CN104680793B - A kind of truck overhead alarm method violating the regulations - Google Patents
A kind of truck overhead alarm method violating the regulations Download PDFInfo
- Publication number
- CN104680793B CN104680793B CN201510058285.9A CN201510058285A CN104680793B CN 104680793 B CN104680793 B CN 104680793B CN 201510058285 A CN201510058285 A CN 201510058285A CN 104680793 B CN104680793 B CN 104680793B
- Authority
- CN
- China
- Prior art keywords
- truck
- visible parts
- outside vehicle
- confidence level
- regulations
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
Abstract
The present invention relates to a kind of truck overhead alarm method violating the regulations, comprise the following steps: certain form of outside vehicle visible parts confidence level at an arbitrary position is trained according to lorry data by (1)；(2) obtain picture to be identified, use road surface grader to obtain the region, road surface of picture to be identified；(3) region, non-road surface is carried out the detection of outside vehicle visible parts, according to the training result of step (1), calculates the confidence level of each outside vehicle visible parts combination；(4) final position of each outside vehicle visible parts is obtained according to described confidence level；(5) picture to be identified obtaining final position is post-processed, it is thus achieved that the final recognition result of vehicle, if final recognition result is truck, then report to the police.Compared with prior art, the present invention is simple to operate, it is possible to find that in time truck is broken rules and regulations situation.
Description
Technical field
The present invention relates to intelligent transportation field, especially relate to a kind of truck overhead alarm method violating the regulations.
Background technology
On overpass, along with travel speed is accelerated and the increase of lorry weight, bridge floor stress is the biggest, thus contracts
Short overhead service life.Therefore to try to forestall traffic accidents, urban district is overhead forbids that truck passes through.But shipping department
Machine, in order to cost-effective, usually overloads, and threatens other people life.
At present, traffic police mostly takes artificially to check to intercept or set height-limiting bar mode and stops on truck overhead.But it is on-the-spot
Intercepting dangerous high, be only applicable to daytime, a lot of trucks are taken advantage of and are collided height-limiting bar night and cause height-limiting bar to damage.I.e.
Make crossing monitor, from the night up to several hours, video having found, lorry violating the regulations is difficult to the most very much, and cannot accomplish
Rapid Alarm.
Prior art by adding camera at crossing, process video pictures, can detect from video car speed,
The information such as position, thus the behaviors such as traveling, road occupying in violation of rules and regulations are detected automatically, but not for overhead on truck
Behavior is made detection and reports to the police.
Summary of the invention
Defect that the purpose of the present invention is contemplated to overcome above-mentioned prior art to exist and providing a kind of identifies accurately, from
Dynamic truck efficiently overhead alarm method violating the regulations.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of truck overhead alarm method violating the regulations, comprises the following steps:
(1) according to lorry data, certain form of outside vehicle visible parts confidence level at an arbitrary position is carried out
Training；
(2) obtain picture to be identified, use road surface grader to obtain the region, road surface of picture to be identified；
(3) region, non-road surface is carried out the detection of outside vehicle visible parts, according to the training result of step (1),
Calculate the confidence level of each outside vehicle visible parts combination；
(4) final position of each outside vehicle visible parts is obtained according to described confidence level；
(5) picture to be identified obtaining final position is post-processed, it is thus achieved that the final recognition result of vehicle, if
Final recognition result is truck, then report to the police.
In described step (1), each visible parts uses the expression way of HOG, is instructed by SVM classifier
Get certain form of each visible parts confidence level at an arbitrary position.
In described step (2), road surface grader is trained by texture and color characteristic and is obtained.
In described step (3), calculate each outside vehicle visible parts combination confidence level particularly as follows:
(3-1) using N-Best mode to obtain the form families of each outside vehicle visible parts, combined number is
(N*P)kKind, wherein, K is part count, and P is the form number of each parts, and N is the possible position of every kind of form
Put number；
(3-2) branch-and-bound mode is used to carry out lopping process, it is thus achieved that final form families；
(3-3) confidence level of the every kind of combination obtained in calculation procedure (3-2)；
(3-4) carrying out non-maxima suppression, same position only exports the result that a confidence level is maximum.
In described step (4), using the maximum combination of confidence level as the final position of each outside vehicle visible parts.
In described step (5), post processing particularly as follows:
(5-1) judge whether picture to be identified is small-sized according to the final position relation of each outside vehicle visible parts
Vehicle；
(5-2) compartment shape and color are identified, it is judged that picture to be identified whether motor bus.
In described step (5-2), compartment shape and color are identified particularly as follows:
Use GraphCut algorithm to cut out compartment in upper windscreen interval interval, extract LBP and Colar
Histogram feature, uses AbaBoosting Classification and Identification.
Described outside vehicle visible parts includes car light, windshield and car front gate.
Compared with prior art, the invention have the advantages that
1, anti-block.Truck is blocked by other vehicles, and it can preferably process part and hide the detection of each parts
Gear situation.
2, without wrong report: use post processing mode, substantially without wrong report, improve warning accuracy rate.
3, applied widely: different weather, the camera of different angles all can use.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the present invention.
Detailed description of the invention
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with the technology of the present invention side
Implement premised on case, give detailed embodiment and concrete operating process, but the protection model of the present invention
Enclose and be not limited to following embodiment.
As it is shown in figure 1, the present embodiment provides a kind of truck overhead alarm method violating the regulations, comprise the following steps:
Step S01, according to lorry data to certain form of outside vehicle visible parts confidence level at an arbitrary position
It is trained.Owing to the component shape of different brands vehicle is different, the most each parts have variform, each
Visible parts uses the expression way of HOG, is obtained certain shape of each visible parts by SVM classifier training
State confidence level at an arbitrary position.Outside vehicle visible parts includes car light, windshield and car front gate etc..
Step S02, obtains picture to be identified, uses road surface grader to obtain the region, road surface of picture to be identified.Road
Face grader is trained by texture and color characteristic and is obtained, and uses classical Texton Boost algorithm to obtain.Texton
Boost algorithm is a kind of semi-supervised machine learning method.
Step S03, carries out the detection of outside vehicle visible parts, according to the instruction of step (1) to region, non-road surface
Practice result, calculate the confidence level of each outside vehicle visible parts combination.
(3-1) using N-Best mode to obtain the form families of each outside vehicle visible parts, combined number is
(N*P)kKind, wherein, K is part count, and P is the form number of each parts, and N is the possible position of every kind of form
Put number；
(3-2) branch-and-bound mode is used to carry out lopping process, it is thus achieved that final form families；
(3-3) confidence level of the every kind of combination obtained in calculation procedure (3-2)；
Consider in following information when calculating confidence level: a) confidence level of each parts；B) between different parts
Position relationship；C) different shape constraint (the such as form of left and right car light is consistent) of different parts, improves
The speed of confidence calculations and accuracy rate.
(3-4) carrying out non-maxima suppression, same position only exports the result that a confidence level is maximum.
Step S04, obtains the final position of each outside vehicle visible parts according to described confidence level, with confidence level
Big combination is as the final position of each outside vehicle visible parts.
Step S05, post-processes the picture to be identified obtaining final position, it is thus achieved that the final recognition result of vehicle,
If final recognition result is truck, then report to the police.Wherein, the effect of post processing is to remove dilly and motor bus
Impact, particularly as follows:
(5-1) judge whether picture to be identified is small-sized according to the final position relation of each outside vehicle visible parts
Vehicle, mainly judges according to the spacing of two car lights and the size of windshield；
(5-2) compartment shape and color are identified, it is judged that picture to be identified whether motor bus, particularly as follows:
Use GraphCut algorithm to cut out compartment in upper windscreen interval interval, extract LBP and Colar
Histogram feature, uses AbaBoosting Classification and Identification.Adaboost is a kind of iterative algorithm, is Boosting
Representing algorithm in algorithm family, its core concept is the grader (weak typing different for the training of same training set
Device), then these weak classifier set are got up, constitute a higher final grader (strong classifier).
Claims (7)
1. a truck overhead alarm method violating the regulations, it is characterised in that comprise the following steps:
(1) according to lorry data, certain form of outside vehicle visible parts confidence level at an arbitrary position is carried out
Training；
(2) obtain picture to be identified, use road surface grader to obtain the region, road surface of picture to be identified；
(3) region, non-road surface is carried out the detection of outside vehicle visible parts, according to the training result of step (1),
Calculate the confidence level of each outside vehicle visible parts combination；
(4) final position of each outside vehicle visible parts is obtained according to described confidence level；
(5) picture to be identified obtaining final position is post-processed, it is thus achieved that the final recognition result of vehicle, if
Final recognition result is truck, then report to the police；
In described step (3), calculate each outside vehicle visible parts combination confidence level particularly as follows:
(3-1) using N-Best mode to obtain the form families of each outside vehicle visible parts, combined number is
(N*P)kKind, wherein, K is part count, and P is the form number of each parts, and N is the possible position of every kind of form
Put number；
(3-2) branch-and-bound mode is used to carry out lopping process, it is thus achieved that final form families；
(3-3) confidence level of the every kind of combination obtained in calculation procedure (3-2)；
(3-4) carrying out non-maxima suppression, same position only exports the result that a confidence level is maximum.
Truck the most according to claim 1 overhead alarm method violating the regulations, it is characterised in that described step
Suddenly in (1), each visible parts uses the expression way of HOG, obtains respectively may be used by SVM classifier training
See certain forms of parts confidence level at an arbitrary position.
Truck the most according to claim 1 overhead alarm method violating the regulations, it is characterised in that described step
Suddenly, in (2), road surface grader is trained by texture and color characteristic and is obtained.
Truck the most according to claim 1 overhead alarm method violating the regulations, it is characterised in that described step
Suddenly in (4), using the maximum combination of confidence level as the final position of each outside vehicle visible parts.
Truck the most according to claim 1 overhead alarm method violating the regulations, it is characterised in that described step
Suddenly in (5), post processing particularly as follows:
(5-1) judge whether picture to be identified is small-sized according to the final position relation of each outside vehicle visible parts
Vehicle；
(5-2) compartment shape and color are identified, it is judged that picture to be identified whether motor bus.
Truck the most according to claim 5 overhead alarm method violating the regulations, it is characterised in that described step
Suddenly in (5-2), compartment shape and color are identified particularly as follows:
Use GraphCut algorithm to cut out compartment in upper windscreen interval interval, extract LBP and Colar
Histogram feature, uses AbaBoosting Classification and Identification.
7. according to the arbitrary described truck of claim 1-6 overhead alarm method violating the regulations, it is characterised in that
Described outside vehicle visible parts includes car light, windshield and car front gate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510058285.9A CN104680793B (en) | 2015-02-04 | 2015-02-04 | A kind of truck overhead alarm method violating the regulations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510058285.9A CN104680793B (en) | 2015-02-04 | 2015-02-04 | A kind of truck overhead alarm method violating the regulations |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104680793A CN104680793A (en) | 2015-06-03 |
| CN104680793B true CN104680793B (en) | 2016-09-07 |
Family
ID=53315775
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510058285.9A Active CN104680793B (en) | 2015-02-04 | 2015-02-04 | A kind of truck overhead alarm method violating the regulations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104680793B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108052899A (en) * | 2017-12-12 | 2018-05-18 | 成都睿码科技有限责任公司 | A kind of method that electric bicycle and motorcycle are distinguished by video |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295017A (en) * | 2013-04-25 | 2013-09-11 | 哈尔滨工程大学 | Vehicle type identification method based on road videos |
| EP2686841A2 (en) * | 2011-03-14 | 2014-01-22 | The Regents of the University of California | Method and system for vehicle classification |
| CN103593981A (en) * | 2013-01-18 | 2014-02-19 | 西安通瑞新材料开发有限公司 | Vehicle model identification method based on video |
| CN103646546A (en) * | 2013-11-23 | 2014-03-19 | 安徽蓝盾光电子股份有限公司 | An intelligent traffic system with a large-scale vehicle passing-forbidding function |
-
2015
- 2015-02-04 CN CN201510058285.9A patent/CN104680793B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2686841A2 (en) * | 2011-03-14 | 2014-01-22 | The Regents of the University of California | Method and system for vehicle classification |
| CN103593981A (en) * | 2013-01-18 | 2014-02-19 | 西安通瑞新材料开发有限公司 | Vehicle model identification method based on video |
| CN103295017A (en) * | 2013-04-25 | 2013-09-11 | 哈尔滨工程大学 | Vehicle type identification method based on road videos |
| CN103646546A (en) * | 2013-11-23 | 2014-03-19 | 安徽蓝盾光电子股份有限公司 | An intelligent traffic system with a large-scale vehicle passing-forbidding function |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104680793A (en) | 2015-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11024165B2 (en) | Driver behavior monitoring | |
| CN105922991B (en) | Based on the lane departure warning method and system for generating virtual lane line | |
| CN110550038B (en) | Intelligent driving decision information acquisition system and method | |
| US20220327406A1 (en) | Systems and methods for classifying driver behavior | |
| CN102298845B (en) | Far-beam light video detection method and system | |
| JP2021517675A (en) | Methods and equipment for recognizing and assessing environmental impacts based on road conditions and weather | |
| WO2017123665A1 (en) | Driver behavior monitoring | |
| CN107284355A (en) | A kind of Safety vehicle door opens processing method and system | |
| CN103984950A (en) | Moving vehicle stop lamp state recognition method adaptable to day detection | |
| US20170282869A1 (en) | Road surface condition detection with multi-scale fusion | |
| CN108108680A (en) | A kind of front vehicle identification and distance measuring method based on binocular vision | |
| CN106183981A (en) | Obstacle detection method based on automobile, device and automobile | |
| CN106004668A (en) | Car rear-end -collision prevention active safety system and control method | |
| CN114379574A (en) | Object reminding device, system and method for vehicle | |
| CN110103954B (en) | Electric control-based automobile rear-end collision prevention early warning device and method | |
| CN116142186A (en) | Early warning method, device, medium and equipment for safe running of vehicle in bad environment | |
| CN104680793B (en) | A kind of truck overhead alarm method violating the regulations | |
| KR102039723B1 (en) | Vehicle's behavior analyzing system using aerial photo and analyzing method using the same | |
| Boumediene et al. | Vehicle detection algorithm based on horizontal/vertical edges | |
| CN105426852A (en) | Method for identifying pedestrians by vehicle-mounted monocular long-wave infrared camera | |
| CN110647863A (en) | Visual signal acquisition and analysis system for intelligent driving | |
| CN114056235A (en) | Ghost probe early warning system and method based on brake lamp identification | |
| Oikawa et al. | Characteristics of collision damage mitigation braking system for pedestrian protection | |
| Jyothi et al. | Driver assistance for safe navigation under unstructured traffic environment | |
| Tsuchiya et al. | Real-time vehicle detection using a single rear camera for a blind spot warning system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |