CN106778569A - Train preceding object object detecting method based on video image - Google Patents
Train preceding object object detecting method based on video image Download PDFInfo
- Publication number
- CN106778569A CN106778569A CN201611103510.7A CN201611103510A CN106778569A CN 106778569 A CN106778569 A CN 106778569A CN 201611103510 A CN201611103510 A CN 201611103510A CN 106778569 A CN106778569 A CN 106778569A
- Authority
- CN
- China
- Prior art keywords
- obstacle
- image
- track
- occupied
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Train Traffic Observation, Control, And Security (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of train preceding object object detecting method based on video image, its implementation process is：Pre-processing image data is carried out to captured image first；Secondly, track fitting detection is carried out according to binary image, and set up detection image window；Then, anti-perspective transform treatment is carried out to pretreated image, the binary image after analyzing anti-perspective transform treatment is arranged image to cumulative；Transverse area is divided according to peak value simultaneously, row is done to sectional image to cumulative, it is comprehensive to determine the parameter that barrier is occupied；Finally, carry out converse perspective transform and mark barrier particular location.This method can effectively detect that obstacle is occupied in real time for the railway detection of obstacles under straight way scene, and judge the barrer types, provide the information such as obstacle location and dimensional parameters, effectively reduce difficulty and the danger of manual detection, and improve efficiency.
Description
Technical field
The present invention relates to a kind of train preceding object object detecting method based on video image, belong to image processing techniques neck
Domain.
Background technology
As the passenger-cargo freight volume of China railways is continuously increased, station train reception and departure quantity and shunting service amount also increase therewith,
The outer casualty accident of railway is in growing trend, and railway barrier badly influences the operation safety of train.So, research train front
The technology of detection of obstacles provides effective technological approaches to ensure the harm that train operating safety and reduction accident occur, and has
Standby significant academic theory and social benefit are worth.According to prior art, train preceding object analyte detection and other environment
Detection of obstacles is divided into active detecting and passive detection.
Active detecting：I.e. some form of signal is sent to orientation to be detected, and the signal for reflecting is examined by sensor
Survey, thus detect barrier.Active detecting mainly includes detections of radar, ultrasound examination and laser detection, and algorithm is simple, holds
Easily realize, amount of calculation is small, is protected from weather influences and accurately detects the position of barrier, but there is also some shortcomings, such as detect
Distance it is limited, coverage rate is limited, resolution ratio is not high, and detections of radar principle is close with laser detection, there is also spatial coverage
Limited, resolution ratio problem not high, active detecting is a kind of infringement formula detection in addition, ambient noise is improve, between sensor
Also interference can be produced.
Passive detection：It is mainly a kind of image detection based on computer vision, has many excellent compared to active detecting
Point, space covering is wide, does not increase ambient noise, is detected in non-infringement mode, and interference etc. will not be produced between imageing sensor.But
It is that there is also defect, this method will fail under the conditions of illumination not enough dense fog, night etc., computationally intensive, and algorithm is complicated, no
Easily realize.Along with computer software and hardware technological development, progressively improve and realized difficult situation.
At present, the judgement for train preceding object thing is mainly still manually completed.Automatic obstacle is carried out using image
Detection there is problems and need further solution：
1) conventional obstacle detection method is judged using image statisticses feature mostly, it is impossible to quickly determine obstacle location
With type；
2) when mounted in order to be able to the wider image that collects must have certain height and angle, this is adopted video camera
The process for collecting picture have passed through perspective transform, and object that will be in three-dimensional world is transformed into two dimensional image.Perspective transform is deposited
Causing rail that original parallel relation is lost in picture, be changed into intersecting lines, the image distal end that is spaced in of rail is becoming
Very little is obtained, this brings difficulty to the barrier of segmentation diverse location and size.
3) under train obstacle detection application scenarios, track structure feature is obvious, in conventional road barricade detection technique
On the basis of, if using track structure feature as priori, being occupied in obstacle can be further in detection and obstacle positioning
Improve performance.
The content of the invention
To solve the deficiencies in the prior art, it is an object of the invention to provide a kind of train front barrier based on video image
Hinder object detecting method, obstacle location can rapidly and accurately be judged with type according to acquired image.
In order to realize above-mentioned target, the present invention is adopted the following technical scheme that：
A kind of train preceding object object detecting method based on video image, it is characterized in that, comprise the following steps：
1) video image M0 is pre-processed, obtains pretreated binary image M1；
2) obstacle detection window is set up：According to the rectilinear orbit aspect of model, track fitting detection is carried out to M1 images, according to
Track detecting result, intercepts the detection window image M2 based on orbital region；
3) according to two in image M2 the four of track extreme coordinates, image M2 is carried out instead according to anti-perspective transform algorithm
Perspective transform obtains image M3；
4) vertical and horizontal respectively to image M3 pixel values add up, and judge whether that obstacle is occupied, and to barrier
Hinder to occupy and positioned；
5) occupying size according to obstacle carries out obstacle danger early warning；
6) according to step 5) result in video image in real time mark obstacle occupy region.
Further, the step 1) in particular content be：A two field picture M0 is extracted from video successively, to image M0
Noise reduction, enhancing treatment are carried out, rim detection and binary conversion treatment then are carried out to enhanced image, color video frequency image is turned
Change black white image into, Morphological scale-space is carried out to the image after binaryzation using corrosion and expanding method successively, so that further
Discrete noise is filtered, pretreated binary image M1 is finally given.
Further, the step 4) concretely comprise the following steps：
401) longitudinal direction is done to image M3 cumulative, obtains cumulative projection of the image in X-axis, according to the sharp peaks characteristic of projection,
It is three detection zones by image M3 points：Orbital region, track exterior lateral area, track inside region；
402) according to provincial characteristics, judge to be occupied with the presence or absence of obstacle in region：
In the case of accessible, the feature of three detection zones is：
Orbital region, two track longitudinal direction accumulated values should have close peak value, and highly be matched with detection window；
Track exterior lateral area, longitudinal accumulated value is flat without saltus step；
Track inside region, longitudinal accumulated value is flat without saltus step；
According to features above, the obstacle of detection is divided into two classes, the first kind is that the obstacle on the inside of track and outside is occupied, the
Two classes are that track blocks obstacle；If track inner side and outer side region has obstacle and occupies, will there is saltus step in longitudinal accumulated value,
Hopping amplitude is relevant with the longitudinal size that obstacle is occupied；If track has obstacle occupied, track regions longitudinal direction accumulated value will be less than
The height value of detection window, it is relevant less than part and the track area size that is blocked；
If 403) three detection zones have obstacle and occupy, the X-axis coordinate that disturbance in judgement is occupied is interval：
Firstly, for orbital region image, do binaryzation and negate treatment, it is 1 that track blocks barrier pixel value, and normal
The pixel value of track regions is 0；Then, the longitudinal direction of original track regions is substituted with orbital region binaryzation longitudinal accumulated value of the inverted
Accumulated value；Finally, the size requirement for being occupied according to detection obstacle, the longitudinal accumulated value transition detection thresholding of setting, for beyond door
The hop region of limit extracts its lateral coordinates interval { (Sxn,Exn), n=1,2 ..., N, wherein, SxnRepresent n-th saltus step area
Initial x-axis coordinate, ExnN-th termination x-axis coordinate in saltus step area is represented, N represents saltus step area number, that is, there is obstacle and occupy
Lateral coordinates interval number；
404) occupying interval for obstacle carries out laterally adding up respectively, judges that the Y-axis coordinate that each obstacle is occupied is interval：From
From left to right, the obstacle successively in interception image M3 occupies the interval image A of lateral coordinatesn, n=1,2 ..., N；
Truncated picture is carried out respectively laterally to add up, cumulative projection of the image in Y-axis is obtained, it is cumulative according to each image
The saltus step of value, can respectively obtain each lateral coordinates interval (Sxn,Exn) on Y-axis coordinate the interval { (Sy that occupies of obstacleN, k,
EyN, k), n=1,2 ..., N, k=1,2 ..., Kn, wherein Syn,kAnd Eyn,kLateral coordinates interval (Sx is represented respectivelyn,Exn) on
Starting and terminate y-axis coordinate, K that k-th obstacle is occupiednRepresent lateral coordinates interval (Sxn,Exn) present on obstacle occupy
Number；The obstacle for more than detecting occupies sum
Further, the step 5) particular content is：
It is rectangular area arrangement set { R1 to define obstacle and occupyn,k, R1n,kRepresent n-th lateral coordinates interval (Sxn,
Exn) on k-th obstacle occupy rectangular area, two diagonal point coordinates of rectangular area are (Sxn, SyN, k) and (Exn,
EyN, k), region R1 is occupied by obstaclen,kArea carry out the danger classes that occupies of disturbance in judgement according to sequence from big to small.
Further, the step 6) specific steps：Rectangular area arrangement set is occupied to the obstacle in image M3
{R1n,kConverse perspective transform is carried out, and obstacle is occupied into coordinate and track blocks obstacle coordinate and is mapped in image M2, mark out
Barrier zone, and the parameter of obstacle is exported in real time, it is that train automatic danger-avoiding or manual intervention provide reference.
Further, the parameter of the obstacle includes number, type and the size of obstacle.
The beneficial effect that the present invention is reached：(1) algorithm that the present invention is given, for the railway barrier under straight way scene
Detection, can effectively detect that obstacle is occupied in real time, and judge the barrer types, provide the letter such as obstacle location and dimensional parameters
Breath, effectively reduces difficulty and the danger of manual detection, and improve efficiency；(2) present invention is using in track regions, track
Side and the different structure feature of exterior lateral area, subregion disturbance in judgement are occupied.Processed by analyzing anti-perspective, effectively eliminated figure
As the influence that vision deformation is extracted to Obstacle Position and dimensional parameters, barrier can be more accurately confirmed compared with prior art
Position and size.
Brief description of the drawings
Fig. 1 is algorithm flow chart of the invention；
Fig. 2 is train operation ahead video image in test site；
Fig. 3 is by pretreated detection window image；
Fig. 4 is that the image after anti-perspective transform is carried out to detection window image；
Fig. 5 is longitudinal accumulation result of image pixel value after anti-perspective transform；
Fig. 6 is image pixel value subregion transverse direction accumulation result after anti-perspective transform, and wherein Fig. 6 (a) is horizontal left rail area
To pixel value accumulation result, Fig. 6 (b) is track inner side horizontal pixel value accumulation result, and Fig. 6 (c) is left rail area transverse direction picture
Element value accumulation result；
Fig. 7 is detection window image obstacle marking result.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following examples are only used for clearly illustrating the present invention
Technical scheme, and can not be limited the scope of the invention with this.
A kind of train preceding object object detecting method based on video image involved in the present invention, specifically includes following step
Suddenly：
1) video image M0 is pre-processed, extracts a two field picture M0 from video successively, image M0 is carried out noise reduction,
Enhancing is processed, and then carries out rim detection and binary conversion treatment to enhanced image, and color video frequency image is converted into black and white
Image, carries out Morphological scale-space to the image after binaryzation using corrosion and expanding method successively, discrete so as to further filter out
Noise, finally gives pretreated binary image M1.
2) obstacle detection window is set up：According to the rectilinear orbit aspect of model, track fitting detection is carried out to M1 images, according to
Track detecting result, intercepts the detection window image M2 based on orbital region；
3) according to two in image M2 the four of track extreme coordinates, image M2 is carried out instead according to anti-perspective transform algorithm
Perspective transform obtains image M3；
4) vertical and horizontal respectively to image M3 pixel values add up, and judge whether that obstacle is occupied, and to barrier
Hinder to occupy and positioned.Comprise the following steps that：
401) longitudinal direction is done to image M3 cumulative, obtains cumulative projection of the image in X-axis, according to the sharp peaks characteristic of projection,
It is three detection zones by image M3 points：Orbital region, track exterior lateral area, track inside region；
402) according to provincial characteristics, judge to be occupied with the presence or absence of obstacle in region：
In the case of accessible, the feature of three detection zones is：
Orbital region, two track longitudinal direction accumulated values should have close peak value, and highly be matched with detection window；
Track exterior lateral area, longitudinal accumulated value is flat without saltus step；
Track inside region, longitudinal accumulated value is flat without saltus step；
According to features above, the obstacle of detection is divided into two classes, the first kind is that the obstacle on the inside of track and outside is occupied, the
Two classes are that track blocks obstacle；If track inner side and outer side region has obstacle and occupies, will there is saltus step in longitudinal accumulated value,
Hopping amplitude is relevant with the longitudinal size that obstacle is occupied；If track has obstacle occupied, track regions longitudinal direction accumulated value will be less than
The height value of detection window, it is relevant less than part and the track area size that is blocked；
If 403) three detection zones have obstacle and occupy, the X-axis coordinate that disturbance in judgement is occupied is interval：
Firstly, for orbital region image, do binaryzation and negate treatment, it is 1 that track blocks barrier pixel value, and normal
The pixel value of track regions is 0；Then, the longitudinal direction of original track regions is substituted with orbital region binaryzation longitudinal accumulated value of the inverted
Accumulated value；Finally, the size requirement for being occupied according to detection obstacle, the longitudinal accumulated value transition detection thresholding of setting, for beyond door
The hop region of limit extracts its lateral coordinates interval { (Sxn,Exn), n=1,2 ..., N, wherein, SxnRepresent n-th saltus step area
Initial x-axis coordinate, ExnN-th termination x-axis coordinate in saltus step area is represented, N represents saltus step area number, that is, there is obstacle and occupy
Lateral coordinates interval number；
404) occupying interval for obstacle carries out laterally adding up respectively, judges that the Y-axis coordinate that each obstacle is occupied is interval：From
From left to right, the obstacle successively in interception image M3 occupies the interval image A of lateral coordinatesn, n=1,2 ..., N；
Truncated picture is carried out respectively laterally to add up, cumulative projection of the image in Y-axis is obtained, it is cumulative according to each image
The saltus step of value, can respectively obtain each lateral coordinates interval (Sxn,Exn) on Y-axis coordinate the interval { (Sy that occupies of obstacleN, k,
EyN, k), n=1,2 ..., N, k=1,2 ..., Kn, wherein Syn,kAnd Eyn,kLateral coordinates interval (Sx is represented respectivelyn,Exn) on
Starting and terminate y-axis coordinate, K that k-th obstacle is occupiednRepresent lateral coordinates interval (Sxn,Exn) present on obstacle occupy
Number；The obstacle for more than detecting occupies sum
5) occupying size according to obstacle carries out obstacle danger early warning, and it is rectangular area arrangement set to define obstacle and occupy
{R1n,k, R1n,kRepresent n-th lateral coordinates interval (Sxn,Exn) on k-th obstacle occupy rectangular area, rectangular area
Two diagonal point coordinates are (Sxn, SyN, k) and (Exn, EyN, k), region R1 is occupied by obstaclen,kArea according to from big to small
Sequence carrys out the danger classes that disturbance in judgement is occupied.
6) according to step 5) result in video image in real time mark obstacle occupy region.Concretely comprise the following steps：To image
Obstacle in M3 occupies rectangular area arrangement set { R1n,kConverse perspective transform is carried out, obstacle is occupied into coordinate and track is blocked
Obstacle coordinate is mapped in image M2, marks out barrier zone, and exports the parameter of obstacle in real time, be train automatic danger-avoiding or
Manual intervention provides reference, and the wherein parameter of obstacle includes number, type and the size of obstacle.
The feasibility of inventive algorithm is illustrated below by a train preceding object analyte detection example.
As shown in Fig. 2 being frame train front video image, two carton obstacles are included in image, one of them is located at
Track inside region, another causes to block to right rail.
As shown in figure 3, being pretreated image detection window.Its processing procedure be video image is carried out noise reduction and
After the Morphological scale-space of enhancing, the rim detection of image, the binary conversion treatment of image and image, by track fitting detection simultaneously
Image detection window is set up by main body of orbital region.
As shown in figure 4, to detection window image, carry out anti-perspective transform, now image middle orbit be approximately two it is parallel
Straight line.
As shown in figure 5, longitudinal direction is carried out to the image pixel value after anti-perspective transform adding up, and project to X-axis.Can from Fig. 5
To find out, there are three main crests, left rail area, track inner side obstacle occupied zone and right rail area are corresponded to respectively.
Longitudinal accumulation result according to Fig. 5, finds out the transition point of image, then can occupy region as rail region and obstacle
X-axis coordinate.
Occupying interval for obstacle carries out laterally adding up respectively, the binary map of subregion transverse direction accumulation result is drawn, such as Fig. 6 institutes
Show, wherein Fig. 6 (a) is left rail area horizontal pixel value accumulation result, Fig. 6 (b) is the track inner side cumulative knot of horizontal pixel value
Really, Fig. 6 (c) is left rail area horizontal pixel value accumulation result.Occupy what is caused by obstacle from Fig. 6 (b) as can be seen that existing
Pixel accumulated value saltus step, may determine that the Y-axis coordinate that obstacle is occupied is interval by trip point.From Fig. 6 (c) as can be seen that track is deposited
In discontinuity interval, the interval is the Y-axis coordinate interval that track blocks obstacle location.
It is comprehensive, X-axis and Y-axis coordinate interval parameter that the obstacle from obtained by Fig. 5, Fig. 6 is occupied, it can be determined that exist at two
Obstacle, for track inner side obstacle is occupied at first, for track blocks obstacle at second, barrier size can also have coordinate to calculate
Obtain.The coordinate occupied to obstacle carries out converse perspective transform, and marks barrier in detection window image, such as Fig. 7 institutes
Show.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, on the premise of the technology of the present invention principle is not departed from, some improvement and deformation can also be made, these improve and deform
Also should be regarded as protection scope of the present invention.
Claims (6)
1. a kind of train preceding object object detecting method based on video image, it is characterized in that, comprise the following steps：
1) video image M0 is pre-processed, obtains pretreated binary image M1；
2) obstacle detection window is set up：According to the rectilinear orbit aspect of model, track fitting detection is carried out to M1 images, according to track
Testing result, intercepts the detection window image M2 based on orbital region；
3) according to two in image M2 the four of track extreme coordinates, had an X-rayed according to anti-perspective transform algorithm is counter to image M2
Conversion obtains image M3；
4) vertical and horizontal respectively to image M3 pixel values add up, and judge whether that obstacle is occupied, and obstacle is accounted for
According to being positioned；
5) occupying size according to obstacle carries out obstacle danger early warning；
6) according to step 5) result in video image in real time mark obstacle occupy region.
2. the train preceding object object detecting method based on video image according to claim 1, it is characterized in that, the step
It is rapid 1) in particular content be：A two field picture M0 is extracted from video successively, noise reduction, enhancing treatment is carried out to image M0, then
Rim detection and binary conversion treatment are carried out to enhanced image, color video frequency image is converted into black white image, used successively
Corrosion and expanding method carry out Morphological scale-space to the image after binaryzation, so as to further filter out discrete noise, finally give
Pretreated binary image M1.
3. the train preceding object object detecting method based on video image according to claim 1, it is characterized in that, the step
Rapid concretely comprising the following steps 4)：
401) longitudinal direction is done to image M3 cumulative, obtains cumulative projection of the image in X-axis, according to the sharp peaks characteristic of projection, will schemed
It is three detection zones as M3 divides：Orbital region, track exterior lateral area, track inside region；
402) according to provincial characteristics, judge to be occupied with the presence or absence of obstacle in region：
In the case of accessible, the feature of three detection zones is：
Orbital region, two track longitudinal direction accumulated values should have close peak value, and highly be matched with detection window；
Track exterior lateral area, longitudinal accumulated value is flat without saltus step；
Track inside region, longitudinal accumulated value is flat without saltus step；
According to features above, the obstacle of detection is divided into two classes, the first kind is that the obstacle on the inside of track and outside is occupied, Equations of The Second Kind
It is that track blocks obstacle；If track inner side and outer side region has obstacle and occupies, will there is saltus step, saltus step in longitudinal accumulated value
Amplitude is relevant with the longitudinal size that obstacle is occupied；If track has obstacle occupied, track regions longitudinal direction accumulated value will be less than detection
The height value of window, it is relevant less than part and the track area size that is blocked；
If 403) three detection zones have obstacle and occupy, the X-axis coordinate that disturbance in judgement is occupied is interval：
Firstly, for orbital region image, do binaryzation and negate treatment, it is 1 that track blocks barrier pixel value, and normal orbit
The pixel value in area is 0；Then, add up the longitudinal direction for substituting original track regions with orbital region binaryzation longitudinal accumulated value of the inverted
Value；Finally, the size requirement for being occupied according to detection obstacle, the longitudinal accumulated value transition detection thresholding of setting, for beyond thresholding
Hop region extracts its lateral coordinates interval { (Sxn,Exn), n=1,2 ..., N, wherein, SxnRepresent rising for n-th saltus step area
Beginning x-axis coordinate, ExnN-th termination x-axis coordinate in saltus step area is represented, N represents saltus step area number, that is, there is the horizontal stroke that obstacle is occupied
To coordinate interval number；
404) occupying interval for obstacle carries out laterally adding up respectively, judges that the Y-axis coordinate that each obstacle is occupied is interval：It is past from a left side
The right side, successively the obstacle in interception image M3 occupy the interval image A of lateral coordinatesn, n=1,2 ..., N；
Truncated picture is carried out respectively laterally to add up, cumulative projection of the image in Y-axis is obtained, according to each image accumulated value
Saltus step, can respectively obtain each lateral coordinates interval (Sxn,Exn) on Y-axis coordinate the interval { (Sy that occupies of obstaclen,k,
Eyn,k), n=1,2 ..., N, k=1,2 ..., Kn, wherein Syn,kAnd Eyn,kLateral coordinates interval (Sx is represented respectivelyn,Exn) on
Starting and terminate y-axis coordinate, K that k-th obstacle is occupiednRepresent lateral coordinates interval (Sxn,Exn) present on obstacle occupy
Number；The obstacle for more than detecting occupies sum。
4. the train preceding object object detecting method based on video image according to claim 3, it is characterized in that, the step
Rapid 5) particular content is：
It is rectangular area arrangement set { R1 to define obstacle and occupyn,k, R1n,kRepresent n-th lateral coordinates interval (Sxn, Exn) on
K-th obstacle occupy rectangular area, two diagonal point coordinates of rectangular area are (Sxn, Syn,k) and (Exn, Eyn,k), pass through
Obstacle occupies region R1n,kArea carry out the danger classes that occupies of disturbance in judgement according to sequence from big to small.
5. the train preceding object object detecting method based on video image according to claim 4, it is characterized in that, the step
Rapid specific steps 6)：Arrangement set { R1 in rectangular area is occupied to the obstacle in image M3n,kConverse perspective transform is carried out, will
Obstacle occupies coordinate and track blocks obstacle coordinate and is mapped in image M2, marks out barrier zone, and exports obstacle in real time
Parameter, is that train automatic danger-avoiding or manual intervention provide reference.
6. the train preceding object object detecting method based on video image according to claim 5, it is characterized in that, the barrier
The parameter for hindering includes number, type and the size of obstacle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611103510.7A CN106778569A (en) | 2016-12-05 | 2016-12-05 | Train preceding object object detecting method based on video image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611103510.7A CN106778569A (en) | 2016-12-05 | 2016-12-05 | Train preceding object object detecting method based on video image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106778569A true CN106778569A (en) | 2017-05-31 |
Family
ID=58884549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611103510.7A Pending CN106778569A (en) | 2016-12-05 | 2016-12-05 | Train preceding object object detecting method based on video image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106778569A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110070026A (en) * | 2019-04-17 | 2019-07-30 | 中国科学院自动化研究所 | Video image danger Situation Awareness method, system, device based on pattern-recognition |
| CN110716504A (en) * | 2019-10-21 | 2020-01-21 | 同济大学 | Rolling ball system motion control method based on multi-closed-loop cascade control |
| CN112464906A (en) * | 2020-12-17 | 2021-03-09 | 上海媒智科技有限公司 | Method and device for detecting rail foreign matter intrusion based on computer vision |
| CN112488056A (en) * | 2020-12-17 | 2021-03-12 | 上海媒智科技有限公司 | Linear track foreign matter intrusion detection method and device based on computer vision |
| CN112896231A (en) * | 2021-03-01 | 2021-06-04 | 宁夏大学 | Railway track sand burying degree monitoring device and method |
| CN114248819A (en) * | 2020-09-25 | 2022-03-29 | 中车株洲电力机车研究所有限公司 | Railway intrusion foreign matter unmanned aerial vehicle detection method, device and system based on deep learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331910A (en) * | 2014-11-24 | 2015-02-04 | 沈阳建筑大学 | Track obstacle detection system based on machine vision |
| CN105701453A (en) * | 2016-01-04 | 2016-06-22 | 中南大学 | Railway ballast vehicle with obstacle identification system and obstacle identification method |
-
2016
- 2016-12-05 CN CN201611103510.7A patent/CN106778569A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331910A (en) * | 2014-11-24 | 2015-02-04 | 沈阳建筑大学 | Track obstacle detection system based on machine vision |
| CN105701453A (en) * | 2016-01-04 | 2016-06-22 | 中南大学 | Railway ballast vehicle with obstacle identification system and obstacle identification method |
Non-Patent Citations (4)
| Title |
|---|
| 朱正平 等: "基于机器视觉的轨道异物侵限检测方法研究", 《甘肃高校学报》 * |
| 陈爱林: "一种基于视频图像的列车前方障碍物检测算法", 《工程技术》 * |
| 陈若望: "列车前方障碍物图像检测算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 黄涛 等: "基于多技术融合的地铁列车障碍物检测系统设计", 《制造业自动化》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110070026A (en) * | 2019-04-17 | 2019-07-30 | 中国科学院自动化研究所 | Video image danger Situation Awareness method, system, device based on pattern-recognition |
| CN110070026B (en) * | 2019-04-17 | 2021-01-29 | 中国科学院自动化研究所 | Method, system and device for sensing dangerous situation of video image based on pattern recognition |
| CN110716504A (en) * | 2019-10-21 | 2020-01-21 | 同济大学 | Rolling ball system motion control method based on multi-closed-loop cascade control |
| CN110716504B (en) * | 2019-10-21 | 2022-06-14 | 同济大学 | Rolling ball system motion control method based on multi-closed-loop cascade control |
| CN114248819A (en) * | 2020-09-25 | 2022-03-29 | 中车株洲电力机车研究所有限公司 | Railway intrusion foreign matter unmanned aerial vehicle detection method, device and system based on deep learning |
| CN114248819B (en) * | 2020-09-25 | 2023-12-29 | 中车株洲电力机车研究所有限公司 | Railway intrusion foreign matter unmanned aerial vehicle detection method, device and system based on deep learning |
| CN112464906A (en) * | 2020-12-17 | 2021-03-09 | 上海媒智科技有限公司 | Method and device for detecting rail foreign matter intrusion based on computer vision |
| CN112488056A (en) * | 2020-12-17 | 2021-03-12 | 上海媒智科技有限公司 | Linear track foreign matter intrusion detection method and device based on computer vision |
| CN112488056B (en) * | 2020-12-17 | 2024-08-23 | 上海媒智科技有限公司 | Linear track foreign matter intrusion detection method and device based on computer vision |
| CN112464906B (en) * | 2020-12-17 | 2024-08-23 | 上海媒智科技有限公司 | Method and device for detecting invasion of track foreign matters based on computer vision |
| CN112896231A (en) * | 2021-03-01 | 2021-06-04 | 宁夏大学 | Railway track sand burying degree monitoring device and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106778569A (en) | Train preceding object object detecting method based on video image | |
| CN104331910B (en) | A kind of track obstacle detecting system based on machine vision | |
| WO2022199472A1 (en) | Obstacle detection method, and vehicle, device and computer storage medium | |
| CN107253485B (en) | Foreign matter invades detection method and foreign matter invades detection device | |
| US9292750B2 (en) | Method and apparatus for detecting traffic monitoring video | |
| CN104574376B (en) | Avoiding collision based on binocular vision and laser radar joint verification in hustle traffic | |
| CN103021177B (en) | Method and system for processing traffic monitoring video image in foggy day | |
| CN106960179B (en) | Rail line Environmental security intelligent monitoring method and device | |
| EP1796043A2 (en) | Object detection | |
| CN110298300B (en) | Method for detecting vehicle illegal line pressing | |
| CN102799857B (en) | Video multi-vehicle outline detection method | |
| CN105203552A (en) | 360-degree tread image detecting system and method | |
| CN105632186A (en) | Method and device for detecting vehicle queue jumping behavior | |
| JP2003067752A (en) | Vehicle periphery monitoring device | |
| CN111461088A (en) | Rail transit obstacle avoidance system based on image processing and target recognition | |
| US9189691B2 (en) | Three-dimensional object detection device and three-dimensional object detection method | |
| CN107274678B (en) | A kind of night vehicle flowrate and model recognizing method based on Kinect | |
| CN103488975A (en) | Zebra crossing real-time detection method based in intelligent driving | |
| JP2002074368A (en) | Moving object recognizing and tracking device | |
| CN105512641B (en) | A method of dynamic pedestrian and vehicle under calibration sleet state in video | |
| KR101406316B1 (en) | Apparatus and method for detecting lane | |
| Wu et al. | Adjacent lane detection and lateral vehicle distance measurement using vision-based neuro-fuzzy approaches | |
| Wang et al. | An improved hough transform method for detecting forward vehicle and lane in road | |
| Wang et al. | Geometry constraints-based visual rail track extraction | |
| Guido et al. | Vehicle tracking system based on videotaping data |
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: 20170531 |