | 国网宁夏电力有限公司中卫供电公司 | Method applied to intelligent identification of dangerous actions of distribution network uninterrupted operation |
CN113283496B (en) * | 2021-05-21 | 2022-07-29 | 国网宁夏电力有限公司中卫供电公司 | Method applied to intelligent identification of dangerous actions of distribution network uninterrupted operation |
CN113456033A (en) * | 2021-06-24 | 2021-10-01 | 江西科莱富健康科技有限公司 | Physiological index characteristic value data processing method and system and computer equipment |
CN113456033B (en) * | 2021-06-24 | 2023-06-23 | 江西科莱富健康科技有限公司 | Physiological index characteristic value data processing method, system and computer equipment |
CN113822249A (en) * | 2021-11-23 | 2021-12-21 | 山东信通电子股份有限公司 | Method and equipment for monitoring position of hidden danger of overhead line |
CN114639220A (en) * | 2022-03-16 | 2022-06-17 | 国能榆林能源有限责任公司 | Coal mining area alarm method, system and storage medium |
CN114894091A (en) * | 2022-05-09 | 2022-08-12 | 上海倍肯智能科技有限公司 | Circuit monitoring device and system with binocular vision ranging function |
CN114894091B (en) * | 2022-05-09 | 2024-04-19 | 上海倍肯智能科技有限公司 | Line monitoring device and system with binocular vision ranging function |
CN116189100A (en) * | 2023-04-27 | 2023-05-30 | 江苏三棱智慧物联发展股份有限公司 | Gas hazard source detection and identification method and system based on spectral image |
CN116189100B (en) * | 2023-04-27 | 2023-07-18 | 江苏三棱智慧物联发展股份有限公司 | Gas hazard source detection and identification method and system based on spectral image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110060299A (en) | Danger source identifies and positions method in passway for transmitting electricity based on binocular vision technology | |
CN110097553B (en) | Semantic mapping system based on instant positioning mapping and three-dimensional semantic segmentation | |
WO2022188379A1 (en) | Artificial intelligence system and method serving electric power robot | |
CN113450408B (en) | Irregular object pose estimation method and device based on depth camera | |
CN111340797A (en) | Laser radar and binocular camera data fusion detection method and system | |
CN104881029B (en) | Mobile Robotics Navigation method based on a point RANSAC and FAST algorithms | |
CN107481315A (en) | A kind of monocular vision three-dimensional environment method for reconstructing based on Harris SIFT BRIEF algorithms | |
CN111998862B (en) | BNN-based dense binocular SLAM method | |
CN103530599A (en) | Method and system for distinguishing real face and picture face | |
CN107397658B (en) | Multi-scale full-convolution network and visual blind guiding method and device | |
CN113516108B (en) | Construction site dust suppression data matching processing method based on data identification | |
CN112750198B (en) | Dense correspondence prediction method based on non-rigid point cloud | |
CN109816780B (en) | Power transmission line three-dimensional point cloud generation method and device of binocular sequence image | |
CN103198477A (en) | Apple fruitlet bagging robot visual positioning method | |
CN106097431A (en) | A kind of object global recognition method based on 3 d grid map | |
CN113284144B (en) | Tunnel detection method and device based on unmanned aerial vehicle | |
CN113393524B (en) | Target pose estimation method combining deep learning and contour point cloud reconstruction | |
CN116222577B (en) | Closed loop detection method, training method, system, electronic equipment and storage medium | |
CN115032648A (en) | Three-dimensional target identification and positioning method based on laser radar dense point cloud | |
Feng et al. | Object detection and localization based on binocular vision for autonomous vehicles | |
CN116659516B (en) | Depth three-dimensional attention visual navigation method and device based on binocular parallax mechanism | |
Liu et al. | Deep learning of directional truncated signed distance function for robust 3D object recognition | |
Yang et al. | Research and application of 3D face modeling algorithm based on ICP accurate alignment | |
CN113221744B (en) | Monocular image 3D object detection method based on deep learning | |
CN115761265A (en) | Method and device for extracting substation equipment in laser radar point cloud |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190726 |