CN109747681A - A kind of train positioning device and method - Google Patents
A kind of train positioning device and method Download PDFInfo
- Publication number
- CN109747681A CN109747681A CN201910032556.1A CN201910032556A CN109747681A CN 109747681 A CN109747681 A CN 109747681A CN 201910032556 A CN201910032556 A CN 201910032556A CN 109747681 A CN109747681 A CN 109747681A
- Authority
- CN
- China
- Prior art keywords
- train
- kilometre
- post
- image
- board
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of train positioning device and methods, positioning device includes kilometre post board, train top side camera, train side cameras and background server, kilometre post board is mounted on by line track, train top side camera is mounted on the top position of train head, train side cameras is mounted on the lateral location of train head, and background server is connect with train top side camera, train side cameras respectively.Compared with prior art, the present invention is based on convolutional neural networks, are positioned by image recognition to train position, while odometer can be cooperated to be corrected train position, and cost is relatively low for installation maintenance, and have superior accuracy and robustness.
Description
Technical field
The present invention relates to train Positioning Technology fields, more particularly, to a kind of train positioning device and method.
Background technique
Train positioning is a key technology of Train Detection and Identification, and important work is played in rail traffic signal system
With.Train position information is real-time and accurately obtained, is the guarantee that rail traffic is efficient, is safely operated, is to improve efficiency, play fortune
The premise of energy.Train operation organization is reduced by train positioning, is conducive to rail traffic and is sent out toward the direction of high density, large conveying quantity
Exhibition.
Currently used train Positioning Technology includes: the train positioning based on track circuit, and the train based on transponder is fixed
Position, train positioning based on global position system etc..Wherein, the train Positioning Technology based on track circuit is a kind of rough grade
Localization method, positioning accuracy depend entirely on track section length；Train positioning based on transponder is a kind of point type positioning,
Have the characteristics that strong interference immunity, without blind location area, the operation is stable, but in order to realize that train is continuously accurately positioned, needs in-orbit
It is laid with a large amount of transponders in road, therefore invests huge；It is high, easy to maintain using the train positioning accuracy of global position system, still
There are blind location areas in the places such as the place around more than obstacle, such as mountain forest, city, tunnel.
Train can obtain millimetre-sized continuous position using high precision odometer using odometer as the main foundation of positioning
Confidence breath due to wheel-slip, idle running or wheel wear etc., is easy to make odometer that there are accumulated errors but in practice.
In train control system, the error usually is eliminated using the track circuit separation or transponder with fixed exact position.
And high density laying transponder is expensive, and some routes do not have track circuit, can not lay transponder.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide a kind of train positioning devices
And method identifies kilometer post data information and train by kilometer post by carrying out discriminance analysis to kilometre post image
Know the opportunity of board, and train position positioning and correction are carried out with this.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of train positioning device, comprising: kilometre post board, train top side camera, train side cameras and backstage
Server, the kilometre post board are mounted on by line track, and the train top side camera is mounted on the top of train head
Position, the train side cameras are mounted on the lateral location of train head, and the background server is respectively and at the top of train
Video camera, the connection of train side cameras；
The kilometre post board, for marking kilometer post information；
The train top side camera, for capturing the image of kilometre post board；
The train side cameras passes through one group of consecutive image of kilometre post board for capturing train；
The background server, for receiving and handling the figure of train top side camera and the capture of train side cameras
Picture carries out image recognition and analysis later, judges that train passes through the opportunity of sign board, and send the kilometer of corrective command and correction
Data are marked to ATP/ATO.
Preferably, the train top side camera and train side cameras are vehicle-mounted high-speed camera.
Preferably, described image is identified as based on convolutional neural networks to the kilometre post board and kilometer post number in image
According to being identified respectively.
A kind of train locating method, comprising:
Step 1, image capture: train top side camera captures kilometre post board image A, and captured image A is transmitted
To background server, while train side cameras starts to capture one group of consecutive image SA that train passes through kilometre post board, it
One group of image SA of capture is also transferred to background server afterwards；
Step 2, image procossing: background server is respectively to kilometre post board image A and train by kilometre post board
One group of consecutive image SA is handled, and the image after handling is respectively kilometre post board image PA and train by kilometre post
One group of consecutive image PSA of board；
Step 3, kilometer post data identify: background server identifies the image PA of kilometre post board, if in image
Kilometer post data can be identified in PA, then follow the steps 4, otherwise return step 1；
Step 4, kilometre post board identifies: background server passes through one group of consecutive image PSA of kilometre post board to train
It is identified, if can identify kilometre post board in this group of image PSA, thens follow the steps 5, otherwise return step 1；
Step 5, train judges by the opportunity of kilometre post board: background server analyzes and determines that train passes through kilometre post
The opportunity of board；
Step 6, train position correction triggering: background server passes through the opportunity of kilometre post board according to train, sends school
Positive order and correction kilometer coupon give train ATP/ATO equipment.
Preferably, the identification kilometer post data in the step 3, specifically include:
Step 3.1, first sliding window is set on the image PA of kilometre post board, and the first sliding window is from the upper left of image PA
Angle starts gradually to move according to the step-length of setting from left to right, from top to bottom；
Step 3.2, based on each movement of the first sliding window, being identified in the first sliding window using convolutional neural networks whether there is
Kilometre post board obtains location information of the kilometre post board in image PA with this；
Step 3.3, according to the location information of acquisition, kilometer post data information is identified using convolutional neural networks.
Preferably, the identification kilometre post board in the step 4, specifically includes:
Step 4.1, second sliding window is set on one group of consecutive image PSA of the train by kilometre post board, successively
Last image is moved to from first image；
Step 4.2, it using convolutional neural networks, identifies and whether there is kilometre post board in the second sliding window.
Preferably, kilometre post board is recognized whether in the step 3.2 and step 4.2, there are kilometers for judgement
The following conditions that Sign Board need to meet simultaneously: kilometer post is completely contained in picture boundary, and the width of kilometre post board is greater than figure
Image width degree 60% and be less than the 80% of picture traverse, the height of kilometre post board is greater than the 60% of picture altitude and is less than image
The 87.5% of height.
Preferably, the convolutional neural networks in the step 3.3 and step 4.2, network model is by input layer, convolution
Layer, pond layer and output layer composition, the input layer include the first input layer and the second input layer, and the convolutional layer includes first
Convolutional layer, the second convolutional layer and third convolutional layer, the pond layer include the first pond layer, the second pond layer and third pond
Layer, the output layer include quantity be the first node of N, second node ... nth node.
Preferably, the first node is the output node that whether there is of kilometre post board, second node to nth node according to
The secondary output node for first character in kilometre post board data information to n-th character.
Compared with prior art, the invention has the following advantages:
1, train positioning device proposed by the present invention is laid with transponder compared to high density, and installation maintenance is at low cost, energy
It is enough to be used cooperatively with odometer, the supplementary means as train position error correction；
2, train locating method proposed by the present invention, uses convolutional neural networks to identify image, can eliminate not
The unfavorable factors bring negative effects such as, light variation complicated with scene under line condition, have superior accuracy and robust
Property；
3, present invention could apply to the region more than mountain forest, city, tunnel and surrounding obstacle, satellite positioning is compensated for
Haves the defects that blind area.
Detailed description of the invention
Fig. 1 is train positioning device structure schematic diagram of the invention；
Fig. 2 is train locating method flow chart of the invention；
Fig. 3 is the convolutional neural networks model structure of the embodiment of the present invention；
Fig. 4 is the training convergence curve figure of the convolutional neural networks of the embodiment of the present invention；
Fig. 5 is the test recognition accuracy curve graph of the convolutional neural networks of the embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is a part of the embodiments of the present invention, rather than whole embodiments.Based on this hair
Embodiment in bright, those of ordinary skill in the art's every other reality obtained without making creative work
Example is applied, all should belong to the scope of protection of the invention.
Embodiment one
As shown in Figure 1, train positioning device of the invention, comprising: kilometre post board 4, train top side camera 1, train
Side cameras 2 and background server 3, the kilometre post board 4 are mounted on by line track, the train top side camera 1
It is mounted on the top position of train head, the train side cameras 2 is mounted on the lateral location of train head, the backstage
Server 3 is connect with train top side camera 1, train side cameras 2 respectively.
The kilometre post board, for marking kilometer post information, Sign Board will be fabricated to 3 face bodies or tetrahedron, have simultaneously
Image is captured conducive to top and lateral video camera；
The train top side camera is responsible for discovery sign board, captures the image of distinguishing mark board.If in the feelings of low speed
Under condition, such as only in standing when train positioning, top side camera can be not provided with；
The train side cameras passes through one group of consecutive image of kilometre post board for capturing train, is responsible for capture
Judge train by the image on the opportunity of sign board；
The background server, for receiving and handling the figure of train top side camera and the capture of train side cameras
Picture carries out image recognition and analysis later, and carries out data communication with train ATP/ATO equipment；
Further, the train top side camera and train side cameras are vehicle-mounted high-speed camera,；
Further, described image is identified as based on convolutional neural networks to the kilometre post board and kilometer post in image
Data are identified respectively；
Further, the data communication is to send corrective command and correction kilometer post data.
The principle of train positioning of the present invention is specific as follows:
Train is positioned main foundation using odometer as train and is corrected when by kilometer post using kilometer post, is offset
The error as caused by wheel-slip, idle running.Top side camera finds witness marker bridge queen, and captured image is transferred to backstage
Server identification, while lateral high-speed camera being notified to start to work, judge that train passes through the opportunity of sign board, and send correction
The kilometer post data of order and correction are to ATP/ATO.
Image capture of the present invention and pretreatment specifically:
The image for obtaining high quality is the premise for utilizing framing.Train running speed is fast, and car body shakes in addition, be
Obtain clear image on the train of traveling, time for exposure of camera short enough (must can use high speed video system, when exposure
Between be less than 10us) obtain picture.In practical application, due to be illuminated by the light and external condition limitation, camera obtain picture quality
It has differences.In order to weaken the influence that illumination handles subsequent image, system carries out gray proces to image.In complex environment
Under, such as night, the picture quality that camera obtains is still poor, this can be made up by illumination and automatic diaphragm lens.
Kilometer post identification of the present invention specifically:
Kilometer post is identified using CNN network.After the picture that system obtains video camera capture, it is arranged on picture
The sliding window of one (128*64) size, moves down to the right since the upper left corner of picture, and mobile 8 pixels, utilize cunning every time
Window judges position of the kilometer post in picture, and with the letter and number of kilometer post in trained CNN Network Recognition sliding window.It deposits
In the Rule of judgment of kilometer post are as follows: (1) kilometer post is completely contained in picture boundary.(2) it is wide to be less than picture for the width of kilometer post
The 80% of degree, and the height of kilometer post is less than the 87.5% of picture height.(3) width of kilometer post is greater than picture width
60% or kilometer post height be greater than picture height 60%.
Training network is made of 3 convolutional layers and 3 pooling layers.The convolution kernel size of 1st convolutional layer is 5 × 5,
And there are 32 outputs.Similarly, the convolution kernel size of the 2nd convolution is also 5 × 5 and has 64 output figures.The last one convolution
Layer has 128 output figures, and convolution kernel size is 5 × 5.In succession after each convolutional layer 2 × 2 Max-pooling layer.Input picture
The size of slice is 128 × 64, they become 128 × 64 after the 1st convolutional layer, become 64 after the 1st pooling layers
× 32, it moves in circles.
Output layer has a node (left side) to be used as the indicator that kilometer post whether there is；Remaining node is used to
Encode the probability of a specific kilometer post.Each column in output layer are consistent with each number in kilometer post, each section
Point provides the probability being consistent with existing character.For example, being located at the 2nd node for arranging the 2nd row provides in kilometer post the second number
Code is the probability of character 1.
It detects network and the difference of network is trained to be to use convolutional layer rather than full articulamentum last two layers, in this way
The input picture size for detecting network can be made to be not limited only to 128*64.One complete picture is put into a kind of specific dimensions
In network, then returns to each node and possess one there are the pictures of character probabilities value.Because adjacent sliding window can be shared
Many convolution features, so these particular pictures are involved in the same network can be to avoid repeatedly calculating same feature.
Model described herein is only applicable to specific kilometer post form.Especially, network structure clearly assumes output only 7
A character, and it is only applicable to specific font.Therefore, it when designing kilometer post mark, needs kilometer post numerical value length standard
To change, a high position mends 0 when curtailment, meanwhile, kilometer post must use standard letter.
The present invention corrects trigger timing specifically:
Train is captured by the time of kilometer post using one group of continuous picture of high-speed camera shooting, works as figure
Occurs kilometer post mark in piece group, then it is assumed that train head is aligned with kilometer post, is corrected immediately to train position.For
It avoids accidentally correcting, host judges whether front video also detects that kilometer post after receiving corrective command, and has succeeded
Identification.Since lateral video camera only needs to judge that kilometer post whether there is, identification particular content is not needed, in order to improve identification effect
Output layer is revised as 1 node by rate, indicates to include suitable kilometer post sign board in picture.
As shown in Fig. 2, train locating method of the invention, comprising:
Step 1, image capture: train top side camera captures kilometre post board image A, and captured image A is transmitted
To background server, while train side cameras starts to capture one group of consecutive image SA that train passes through kilometre post board, it
One group of image SA of capture is also transferred to background server afterwards；
Step 2, image procossing: background server is respectively to kilometre post board image A and train by kilometre post board
One group of consecutive image SA is handled, and the image after handling is respectively kilometre post board image PA and train by kilometre post
One group of consecutive image PSA of board；
Step 3, kilometer post data identify: background server identifies the image PA of kilometre post board, if in image
Kilometer post data can be identified in PA, then follow the steps 4, otherwise return step 1；
Step 4, kilometre post board identifies: background server passes through one group of consecutive image PSA of kilometre post board to train
It is identified, if can identify kilometre post board in this group of image PSA, thens follow the steps 5, otherwise return step 1；
Step 5, train judges by the opportunity of kilometre post board: background server analyzes and determines that train passes through kilometre post
The opportunity of board；
Step 6, train position correction triggering: background server passes through the opportunity of kilometre post board according to train, sends school
Positive order and correction kilometer coupon give train ATP/ATO equipment.
Identification kilometer post data in the step 3, specifically include:
Step 3.1, first sliding window is set on the image PA of kilometre post board, and the first sliding window is from the upper left of image PA
Angle starts gradually to move according to certain step-length from left to right, from top to bottom；
Step 3.2, based on each movement of the first sliding window, being identified in the first sliding window using convolutional neural networks whether there is
Kilometre post board obtains location information of the kilometre post board in image PA with this；
Step 3.3, according to the location information of acquisition, kilometer post data information is identified using convolutional neural networks.
Identification kilometre post board in the step 4, specifically includes:
Step 4.1, second sliding window is set on one group of consecutive image PSA of the train by kilometre post board, successively
Last image is moved to from first image；
Step 4.2, it using convolutional neural networks, identifies and whether there is kilometre post board in the second sliding window.
Kilometre post board is recognized whether in the step 3.2 and step 4.2, and there are kilometre post boards for judgement
Condition are as follows: the width of kilometre post board is greater than the 60% of picture traverse and is less than 80% or kilometre post board of picture traverse
Height is greater than the 60% of picture altitude and is less than the 87.5% of picture altitude.
As shown in figure 3, the present invention be used for image recognition convolutional neural networks, network model by input layer, convolutional layer,
Pond layer and output layer composition, the input layer includes the first input layer IN1 and the second input layer IN2, the convolutional layer include
First convolutional layer CON1, the second convolutional layer CON2 and third convolutional layer CON3, the pond layer include the first pond layer POOL1,
Second pond layer POOL2 and third pond layer POOL3, the output layer include first node Node1, the second section that quantity is N
Point Node2 ... nth node NodeN；
Wherein, first node Node1 is the output node that kilometre post board whether there is, second node Node2 to N section
Point NodeN is followed successively by kilometre post board data information first character to the output node of n-th character.
Specifically, the input of the first input layer IN1 is the image PA, the second input layer IN2 for the kilometre post board handled well
Input be one group of consecutive image PSA that the train handled well passes through kilometre post board, the output of the first input layer IN1 connects respectively
It is connected to the input of the first convolutional layer CON1, the second convolutional layer CON2 and third convolutional layer CON3, the output point of the second input layer IN2
It is not connected to the input of the first convolutional layer CON1, the second convolutional layer CON2 and third convolutional layer CON3, the first convolutional layer CON1's
Output is connected to the input of the first pond layer POOL1, and the output of the second convolutional layer CON2 is connected to the defeated of the second pond layer POOL2
Enter, the output of third convolutional layer CON3 is connected to the input of third pond layer POOL3, the first pond layer POOL1, the second pond layer
The output of POOL2 and third pond layer POOL3 is all connected to the input of first node Node1, the output point of first node Node1
It is not connected to the input of second node Node2 to nth node NodeN.
The present invention is used for the convolutional neural networks of image recognition, specific training process are as follows:
The training pictures that number is 100,000 and the survey that number is 50 are obtained by the method for simulating actual scene first
Try pictures；
Then according to the data of training pictures, 30 hours of training, training convergence curve is as shown in figure 4, final receive
It holds back curve and levels off to 0；
Trained convolutional neural networks are finally utilized, image recognition are carried out to test pictures, recognition accuracy is bent
As figure 5 illustrates, recognition accuracy mean value is 97.6% to line.
Wherein, in order to which the light for simulating actual scene changes, the kilometer post information and background color on kilometre post board are equal
It is randomly selected, in addition, being used a kind of based on random rolling to simulate the captured various situations of kilometre post board image
Turn, inclination, deflection, the affine transformation method for translating and scaling, so that kilometre post board converts.
In actual application, when train passes through kilometer post, high speed camera meeting continuous capturing to multiple pictures, the device
The kilometer post that will identify that is classified by kilometer post numerical value (unidentified is counted by full 0), and only discrimination is more than certain
Threshold value just judges to identify successfully, as basis on location, to make up the case where individual picture recognitions fail.Simultaneously because in train row
During sailing, need quickly to handle a large amount of picture, it can be by introducing Haar cascade, the method for HOG detector, parallel plan
The equipment omited and use higher performance improves detection efficiency, meets the requirements testing result.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (9)
1. a kind of train positioning device characterized by comprising kilometre post board, train top side camera, the camera shooting of train side
Machine and background server, the kilometre post board are mounted on by line track, and the train top side camera is mounted on train vehicle
The top position of head, the train side cameras are mounted on the lateral location of train head, the background server respectively with
Train top side camera, the connection of train side cameras；
The kilometre post board, for marking kilometer post information；
The train top side camera, for capturing the image of kilometre post board；
The train side cameras passes through one group of consecutive image of kilometre post board for capturing train；
The background server, for receiving and handling train top side camera and train side cameras captured image, it
Image recognition and analysis are carried out afterwards, judge that train passes through the opportunity of sign board, and send the kilometer post number of corrective command and correction
According to ATP/ATO.
2. a kind of train positioning device according to claim 1, which is characterized in that the train top side camera and train
Side cameras is vehicle-mounted high-speed camera.
3. a kind of train positioning device according to claim 1, which is characterized in that described image is identified as based on convolution mind
Through network in image kilometre post board and kilometer post data identify respectively.
4. a kind of train locating method characterized by comprising
Step 1, image capture: train top side camera captures kilometre post board image A, and after captured image A is transferred to
Platform server, while train side cameras starts to capture one group of consecutive image SA, Zhi Houye that train passes through kilometre post board
One group of image SA of capture is transferred to background server；
Step 2, image procossing: background server passes through one group of kilometre post board to kilometre post board image A and train respectively
Consecutive image SA is handled, and the image after handling is respectively kilometre post board image PA and train by kilometre post board
One group of consecutive image PSA；
Step 3, kilometer post data identify: background server identifies the image PA of kilometre post board, if in image PA
It can identify kilometer post data, then follow the steps 4, otherwise return step 1；
Step 4, kilometre post board identifies: background server carries out train by one group of consecutive image PSA of kilometre post board
Identification, if can identify kilometre post board in this group of image PSA, thens follow the steps 5, otherwise return step 1；
Step 5, train judges by the opportunity of kilometre post board: background server analyzes and determines train by kilometre post board
Opportunity；
Step 6, train position correction triggering: background server passes through the opportunity of kilometre post board according to train, sends correction life
It enables and correction kilometer coupon gives train ATP/ATO equipment.
5. a kind of train locating method according to claim 4, which is characterized in that the identification kilometer post in the step 3
Data specifically include:
Step 3.1, first sliding window is set on the image PA of kilometre post board, and the first sliding window is opened from the upper left corner of image PA
Beginning gradually moves according to the step-length of setting from left to right, from top to bottom；
Step 3.2, based on each movement of the first sliding window, being identified using convolutional neural networks whether there is kilometer in the first sliding window
Sign Board obtains location information of the kilometre post board in image PA with this；
Step 3.3, according to the location information of acquisition, kilometer post data information is identified using convolutional neural networks.
6. a kind of train locating method according to claim 4, which is characterized in that the identification kilometer post in the step 4
Know board, specifically include:
Step 4.1, second sliding window is set on one group of consecutive image PSA of the train by kilometre post board, successively from the
One image is moved to last image；
Step 4.2, it using convolutional neural networks, identifies and whether there is kilometre post board in the second sliding window.
7. a kind of train locating method according to claim 5 or 6, which is characterized in that the step 3.2 and step 4.2
In recognize whether kilometre post board, there are the following conditions that kilometre post board need to meet simultaneously for judgement: kilometer post is complete
It is included in picture boundary entirely, the width of kilometre post board is greater than the 60% of picture traverse and is less than the 80% of picture traverse, public affairs
In Sign Board height be greater than picture altitude 60% and be less than picture altitude 87.5%.
8. a kind of train locating method according to claim 5 or 6, which is characterized in that the step 3.3 and step 4.2
In convolutional neural networks, network model is made of input layer, convolutional layer, pond layer and output layer, and the input layer includes
First input layer and the second input layer, the convolutional layer include the first convolutional layer, the second convolutional layer and third convolutional layer, the pond
Changing layer includes the first pond layer, the second pond layer and third pond layer, the output layer include quantity for N first node, the
Two nodes ... nth node.
9. a kind of train locating method according to claim 8, which is characterized in that the first node is kilometre post board
The output node that whether there is, second node to nth node are followed successively by kilometre post board data information first character to N
The output node of a character.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910032556.1A CN109747681A (en) | 2019-01-14 | 2019-01-14 | A kind of train positioning device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910032556.1A CN109747681A (en) | 2019-01-14 | 2019-01-14 | A kind of train positioning device and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109747681A true CN109747681A (en) | 2019-05-14 |
Family
ID=66405476
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910032556.1A Pending CN109747681A (en) | 2019-01-14 | 2019-01-14 | A kind of train positioning device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109747681A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110194200A (en) * | 2019-05-31 | 2019-09-03 | 唐智科技湖南发展有限公司 | It is a kind of run vehicle kilometer post determine method, apparatus and medium |
| CN110287897A (en) * | 2019-06-27 | 2019-09-27 | 清华大学 | A visual positioning system for rail trains based on deep learning |
| CN110641522A (en) * | 2019-09-30 | 2020-01-03 | 中铁第四勘察设计院集团有限公司 | Train positioning system of high-speed magnetic levitation track traffic |
| CN111762237A (en) * | 2020-06-29 | 2020-10-13 | 交控科技股份有限公司 | Rail transit train positioning method, device and system |
| CN112249089A (en) * | 2020-09-27 | 2021-01-22 | 卡斯柯信号有限公司 | A rail transit emergency positioning system and method |
| CN112550377A (en) * | 2020-12-18 | 2021-03-26 | 卡斯柯信号有限公司 | Rail transit emergency positioning method and system based on video identification and IMU (inertial measurement Unit) equipment |
| CN114283399A (en) * | 2021-12-24 | 2022-04-05 | 深圳龙岗智能视听研究院 | A low-cost vehicle localization method based on road sign detection |
| CN115402380A (en) * | 2022-09-29 | 2022-11-29 | 天津市英贝特航天科技有限公司 | A subway train benchmarking device and control method |
| US20230087606A1 (en) * | 2021-09-23 | 2023-03-23 | Cattron North America, Inc. | Vision-Based Systems And Methods For Locomotive Control And/Or Location Determination |
| CN116208258A (en) * | 2023-02-28 | 2023-06-02 | 中铁检验认证中心有限公司 | Self-adaptive positioning method and device for transponder transmission system |
| CN117341771A (en) * | 2023-08-24 | 2024-01-05 | 浙江众合科技股份有限公司 | A train positioning system and method based on image recognition and speed integration |
| CN119290004A (en) * | 2024-12-12 | 2025-01-10 | 福建城建智控科技有限公司 | A route planning method and device based on subway tunnel inspection |
| CN119705556A (en) * | 2024-12-12 | 2025-03-28 | 卡斯柯信号有限公司 | A visual positioning device, method, equipment and medium for train in tunnel |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB9012366D0 (en) * | 1990-06-02 | 1990-07-25 | Solvit Scient Engineers Limite | Automatic system to read barcode signs from or on a moving vehicle |
| WO2007091072A1 (en) * | 2006-02-07 | 2007-08-16 | Richard Shenton | System for measuring speed and/or position of a train |
| JP2008056179A (en) * | 2006-09-01 | 2008-03-13 | Toshiba Corp | Train position detection device |
| US20090080697A1 (en) * | 2005-04-25 | 2009-03-26 | Geo Technical Laboratory Co., Ltd. A Corporation | Imaging position analyzing method |
| JP2009250718A (en) * | 2008-04-03 | 2009-10-29 | Nissan Motor Co Ltd | Vehicle position detecting apparatus and vehicle position detection method |
| JP2012011867A (en) * | 2010-06-30 | 2012-01-19 | Mitsubishi Heavy Ind Ltd | Vehicle position calculation system and vehicle position calculation method, and program therefor |
| CN102829763A (en) * | 2012-07-30 | 2012-12-19 | 中国人民解放军国防科学技术大学 | Pavement image collecting method and system based on monocular vision location |
| CN105354568A (en) * | 2015-08-24 | 2016-02-24 | 西安电子科技大学 | Convolutional neural network based vehicle logo identification method |
| CN107600116A (en) * | 2017-09-21 | 2018-01-19 | 湖南中车时代通信信号有限公司 | A kind of rail truck localization method and system based on video identification and RFID |
| CN107679452A (en) * | 2017-08-28 | 2018-02-09 | 中国电子科技集团公司第二十八研究所 | Goods train license number real-time identifying system based on convolutional neural networks under big data |
| DE102017109944A1 (en) * | 2017-05-09 | 2018-11-15 | Connaught Electronics Ltd. | Automatic alignment with a petrol pump |
-
2019
- 2019-01-14 CN CN201910032556.1A patent/CN109747681A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB9012366D0 (en) * | 1990-06-02 | 1990-07-25 | Solvit Scient Engineers Limite | Automatic system to read barcode signs from or on a moving vehicle |
| US20090080697A1 (en) * | 2005-04-25 | 2009-03-26 | Geo Technical Laboratory Co., Ltd. A Corporation | Imaging position analyzing method |
| WO2007091072A1 (en) * | 2006-02-07 | 2007-08-16 | Richard Shenton | System for measuring speed and/or position of a train |
| JP2008056179A (en) * | 2006-09-01 | 2008-03-13 | Toshiba Corp | Train position detection device |
| JP2009250718A (en) * | 2008-04-03 | 2009-10-29 | Nissan Motor Co Ltd | Vehicle position detecting apparatus and vehicle position detection method |
| JP2012011867A (en) * | 2010-06-30 | 2012-01-19 | Mitsubishi Heavy Ind Ltd | Vehicle position calculation system and vehicle position calculation method, and program therefor |
| CN102829763A (en) * | 2012-07-30 | 2012-12-19 | 中国人民解放军国防科学技术大学 | Pavement image collecting method and system based on monocular vision location |
| CN105354568A (en) * | 2015-08-24 | 2016-02-24 | 西安电子科技大学 | Convolutional neural network based vehicle logo identification method |
| DE102017109944A1 (en) * | 2017-05-09 | 2018-11-15 | Connaught Electronics Ltd. | Automatic alignment with a petrol pump |
| CN107679452A (en) * | 2017-08-28 | 2018-02-09 | 中国电子科技集团公司第二十八研究所 | Goods train license number real-time identifying system based on convolutional neural networks under big data |
| CN107600116A (en) * | 2017-09-21 | 2018-01-19 | 湖南中车时代通信信号有限公司 | A kind of rail truck localization method and system based on video identification and RFID |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110194200B (en) * | 2019-05-31 | 2021-06-11 | 唐智科技湖南发展有限公司 | Method, device and medium for determining kilometer post of running vehicle |
| CN110194200A (en) * | 2019-05-31 | 2019-09-03 | 唐智科技湖南发展有限公司 | It is a kind of run vehicle kilometer post determine method, apparatus and medium |
| CN110287897A (en) * | 2019-06-27 | 2019-09-27 | 清华大学 | A visual positioning system for rail trains based on deep learning |
| CN110287897B (en) * | 2019-06-27 | 2021-05-14 | 清华大学 | Rail train visual positioning system based on deep learning |
| CN110641522A (en) * | 2019-09-30 | 2020-01-03 | 中铁第四勘察设计院集团有限公司 | Train positioning system of high-speed magnetic levitation track traffic |
| CN111762237A (en) * | 2020-06-29 | 2020-10-13 | 交控科技股份有限公司 | Rail transit train positioning method, device and system |
| CN112249089A (en) * | 2020-09-27 | 2021-01-22 | 卡斯柯信号有限公司 | A rail transit emergency positioning system and method |
| CN112550377A (en) * | 2020-12-18 | 2021-03-26 | 卡斯柯信号有限公司 | Rail transit emergency positioning method and system based on video identification and IMU (inertial measurement Unit) equipment |
| US20230087606A1 (en) * | 2021-09-23 | 2023-03-23 | Cattron North America, Inc. | Vision-Based Systems And Methods For Locomotive Control And/Or Location Determination |
| US12371081B2 (en) * | 2021-09-23 | 2025-07-29 | Cattron North America, Inc. | Vision-based systems and methods for locomotive control and/or location determination |
| CN114283399A (en) * | 2021-12-24 | 2022-04-05 | 深圳龙岗智能视听研究院 | A low-cost vehicle localization method based on road sign detection |
| CN115402380A (en) * | 2022-09-29 | 2022-11-29 | 天津市英贝特航天科技有限公司 | A subway train benchmarking device and control method |
| CN116208258A (en) * | 2023-02-28 | 2023-06-02 | 中铁检验认证中心有限公司 | Self-adaptive positioning method and device for transponder transmission system |
| CN117341771A (en) * | 2023-08-24 | 2024-01-05 | 浙江众合科技股份有限公司 | A train positioning system and method based on image recognition and speed integration |
| CN119290004A (en) * | 2024-12-12 | 2025-01-10 | 福建城建智控科技有限公司 | A route planning method and device based on subway tunnel inspection |
| CN119705556A (en) * | 2024-12-12 | 2025-03-28 | 卡斯柯信号有限公司 | A visual positioning device, method, equipment and medium for train in tunnel |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109747681A (en) | A kind of train positioning device and method | |
| CN104599249B (en) | Cableway platform bridge floor car load is distributed real-time detection method | |
| CN107563419B (en) | Train positioning method combining image matching and two-dimensional code | |
| CN112581795B (en) | A video-based real-time early warning method and system for bridge and ship-to-ship collision | |
| CN111145545A (en) | Road traffic behavior unmanned aerial vehicle monitoring system and method based on deep learning | |
| CN106530794B (en) | Method and system for automatic identification and calibration of driving road | |
| CN102208013A (en) | Scene matching reference data generation system and position measurement system | |
| DE102012215533A1 (en) | Method for determining position of rail vehicle by satellite navigation system, involves providing receiver unit in retainer at track, and comparing recorded data in retainer with stored data of landmarks | |
| CN109509223A (en) | Front vehicles distance measuring method based on deep learning | |
| CN106092123B (en) | A kind of video navigation method and device | |
| CN106767854A (en) | mobile device, garage map forming method and system | |
| CN113591643A (en) | Underground vehicle station entering and exiting detection system and method based on computer vision | |
| CN104794425A (en) | Vehicle counting method based on movement track | |
| CN119492370A (en) | High-precision railway map generation method and system based on one-way trajectory correction | |
| CN107506753B (en) | Multi-vehicle tracking method for dynamic video monitoring | |
| CN112550377A (en) | Rail transit emergency positioning method and system based on video identification and IMU (inertial measurement Unit) equipment | |
| CN112765392B (en) | High-speed rail train control positioning method and system based on image matching | |
| CN113343782B (en) | A highway sign detection method based on UAV remote sensing | |
| CN103605960B (en) | A kind of method for identifying traffic status merged based on different focal video image | |
| CN114037750B (en) | A method for realizing a track virtual balise | |
| CN120766544A (en) | Vehicle speed measurement method during UAV highway inspection based on UAV nest | |
| CN119442145A (en) | A multi-target detection method for complex traffic scenes | |
| CN114217626B (en) | Railway engineering detection method and system based on unmanned aerial vehicle routing inspection video | |
| CN113706617A (en) | Multi-track train positioning method, system, background server and readable storage medium | |
| Belov et al. | Railway traffic lights recognition system |
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: 20190514 |