CN114529811A - Rapid and automatic identification and positioning method for foreign matters in subway tunnel - Google Patents
Rapid and automatic identification and positioning method for foreign matters in subway tunnel Download PDFInfo
- Publication number
- CN114529811A CN114529811A CN202011214199.XA CN202011214199A CN114529811A CN 114529811 A CN114529811 A CN 114529811A CN 202011214199 A CN202011214199 A CN 202011214199A CN 114529811 A CN114529811 A CN 114529811A
- Authority
- CN
- China
- Prior art keywords
- foreign matters
- foreign
- camera
- image
- distance
- 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 32
- 238000001514 detection method Methods 0.000 claims abstract description 34
- 238000012544 monitoring process Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 12
- 238000003384 imaging method Methods 0.000 claims description 8
- 238000005259 measurement Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000003062 neural network model Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 7
- 238000011897 real-time detection Methods 0.000 abstract description 7
- 238000007689 inspection Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 108091008695 photoreceptors Proteins 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000005477 standard model Effects 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
- G06F18/21—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
- 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
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/55—Depth or shape recovery from multiple images
- G06T7/593—Depth or shape recovery from multiple images from stereo images
-
- 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
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
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)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method for quickly and automatically identifying and positioning foreign matters in a subway tunnel, which comprises the following steps: two identical cameras are arranged on the robot; acquiring subway tunnel images of two camera monitoring areas; processing the acquired image and judging whether foreign matters exist in the image or not; when the foreign body is detected, judging the type of the foreign body and calculating the position of the foreign body and the distance between the foreign body and the camera; according to the type, position and distance of the foreign matters, the robot grabs the foreign matters. The foreign matter is identified by using a YOLO v3 target identification algorithm. The method has the advantages that the deep network model trained by the YOLO object detection framework is used for real-time foreign matter detection, and compared with the traditional detection algorithm, the deep network model trained by the YOLO object detection framework is more accurate, and meanwhile, the identification process is very rapid, and the target of real-time detection can be achieved.
Description
Technical Field
The invention belongs to the field related to subway tunnel foreign matter detection, and particularly relates to a subway tunnel foreign matter detection method and system.
Background
In the normal operation process of the subway, the foreign matters invade the safe driving range of the subway and collide and rub with the subway occasionally. As the foreign matter intrusion event has the characteristics of emergencies, uncertainty, great harmfulness and the like, once the train collides with foreign matters, the operation is interrupted if the train collides with the foreign matters, the driving safety is endangered if the train collides with the foreign matters, meanwhile, the damage of vehicle equipment and the casualties are brought, and the safe operation of the subway train is seriously threatened. Therefore, how to timely detect the intrusion limit of the foreign matter in the running process of the train, and perform early warning and alarming before the train arrives, and timely take effective measures to avoid the collision between the subway and the intruding object is an important ring for ensuring the normal running of the subway at present.
At present, the detection of the foreign matters in the subway tunnel is mainly completed by utilizing manual itineration inspection, so that a large amount of manpower and material resources are consumed, meanwhile, the efficiency is low, the subway speed is increased and the track line is prolonged for the subway track traffic which is rapidly developed at present, and workers hardly complete the detection in a limited time.
For tunnel foreign matter detection with a robot as a carrier, an intelligent foreign matter detection device in a subway rail section is disclosed in Chinese invention patent specification CN 109131444A, a monorail trolley is used for running on a rail, image recognition is used for judging whether a subway tunnel is invaded by foreign matters, and the foreign matters are reported to a monitoring center if found; however, the method consumes a large amount of resources for maintenance, and once the trolley is in failure, the system needs manpower and material resources to be checked and repaired. The invention discloses an intelligent detection system for a rail transit tunnel in Chinese patent specification CN 108248635A, which monitors the inside of the tunnel in real time through a real-time monitoring intelligent inspection robot arranged on a guide rail and transmits detected data and information to a central control system of a platform. Potential safety hazards existing inside the rail transit tunnel can be found; however, although this method can detect the tunnel state in real time, it cannot recognize and capture a foreign object. The invention discloses a method for omnibearing real-time monitoring by using a tunnel inspection robot in Chinese patent specification CN 109688388A, wherein the tunnel inspection robot receives an instruction and then conducts real-time omnibearing monitoring on the interior of a tunnel, including illegal driving, traffic accidents, vehicle congestion, road foreign matters and other conditions, so as to timely discover and timely process tunnel abnormity and illegal conditions; however, although this method can recognize the foreign object, it cannot combine recognition with automatic grasping, and cannot recognize and grasp the foreign object at any time. In chinese patent specification CN 108657223 a, an automatic inspection system for urban rail transit and a tunnel deformation detection method are disclosed, wherein a tunnel deformation diagram of a corresponding position point is obtained by comparing a three-dimensional model of a tunnel after real-time detection processing with a standard model. Various parameters including tunnel equipment facilities and cable temperature, tunnel deformation and track damage are detected, and foreign matters in the track can be automatically identified and cleaned, so that the labor cost is greatly saved; the invention discloses an automatic inspection robot, which integrates all system modules together and does not research a specific technical method for automatically identifying and positioning foreign matters.
For the existing railway track foreign matter intrusion detection technology, in Chinese patent specification CN 108549087A, an on-line detection method based on laser radar is disclosed, which judges whether foreign matters exist in a monitoring range through laser radar monitoring; however, this method cannot recognize the type of the foreign object and capture the foreign object in time. The Chinese utility model patent specification CN 207712053U discloses a robot for rail transit, which can realize the rapid inspection of rail transit, can remove foreign matters on the rail, and adopts an inspection device with a special structure to realize the clearer detection of the rail; however, this patent does not identify parts of the system and only illustrates the working principle and mode of the robot. In chinese patent specification CN 206115276U, a robot system for inspecting underground foreign objects is disclosed, which comprises a robot body, a robot controller, a control terminal, and a power supply, wherein the robot is used for performing inspection, and the inspection result is transmitted to a central control system, so that workers can know the current situation in time; the system can not realize the identification and timely grabbing of the foreign matters, and still consumes certain manpower and material resources. In chinese patent specification CN 108197610 a, a track foreign matter detection system based on deep learning is disclosed, which comprises: the vehicle-mounted image acquisition device, the image transmission unit, the foreign matter detection device, the machine learning device and the image data can realize real-time detection of the rail foreign matter; but the system cannot determine the specific position of the foreign body and timely pick up the foreign body.
In summary, the present domestic research on the foreign matters in the subway tunnel has achieved certain achievements, but research and development on the aspects of identifying the types and the position distances of the foreign matters in real time and picking up the foreign matters in real time by using the existing robot as a carrier still need to be explored.
Disclosure of Invention
In view of the above, the invention provides a technology for quickly and automatically identifying and positioning a foreign object in a subway tunnel, which can quickly, effectively and automatically identify and position the type and position of the foreign object, and simultaneously returns the data of the foreign object in real time by combining a robot, so as to realize the grabbing of the foreign object by a manipulator.
The invention adopts the following technical scheme: a quick and automatic identification and positioning method for a foreign body in a subway tunnel comprises the following steps:
s1: two identical cameras are arranged on the robot;
s2, acquiring subway tunnel images of two camera monitoring areas;
s3: processing the acquired image and judging whether foreign matters exist in the image or not;
s4: when the foreign body is detected, judging the type of the foreign body and calculating the position of the foreign body and the distance between the foreign body and the camera;
s5: according to the type, position and distance of the foreign matters, the robot grabs the foreign matters.
The S3 includes the steps of:
acquiring an image by any one of the two cameras, and performing image enhancement and filtering processing;
and judging and predicting the type and the position of the foreign matter in the image through a neural network.
The training of the neural network comprises the following steps:
(1) acquiring image data of the foreign matters in the subway tunnel by two cameras in advance, labeling the foreign matters in the image acquired by each camera, representing the positions of the foreign matters by using a label frame containing the foreign matters, and representing the types of the foreign matters by using label names;
(2) and establishing a YOLO neural network model for foreign matter detection, and carrying out YOLO neural network model training on images acquired and marked by each camera to obtain the trained YOLO neural network model for identifying foreign matters from monitoring images acquired by any camera in real time.
The method comprises the following steps of calculating the distance between the foreign object and the camera, and calculating the distance between the foreign object and the camera by using binocular ranging, wherein the method comprises the following steps:
according to the camera parameters, the parallax error of the same foreign object between the binocular cameras is obtained, and then the depth information of the foreign object in the image, namely the distance between the foreign object and the cameras is obtained.
The method comprises the following steps of calculating the distance between the foreign object and the camera, and calculating the distance between the foreign object and the camera by using binocular ranging, wherein the method comprises the following steps:
calibrating a binocular camera in advance to obtain camera parameters of the two cameras, correcting images respectively obtained by the two cameras to enable the two corrected images to be located on the same plane and to be parallel to each other, calculating the depth of each pixel according to the matching result of the two corrected images to obtain a depth map, and obtaining the distance between the foreign matter and the cameras according to the depth map.
The distance between the foreign object and the camera is obtained by the following formula:
wherein c is the focal length of the camera, b is the length of the connecting line of the imaging points of the two cameras, d is the parallax, and Z is the distance from the foreign object to the midpoint of the connecting line of the two cameras.
A quick automatic identification of subway tunnel foreign matter and positioning system includes:
the image processing module is used for processing the subway tunnel image acquired by the binocular camera and judging whether foreign matters exist in the image;
and the image detection module is used for judging the type of the foreign matters and calculating the positions of the foreign matters and the distance between the foreign matters and the camera when the foreign matters are detected, so that the robot can grab the foreign matters.
Generally, compared with the prior art, the above technical solution according to the present invention mainly has the following technical advantages:
1. foreign objects are identified by using a Yolo v3 target identification algorithm. The method has the advantages that the deep network model trained by the YOLO object detection framework is used for real-time foreign matter detection, and compared with the traditional detection algorithm, the deep network model trained by the YOLO object detection framework is more accurate, and meanwhile, the identification process is very rapid, and the target of real-time detection can be achieved.
2. And calculating the distance of the foreign matters by adopting binocular ranging. Compared with monocular distance measurement, the binocular distance measurement can directly utilize parallax to calculate the distance, the precision is higher than that of a monocular distance measurement, more accurate position distance information is obtained through images, and the identification calculation requirements of people are well met.
3. This technique uses the robot as the carrier, when can accurately discern the location foreign matter position fast, can combine together with the robot, realizes real-time detection, the real-time snatching to the foreign matter, ensures no foreign matter in the subway tunnel at any time, ensures subway safe operation, guarantees driver and passenger's health safety.
Drawings
FIG. 1 is a schematic flow chart of the present technique;
FIG. 2 is a schematic diagram of a pinhole model;
fig. 3 is a schematic diagram of a binocular imaging model.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The invention comprises the following steps:
s1: installing cameras according to requirements, and installing two identical cameras at proper positions of the robot;
s2, acquiring an image of the camera monitoring area;
s3: processing the acquired image and judging whether foreign matters exist in the image or not;
s4: when the foreign body is detected, judging the type of the foreign body and calculating the position and the distance between the foreign body and a camera (a robot);
s5: the robot grabs the foreign matters according to the types, positions and distance data of the foreign matters.
Preferably, for step S1, the system is combined with existing robot technology, and the foreign object is found and captured in time by using the capturing function of the manipulator of the robot.
Preferably, for step S1, the robot as the system carrier should be selected to have a moving function and a grasping function. The moving function should be able to adjust the direction and speed of movement to correctly be close to the foreign object, and the grabbing function should be able to adjust the size, angle and force of the manipulator to successfully grab the foreign object.
Preferably, for step S1, the positions of the two cameras should be installed at the proper positions of the robot, so as to monitor the tunnel global image.
Preferably, in step S3, the image obtained by the camera is first subjected to image enhancement, filtering, and other processing to obtain a picture meeting the identification requirement.
Preferably, for step S3, the object recognition and positioning algorithm YOLO v3 based on the deep neural network is used to implement judgment and prediction of the type and position of the foreign object in the image. The YOLO v3 has the typical characteristics of high running speed, capability of being used for real-time detection, low background false detection rate, high identification accuracy rate and capability of meeting technical requirements.
The implementation of the YOLO recognition algorithm needs to include the following steps: (1) a large amount of image data is collected in advance, and a foreign object is marked in the image, the position of the foreign object is represented by a label box containing the foreign object, and the type of the foreign object is represented by a label name. (2) And establishing a foreign matter detection and identification model, and carrying out YOLO neural network model training by using the acquired tunnel foreign matter image to obtain a trained foreign matter detection model. (3) And testing the detection model by using the trained foreign matter detection and identification model to monitor images collected by the camera, and identifying the foreign matters.
When the robot patrols and examines, the camera catches image data, transmits the data as input to the neural network, detects it, if there is a foreign matter, identifies the specific position and category, if there is no foreign matter, continues to process the next frame of image.
Preferably, for step S4, calculating the distance between the alien material and the camera (robot) is accomplished using binocular ranging. According to the camera imaging model, for the monitoring of the same object, the positions of the object in the image are different when the object is observed at different positions. The principle of binocular ranging is similar to that of human eyes. Human eyes can perceive the distance of an object because the images of the same object presented by the two eyes are different, which is also called as parallax. The farther the object distance is, the smaller the parallax error is; conversely, the greater the parallax. By calculating the parallax of the same object between the binocular cameras, the depth information of the object in the camera image, namely the distance between the object and the cameras, can be calculated. Binocular ranging generally requires several steps: (1) and calibrating the binocular cameras to obtain internal and external parameters and homography matrixes of the two cameras. (2) And correcting the original image according to the calibration result, wherein the two corrected images are positioned on the same plane and are parallel to each other. (3) And matching pixel points of the two corrected images. (4) And calculating the depth of each pixel according to the matching result, thereby obtaining a depth map.
Preferably, for step S5, the foreign object type, position and distance data obtained in step S4 are returned to the robot, and the robot adjusts its position in real time according to the information, approaches and grasps the foreign object.
Fig. 1 illustrates a schematic frame structure diagram of a rapid automatic identification and positioning technology for a foreign matter in a subway tunnel according to the present invention, which comprises the following steps:
s1: installing cameras according to requirements, and installing two identical cameras at proper positions of the robot;
due to the detection requirement, in order to realize the detection of the foreign matters in the subway tunnel, a camera is additionally arranged on the robot. According to the invention, the acquisition of a subway tunnel monitoring image is realized by using two cameras, the image acquired by a single camera is used for identifying and detecting the foreign matters in the subway tunnel, and the calculation of the time distance between the foreign matters and the cameras is realized by using the principle that the two cameras are used for shooting the same foreign matters at the same moment according to the images shot by the two cameras at different positions, and the positions of the foreign matters in the images are different to generate parallax;
meanwhile, the foreign matters are picked up by a robot hand. According to the technology, the robot is used as a carrier, when the system calculates the specific type distance of the foreign matters, data are transmitted into the robot, the robot performs corresponding actions to approach and pick up the foreign matters, and therefore real-time detection and real-time capture of the foreign matters are achieved.
When the camera is additionally arranged on the robot, the structure of the robot is not changed when the camera is additionally arranged on the robot in consideration of cost and structural reasons. The structure is considered, and the installation position and the angle of the camera are also considered. The two camera positions are arranged to be symmetrical about the center line of the robot as much as possible, and the directions are on the same horizontal line as much as possible, so that the longer vision field and the simpler calculation are achieved.
S2, acquiring an image of the camera monitoring area;
the parameters of the cameras are continuously adjusted in the experimental process, and the parameters of the two cameras are fixed before the operation, so that the images collected by the cameras are clear and accurate, and preparation is made for later image identification. The collected image is transmitted to the system through the image transmission unit, so that the image is further processed.
S3: processing the acquired image and judging whether foreign matters exist in the image or not;
s4: when the foreign body is detected, judging the type of the foreign body and calculating the position and the distance between the foreign body and a camera (a robot);
in the step S3, the acquired image is subjected to image enhancement, filtering and other processing, and because the subway tunnel is relatively dark, the acquired image becomes clear through image enhancement, brightness is improved, and foreign matters and a background are more clearly separated; filtering out noise in the background that may interfere with detection provides for identification of foreign objects.
The fast detection and identification of the subway tunnel foreign matter mentioned in steps S3 and S4. And (3) judging and predicting the type and the position of the foreign matter in the image by using an object recognition and positioning algorithm YOLO v3 based on a deep neural network.
For detecting foreign objects using the YOLO algorithm, the following steps are required: 1) a large amount of data sets are collected, subway tunnel pictures with foreign matters are searched or shot on the spot on the network, the foreign matters comprise notebook bags, male and female backpacks, handbags, handbag bags, plastic bags filled with articles, express packaging bags, water cups, medicine bottles, mineral water bottles, cannon-shot cylinders and the like, and the pictures are used as data sets for training. Labeling the collected pictures with labeling tool such as labelimg, surrounding the object with a label frame, identifying the position and size of the foreign matter, filling the type of the foreign matter in the label name, and making the labeled picture file format be xml. 2) And modifying the Makefile, and training the data set by using the GPU, so that the training speed of the user can be greatly improved. Performing network training on our data set, and performing model training by using a network structure and a training algorithm provided by a YOLO framework to obtain the weight required by us; 3) and (3) performing network prediction on the trained model, returning the boundary frame coordinates (the center coordinates of the boundary frame and the length and width of the boundary frame) of the foreign matters and the types of the foreign matters from the test result, continuously adjusting the training parameters according to the test result, and optimizing the network structure so that the final model meets the actual requirement.
The calculation of the distance between the foreign object and the camera (robot) mentioned in step S4 is performed by binocular distance measurement.
The binocular distance measurement can directly utilize parallax to calculate distance, the precision is higher than that of a single eye, and more accurate position distance information is obtained through images; the method has no limit of identification rate, because in principle, identification is not needed first and then measurement is carried out, but all obstacles are directly measured; binocular vision also does not require maintenance of a sample database because there is no notion of samples for binocular.
The imaging mechanism of the camera is approximately described by a pinhole model, as shown in fig. 2. M is an object point in a real scene, O is the optical center of the camera, O ' is the projection of the optical center on an image plane, OO ' is the optical axis of the camera, and M ' is the image point of M on the image plane P.
For simplicity, consider two cameras with parallel optical axes separated by a suitable distance (the camera parameters are identical), which is the most ideal model of the eyes, and the images at the same object point P are shown in fig. 3.
Referring to fig. 3, P is a point on the object to be measured, and O' and O ″ are optical centers of the left and right cameras. The imaging points of the point P on the two camera photoreceptors are respectively P 'and P "(the imaging plane of the camera is placed in front of the lens after rotating), c is the focal length of the camera, b is the center line (base line) of the two cameras, and the distance in the depth direction from the point P to the point O' is set as Z, then the principle of similarity of triangles is adopted:
the X/Y coordinates of the point P are:
in the depth direction:
thus, it can be deduced that:
Where X' and X "are the distances of the point P in the X direction of the image point shift on the imaging plane, respectively, and the difference between the two is defined as the parallax, i.e. the difference between the image positions of the image points in different cameras, and is denoted by d.
The binocular measurement technique determines the target position by the principle of parallax.
Binocular ranging generally requires several steps: (1) and calibrating the binocular cameras to obtain internal and external parameters and homography matrixes of the two cameras. (2) And correcting the original image according to the calibration result, wherein the two corrected images are positioned on the same plane and are parallel to each other. (3) And matching pixel points of the two corrected images. (4) And calculating the depth of each pixel according to the matching result, thereby obtaining a depth map.
S5: the robot grabs the foreign matters according to the types, positions and distance data of the foreign matters.
In conclusion, the invention realizes the rapid identification of the type and the position of the foreign matters through the YOLO v3 target object identification and positioning algorithm, realizes the calculation of the distance of the foreign matters through binocular ranging, integrates the functions by taking a robot as a carrier, realizes the real-time rapid identification and positioning of the foreign matters in the subway tunnel, picks up the foreign matters in time and ensures the driving safety of the subway train.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. A quick and automatic identification and positioning method for foreign matters in a subway tunnel is characterized by comprising the following steps:
s1: two identical cameras are arranged on the robot;
s2, acquiring subway tunnel images of two camera monitoring areas;
s3: processing the acquired image and judging whether foreign matters exist in the image or not;
s4: when the foreign body is detected, judging the type of the foreign body and calculating the position of the foreign body and the distance between the foreign body and the camera;
s5: according to the type, position and distance of the foreign matters, the robot grabs the foreign matters.
2. The method for rapidly and automatically identifying and positioning the foreign matters in the subway tunnel as claimed in claim 1, wherein said S3 comprises the steps of:
acquiring an image by any one of the two cameras, and performing image enhancement and filtering processing;
and judging and predicting the type and the position of the foreign matter in the image through a neural network.
3. The method for rapidly and automatically identifying and positioning the foreign matters in the subway tunnel according to claim 2, wherein the training of the neural network comprises the following steps:
(1) acquiring image data of the foreign matters in the subway tunnel by two cameras in advance, labeling the foreign matters in the image acquired by each camera, representing the positions of the foreign matters by using a label frame containing the foreign matters, and representing the types of the foreign matters by using label names;
(2) and establishing a YOLO neural network model for foreign matter detection, and carrying out YOLO neural network model training on images acquired and labeled by each camera to obtain the trained YOLO neural network model for identifying foreign matters on monitoring images acquired by any camera in real time.
4. The method for rapidly and automatically identifying and positioning the foreign matters in the subway tunnel according to claim 1, wherein the step of calculating the distance between the foreign matters and the camera by using binocular distance measurement comprises the following steps:
according to the camera parameters, the parallax error of the same foreign object between the binocular cameras is obtained, and then the depth information of the foreign object in the image, namely the distance between the foreign object and the cameras is obtained.
5. The method for rapidly and automatically identifying and positioning the foreign matters in the subway tunnel according to claim 1, wherein the step of calculating the distance between the foreign matters and the camera by using binocular distance measurement comprises the following steps:
calibrating a binocular camera in advance to obtain camera parameters of the two cameras, correcting images respectively obtained by the two cameras to enable the two corrected images to be located on the same plane and to be parallel to each other, calculating the depth of each pixel according to the matching result of the two corrected images to obtain a depth map, and obtaining the distance between the foreign matter and the cameras according to the depth map.
6. A method for rapidly and automatically identifying and positioning a foreign object in a subway tunnel according to claim 4 or 5, wherein the distance between the foreign object and the camera is obtained by the following formula:
wherein c is the focal length of the camera, b is the length of the connecting line of the imaging points of the two cameras, d is the parallax, and Z is the distance from the foreign object to the midpoint of the connecting line of the two cameras.
7. The utility model provides a quick automatic identification of subway tunnel foreign matter and positioning system which characterized in that includes:
the image processing module is used for processing the subway tunnel image acquired by the binocular camera and judging whether foreign matters exist in the image;
and the image detection module is used for judging the type of the foreign matters and calculating the positions of the foreign matters and the distance between the foreign matters and the camera when the foreign matters are detected, so that the robot can grab the foreign matters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011214199.XA CN114529811A (en) | 2020-11-04 | 2020-11-04 | Rapid and automatic identification and positioning method for foreign matters in subway tunnel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011214199.XA CN114529811A (en) | 2020-11-04 | 2020-11-04 | Rapid and automatic identification and positioning method for foreign matters in subway tunnel |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114529811A true CN114529811A (en) | 2022-05-24 |
Family
ID=81618622
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011214199.XA Pending CN114529811A (en) | 2020-11-04 | 2020-11-04 | Rapid and automatic identification and positioning method for foreign matters in subway tunnel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114529811A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115035060A (en) * | 2022-06-07 | 2022-09-09 | 贵州聚原数技术开发有限公司 | Tunnel wall deformation detection method based on computer image recognition |
| CN115892131A (en) * | 2023-02-15 | 2023-04-04 | 深圳大学 | Intelligent monitoring method and system for subway tunnel |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004059900A2 (en) * | 2002-12-17 | 2004-07-15 | Evolution Robotics, Inc. | Systems and methods for visual simultaneous localization and mapping |
| 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 |
| CN208868062U (en) * | 2018-07-23 | 2019-05-17 | 中国安全生产科学研究院 | A kind of urban track traffic automatic tour inspection system |
| CN110217271A (en) * | 2019-05-30 | 2019-09-10 | 成都希格玛光电科技有限公司 | Fast railway based on image vision invades limit identification monitoring system and method |
-
2020
- 2020-11-04 CN CN202011214199.XA patent/CN114529811A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004059900A2 (en) * | 2002-12-17 | 2004-07-15 | Evolution Robotics, Inc. | Systems and methods for visual simultaneous localization and mapping |
| 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 |
| CN208868062U (en) * | 2018-07-23 | 2019-05-17 | 中国安全生产科学研究院 | A kind of urban track traffic automatic tour inspection system |
| CN110217271A (en) * | 2019-05-30 | 2019-09-10 | 成都希格玛光电科技有限公司 | Fast railway based on image vision invades limit identification monitoring system and method |
Non-Patent Citations (2)
| Title |
|---|
| 于晓英;苏宏升;姜泽;董昱;: "基于YOLO的铁路侵限异物检测方法", 兰州交通大学学报, no. 02, 15 April 2020 (2020-04-15) * |
| 任新新;胡文韬;吕海翔;刘能;樊绍胜;: "电缆管道巡检清理机器人的研究与设计", 电力学报, no. 02, 25 April 2016 (2016-04-25) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115035060A (en) * | 2022-06-07 | 2022-09-09 | 贵州聚原数技术开发有限公司 | Tunnel wall deformation detection method based on computer image recognition |
| CN115892131A (en) * | 2023-02-15 | 2023-04-04 | 深圳大学 | Intelligent monitoring method and system for subway tunnel |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111958592B (en) | Image semantic analysis system and method for transformer substation inspection robot | |
| CN106808482B (en) | A kind of crusing robot multisensor syste and method for inspecting | |
| KR102065975B1 (en) | Safety Management System Using a Lidar for Heavy Machine | |
| CN106679567A (en) | Contact net and strut geometric parameter detecting measuring system based on binocular stereoscopic vision | |
| KR101602376B1 (en) | A train faulty monitoring system | |
| CN109829908B (en) | Binocular image-based method and device for detecting safety distance of ground object below power line | |
| CN112235537B (en) | Transformer substation field operation safety early warning method | |
| CN110602449A (en) | Intelligent construction safety monitoring system method in large scene based on vision | |
| CN114529811A (en) | Rapid and automatic identification and positioning method for foreign matters in subway tunnel | |
| CN102737236A (en) | Method for automatically acquiring vehicle training sample based on multi-modal sensor data | |
| CN112132896A (en) | Trackside equipment state detection method and system | |
| CN106162144A (en) | A kind of visual pattern processing equipment, system and intelligent machine for overnight sight | |
| CN111160220B (en) | Deep learning-based parcel detection method and device and storage medium | |
| CN110136186B (en) | Detection target matching method for mobile robot target ranging | |
| CN102788572A (en) | Method, device and system for measuring attitude of lifting hook of engineering machinery | |
| CN109986605A (en) | A kind of intelligence automatically tracks robot system and method | |
| CN115752462A (en) | Method, system, electronic equipment and medium for inspecting key inspection targets in building | |
| Sutjaritvorakul et al. | Data-driven worker detection from load-view crane camera | |
| CN112000094A (en) | Single-and-double-eye combined high-voltage transmission line hardware fitting online identification and positioning system and method | |
| Liu et al. | An approach for auto bridge inspection based on climbing robot | |
| CN110702016A (en) | Power transmission line icing measurement system and method | |
| CN107339950A (en) | A kind of track quick track switching operating operation sleeper bolt method for detecting position | |
| CN117197779A (en) | Track traffic foreign matter detection method, device and system based on binocular vision | |
| CN114812403B (en) | Large-span steel structure hoisting deformation monitoring method based on unmanned plane and machine vision | |
| CN117008622A (en) | Visual robot underwater target identification tracking method and underwater visual robot thereof |
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 |