CN105654045A - Method applied in active driving technology for identifying traffic control personnel - Google Patents

Method applied in active driving technology for identifying traffic control personnel Download PDF

Info

Publication number
CN105654045A
CN105654045A CN201511005119.9A CN201511005119A CN105654045A CN 105654045 A CN105654045 A CN 105654045A CN 201511005119 A CN201511005119 A CN 201511005119A CN 105654045 A CN105654045 A CN 105654045A
Authority
CN
China
Prior art keywords
traffic control
control personnel
pedestrian
traffic
driving technology
Prior art date
Application number
CN201511005119.9A
Other languages
Chinese (zh)
Other versions
CN105654045B (en
Inventor
田雨农
吴子章
周秀田
陆振波
于维双
Original Assignee
大连楼兰科技股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 大连楼兰科技股份有限公司 filed Critical 大连楼兰科技股份有限公司
Priority to CN201511005119.9A priority Critical patent/CN105654045B/en
Publication of CN105654045A publication Critical patent/CN105654045A/en
Application granted granted Critical
Publication of CN105654045B publication Critical patent/CN105654045B/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00362Recognising human body or animal bodies, e.g. vehicle occupant, pedestrian; Recognising body parts, e.g. hand
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/62Methods or arrangements for recognition using electronic means
    • G06K9/6217Design or setup of recognition systems and techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
    • G06K9/6256Obtaining sets of training patterns; Bootstrap methods, e.g. bagging, boosting
    • G06K9/6259Obtaining sets of training patterns; Bootstrap methods, e.g. bagging, boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling

Abstract

The invention provides a method applied in active driving technology for identifying traffic control personnel. On the basis of adboost-based passenger detecting technology, large-sample modeling is performed on fluorescent waistcoats which are equipped on the traffic control personnel for realizing characteristic statistics, and furthermore a color histogram statistics template of the fluorescent waistcoats is established. When an active driving system detects a front passenger, color histogram statistics template matching of the fluorescent waistcoat is performed in a square area which is around the trunk part of the passenger and has an edge length of one interval. When the similarity is above 85%, characteristic matching is performed on a candidate area. A part which satisfies a threshold requirement is determined as the fluorescent waistcoat, and the passenger is determined as the traffic control personnel. On special traffic condition in which traffic control personnel exist, environment information can be mastered in a more comprehensive manner, thereby making more coping decisions.

Description

It is applied in the method identifying traffic control personnel in active driving technology

Technical field

The invention belongs to initiatively driving technology field, it is a kind of be applied in the method identifying traffic control personnel in initiatively driving technology specifically.

Background technology

The favor receiving many research and development institutions is driven in the active that intelligent transportation in recent years combines with vision technique, especially after Google and tesla carry out the open test that automobile is initiatively driven in succession, the company of research active driving technology and scientific research institution, especially as emerged rapidly in large numbersBamboo shoots after a spring rain, emerge in an endless stream. At present, existing most initiatively driving technology, substantially it is all by the position & navigation map of automobile self is combined, carry out the planning in path, be combined, with sensors such as lidars, the traffic sign and obstacle information that detect front by vision sensor, to instruct the driving trace of automobile itself, thus observe traffic rules and regulations, and avoid the generation of traffic accident. But when the unexpected happens, the traffic control personnel such as traffic police or association's duty that often have are to carry out maintenance and the process of special screne. If the process initiatively driven can be made prompting or the warning of this respect, it is very helpful undoubtedly to navigating mate.

Summary of the invention

The present invention provides and a kind of is applied in the method identifying traffic control personnel in initiatively driving technology, passes through the method, it is possible to make initiatively to drive to have higher actual use value, the Driving Scene being more applicable in people's daily life.

For achieving the above object, the technical scheme of the present invention is, is applied in the method identifying traffic control personnel in active driving technology, is realized by following step:

S1: fluorescence vest image pattern and its negative sample collecting a large amount of pedestrian and traffic control personnel;

S2: carry out characteristic statistics by adboost, off-line training obtains the sorter about pedestrian detection and the sorter of fluorescence vest detection, and sets up the color histogram figure template of fluorescence vest;

S3: when active driving system by pedestrian detection module be tested with pedestrian target occur time, the torso portion of pedestrian is respectively deviateed 1/N up and down interval, carries out the color histogram figure template matches of fluorescence vest;

S4: multiple candidate regions that measuring similarity reaches after template matches more than M carry out fluorescence vest sorter and accurately detect, chooses region that wherein degree of confidence the is the highest degree of confidence as net result;

S5: when the degree of confidence of net result is greater than threshold value, detects that pedestrian is traffic control personnel.

Further, the color histogram figure template of fluorescence vest is obtained by rbf neural network training.

Further, the span of described M is 80-85%.

Further, described threshold value is 95%;

Further, the method also comprises when detecting that pedestrian is traffic control personnel, and navigating mate is pointed out the step switching to manual drive pattern.

Further, when detecting that pedestrian is traffic control personnel, the cap of traffic control personnel being carried out type identification, its step is as follows:

A, carry out characteristic statistics by adboost, the alert cap picture of the traffic police of off-line training large sample, and then determine the alert cap sorter of traffic police;

B, on the above-mentioned basis detecting out traffic control personnel, by the pedestrian head region detected out being carried out the expansion in each 1/N region up and down;

C, the detection that then head zone after expansion carries out the alert cap of traffic police, utilize and train the alert cap sorter of the traffic police obtained in steps A, judges whether there is the alert cap of traffic police in this region, if had, this traffic control personnel are traffic police; Otherwise for association is diligent.

Further, the span of described N is 2-3.

The present invention is owing to adopting above technical scheme, it is possible to obtain following technique effect: pedestrian detection and dress ornament detection is combined, thus traffic control personnel is detected in active driving procedure, and then personnel on prompting vehicle. Navigating mate on such helping prompt vehicle, under some have the special traffic scene that traffic control personnel exist, can more comprehensively grasp environmental information, more tackle decision-making to make.

The present invention utilizes the color distribution histogram of the fluorescence vest of traffic control personnel, close in the pedestrian's torso portion detected out and in scope, carry out Preliminary detection, then the candidate region detected out is carried out the accurate identification of sorter, thus improves the recognition efficiency of traffic control personnel. By the alert cap of traffic police is set up sorter, when detecting out traffic control personnel, in its head zone neighborhood, carry out the alert cap detection of traffic police, thus differentiate whether this traffic control personnel are traffic police. Follow-up the method can also be used to carry out classifying to the traffic control personnel of other types or other staff and identify.

Accompanying drawing explanation

The present invention has accompanying drawing 2 width:

Fig. 1 is the traffic control personal identification process flow block diagram of the present invention;

Fig. 2 is traffic police's recognition process FB(flow block).

Embodiment

Below by embodiment, and by reference to the accompanying drawings, the technical scheme of the present invention is described in further detail.

Embodiment 1

It is applied in the method identifying traffic control personnel in active driving technology, is realized by following step:

S1: fluorescence vest image pattern and its negative sample collecting a large amount of pedestrian and traffic control personnel;

S2: carry out characteristic statistics by adboost, off-line training obtains the sorter about pedestrian detection and the sorter of fluorescence vest detection, and is obtained the color histogram figure template of fluorescence vest by rbf neural network training;

S3: when active driving system by pedestrian detection module be tested with pedestrian target occur time, the torso portion of pedestrian is respectively deviateed 1/2 up and down interval, carries out the color histogram figure template matches of fluorescence vest;

S4: multiple candidate regions that measuring similarity reaches after template matches more than 80% carry out fluorescence vest sorter and accurately detect, chooses region that wherein degree of confidence the is the highest degree of confidence as net result;

S5: when the degree of confidence of net result is greater than threshold value 95%, detects that pedestrian is that navigating mate is pointed out the step switching to manual drive pattern or carries out other work by traffic control personnel.

When detecting that pedestrian is traffic control personnel, the cap of traffic control personnel being carried out type identification, its step is as follows: carry out characteristic statistics by adboost, the alert cap picture of the traffic police of off-line training large sample, and then determines the alert cap sorter of traffic police; On the above-mentioned basis detecting out traffic control personnel, by the pedestrian head region detected out being carried out the expansion in each 1/2 region up and down; Then the head zone after expansion being carried out the detection of the alert cap of traffic police, utilizes and steps A is trained the alert cap sorter of the traffic police obtained, judging whether there is the alert cap of traffic police in this region, if had, this traffic control personnel are traffic police; Otherwise for association is diligent.

Embodiment 2

It is applied in the method identifying traffic control personnel in active driving technology, is realized by following step:

S1: fluorescence vest image pattern and its negative sample collecting a large amount of pedestrian and traffic control personnel;

S2: carry out characteristic statistics by adboost, off-line training obtains the sorter about pedestrian detection and the sorter of fluorescence vest detection, and is obtained the color histogram figure template of fluorescence vest by rbf neural network training;

S3: when active driving system by pedestrian detection module be tested with pedestrian target occur time, the torso portion of pedestrian is respectively deviateed 1/3 up and down interval, carries out the color histogram figure template matches of fluorescence vest;

S4: multiple candidate regions that measuring similarity reaches after template matches more than 85% carry out fluorescence vest sorter and accurately detect, chooses region that wherein degree of confidence the is the highest degree of confidence as net result;

S5: when the degree of confidence of net result is greater than threshold value 95%, detects that pedestrian is that navigating mate is pointed out the step switching to manual drive pattern or carries out other work by traffic control personnel.

When detecting that pedestrian is traffic control personnel, the cap of traffic control personnel being carried out type identification, its step is as follows: carry out characteristic statistics by adboost, the alert cap picture of the traffic police of off-line training large sample, and then determines the alert cap sorter of traffic police; On the above-mentioned basis detecting out traffic control personnel, by the pedestrian head region detected out being carried out the expansion in each 1/3 region up and down; Then the head zone after expansion being carried out the detection of the alert cap of traffic police, utilizes and steps A is trained the alert cap sorter of the traffic police obtained, judging whether there is the alert cap of traffic police in this region, if had, this traffic control personnel are traffic police; Otherwise for association is diligent.

Pedestrian detection and dress ornament (being fluorescence vest here) detection are combined by the present invention, thus traffic control personnel are detected in active driving procedure, and then personnel on prompting vehicle. The present invention utilizes the color distribution histogram of the fluorescence vest of traffic control personnel, close in the pedestrian's torso portion detected out and in scope, carry out Preliminary detection, then the candidate region detected out is carried out the accurate identification of sorter, thus improves the recognition efficiency of traffic control personnel. Alert to pedestrian detection and traffic police cap detection is combined by the present invention, thus when detecting out traffic control personnel, carries out the alert cap detection of traffic police, thus differentiate whether this traffic control personnel are traffic police in its head zone neighborhood.Follow-up the method can also be used to carry out classifying to the traffic control personnel of other types or other staff and identify.

The above; it is only the present invention's preferably embodiment; but protection scope of the present invention is not limited thereto; any it is familiar with those skilled in the art in the technical scope of present disclosure; technical scheme and invention design thereof according to the present invention are equal to replacement or are changed, and all should be encompassed within protection scope of the present invention.

Claims (7)

1. it is applied in the method identifying traffic control personnel in active driving technology, it is characterised in that, realized by following step:
S1: fluorescence vest image pattern and its negative sample collecting a large amount of pedestrian and traffic control personnel;
S2: carry out characteristic statistics by adboost, off-line training obtains the sorter about pedestrian detection and the sorter of fluorescence vest detection, and sets up the color histogram figure template of fluorescence vest;
S3: when active driving system by pedestrian detection module be tested with pedestrian target occur time, the torso portion of pedestrian is respectively deviateed 1/N up and down interval, carries out the color histogram figure template matches of fluorescence vest;
S4: multiple candidate regions that measuring similarity reaches after template matches more than M carry out fluorescence vest sorter and accurately detect, chooses region that wherein degree of confidence the is the highest degree of confidence as net result;
S5: when the degree of confidence of net result is greater than threshold value, detects that pedestrian is traffic control personnel.
2. according to claim 1 it is applied in the method identifying traffic control personnel in initiatively driving technology, it is characterised in that, the color histogram figure template of fluorescence vest is obtained by rbf neural network training.
3. according to claim 1 it is applied in the method identifying traffic control personnel in initiatively driving technology, it is characterised in that, the span of described M is 80-85%.
4. according to claim 3 it is applied in the method identifying traffic control personnel in initiatively driving technology, it is characterised in that, described threshold value is 95%.
5. according to claim 4 it is applied in the method identifying traffic control personnel in initiatively driving technology, it is characterized in that, the method also comprises when detecting that pedestrian is traffic control personnel, and navigating mate is pointed out the step switching to manual drive pattern.
6. according to claim 5 it is applied in the method identifying traffic control personnel in initiatively driving technology, it is characterised in that, when detecting that pedestrian is traffic control personnel, the cap of traffic control personnel is carried out type identification, its step is as follows:
A, carry out characteristic statistics by adboost, the alert cap picture of the traffic police of off-line training large sample, and then determine the alert cap sorter of traffic police;
B, on the above-mentioned basis detecting out traffic control personnel, by the pedestrian head region detected out being carried out the expansion in each 1/N region up and down;
C, the detection that then head zone after expansion carries out the alert cap of traffic police, utilize and train the alert cap sorter of the traffic police obtained in steps A, judges whether there is the alert cap of traffic police in this region, if had, this traffic control personnel are traffic police; Otherwise for association is diligent.
7. it is applied in, according to claim 1 or 6, the method identifying traffic control personnel in initiatively driving technology, it is characterised in that, the span of described N is 2-3.
CN201511005119.9A 2015-12-29 2015-12-29 Apply the method that traffic control personnel are identified in active driving technology CN105654045B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201511005119.9A CN105654045B (en) 2015-12-29 2015-12-29 Apply the method that traffic control personnel are identified in active driving technology

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201511005119.9A CN105654045B (en) 2015-12-29 2015-12-29 Apply the method that traffic control personnel are identified in active driving technology

Publications (2)

Publication Number Publication Date
CN105654045A true CN105654045A (en) 2016-06-08
CN105654045B CN105654045B (en) 2019-04-05

Family

ID=56478138

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201511005119.9A CN105654045B (en) 2015-12-29 2015-12-29 Apply the method that traffic control personnel are identified in active driving technology

Country Status (1)

Country Link
CN (1) CN105654045B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109263649A (en) * 2018-08-21 2019-01-25 北京汽车股份有限公司 Object identification method and object identification system under vehicle and its automatic driving mode

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040145496A1 (en) * 1996-09-25 2004-07-29 Ellis Christ G. Intelligent vehicle apparatus and method for using the apparatus
CN101872422A (en) * 2010-02-10 2010-10-27 杭州海康威视软件有限公司 People flow rate statistical method and system capable of precisely identifying targets
CN102542260A (en) * 2011-12-30 2012-07-04 中南大学 Method for recognizing road traffic sign for unmanned vehicle
CN103049751A (en) * 2013-01-24 2013-04-17 苏州大学 Improved weighting region matching high-altitude video pedestrian recognizing method
CN103489324A (en) * 2013-09-22 2014-01-01 北京联合大学 Real-time dynamic traffic light detection identification method based on unmanned driving
CN104463146A (en) * 2014-12-30 2015-03-25 华南师范大学 Posture identification method and device based on near-infrared TOF camera depth information

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040145496A1 (en) * 1996-09-25 2004-07-29 Ellis Christ G. Intelligent vehicle apparatus and method for using the apparatus
CN101872422A (en) * 2010-02-10 2010-10-27 杭州海康威视软件有限公司 People flow rate statistical method and system capable of precisely identifying targets
CN102542260A (en) * 2011-12-30 2012-07-04 中南大学 Method for recognizing road traffic sign for unmanned vehicle
CN103049751A (en) * 2013-01-24 2013-04-17 苏州大学 Improved weighting region matching high-altitude video pedestrian recognizing method
CN103489324A (en) * 2013-09-22 2014-01-01 北京联合大学 Real-time dynamic traffic light detection identification method based on unmanned driving
CN104463146A (en) * 2014-12-30 2015-03-25 华南师范大学 Posture identification method and device based on near-infrared TOF camera depth information

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109263649A (en) * 2018-08-21 2019-01-25 北京汽车股份有限公司 Object identification method and object identification system under vehicle and its automatic driving mode

Also Published As

Publication number Publication date
CN105654045B (en) 2019-04-05

Similar Documents

Publication Publication Date Title
US8917904B2 (en) Vehicle clear path detection
Tan et al. Color model-based real-time learning for road following
Betke et al. Real-time multiple vehicle detection and tracking from a moving vehicle
Li et al. Springrobot: A prototype autonomous vehicle and its algorithms for lane detection
Gavrila Traffic sign recognition revisited
Gong et al. The recognition and tracking of traffic lights based on color segmentation and camshift for intelligent vehicles
US8670592B2 (en) Clear path detection using segmentation-based method
JP2013073620A (en) Method and system for detecting road landform for driver support system
US20100097456A1 (en) Clear path detection using a hierachical approach
US20120300078A1 (en) Environment recognizing device for vehicle
TWI430212B (en) Abnormal behavior detection system and method using automatic classification of multiple features
US8995723B2 (en) Detecting and recognizing traffic signs
CN103020623B (en) Method for traffic sign detection and road traffic sign detection equipment
US8634593B2 (en) Pixel-based texture-less clear path detection
EP1782335A1 (en) Method for traffic sign detection
Cheng et al. Lane detection with moving vehicles in the traffic scenes
Fang et al. Road-sign detection and tracking
Keller et al. A new benchmark for stereo-based pedestrian detection
Fang et al. A road sign recognition system based on dynamic visual model
CN107368890A (en) A kind of road condition analyzing method and system based on deep learning centered on vision
Shneier Road sign detection and recognition
Jensen et al. Vision for looking at traffic lights: Issues, survey, and perspectives
US8452053B2 (en) Pixel-based texture-rich clear path detection
Siogkas et al. Traffic Lights Detection in Adverse Conditions using Color, Symmetry and Spatiotemporal Information.
US20130169536A1 (en) Control of a wearable device

Legal Events

Date Code Title Description
PB01 Publication
C06 Publication
SE01 Entry into force of request for substantive examination
C10 Entry into substantive examination
GR01 Patent grant
GR01 Patent grant