CN106203464A - A kind of based on the model recognizing method setting up membership function - Google Patents

A kind of based on the model recognizing method setting up membership function Download PDF

Info

Publication number
CN106203464A
CN106203464A CN201610473212.0A CN201610473212A CN106203464A CN 106203464 A CN106203464 A CN 106203464A CN 201610473212 A CN201610473212 A CN 201610473212A CN 106203464 A CN106203464 A CN 106203464A
Authority
CN
China
Prior art keywords
vehicle
ratio
membership function
rise
feature
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
Application number
CN201610473212.0A
Other languages
Chinese (zh)
Inventor
韩毅
黄帅
陈秋谨
王瞾
王甜甜
胡宁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Changan University
Original Assignee
Changan University
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 Changan University filed Critical Changan University
Priority to CN201610473212.0A priority Critical patent/CN106203464A/en
Publication of CN106203464A publication Critical patent/CN106203464A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/08—Detecting or categorising vehicles

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a kind of based on the model recognizing method setting up membership function, the method first extract the top in vehicle's contour long than, front and back than, rise the vehicle feature than these ratios, then analyze the top extracted long than, front and back than, rise than these vehicle features in excursion corresponding to each vehicle, choose suitable membership function by corresponding excursion, and determine the threshold value of correspondence；Afterwards the vehicle eigenvalue extracted is substituted in membership function, use the principle of maximum membership degree to be identified judging to corresponding vehicle, and set up repeatedly membership function judge according to actual can employing.Relative to prior art, which raises working performance and operation quality.

Description

A kind of based on the model recognizing method setting up membership function
Technical field
The present invention relates to technical field of intelligent traffic, be specifically related to a kind of based on the vehicle cab recognition side setting up membership function Method.
Background technology
Along with the fast development of China's economy and the improving constantly of life Happiness Index, automobile number per capita since reform and opening-up Amount increases considerably the most progressively walks close to thousand outer families so that automobile.For intelligent transportation system, for overall arrangement, anti- Stop loss and the redundancy of personnel of fund, set up a set of perfect traffic system to manage the portions such as each parking lot, charge station, traffic The demand of door is more and more urgent.Vehicle cab recognition is as an important component part of intelligent transport technology, to vehicle cab recognition Technical research be always domestic and international intelligent transport technology research emphasis.
It practice, vehicle cab recognition is a typical target recognition problem, and with the detection of target and identify closely bound up Be clarification of objective, this is just related to a feature extraction and the problem of feature selection.And existing model recognizing method pair Which kind of feature is used to lack effectively evaluating mechanism.How to find and there is the good feature described with classification performance and how to extract These features just become the key solving vehicle cab recognition problem.
Summary of the invention
The defect existed for above-mentioned prior art or deficiency, it is an object of the invention to, it is provided that a kind of based on setting up person in servitude The model recognizing method of genus degree function, the method utilizes computer vision, pattern recognition theories and methods, passes through maximum membership degree Principle reaches quickly to identify the purpose of vehicle.
In order to realize above-mentioned task, the present invention takes following technical solution:
A kind of based on setting up the model recognizing method of membership function, first extract the long ratio in top in vehicle's contour, front and back Than, rise the vehicle feature than these ratios, then analyze extraction the long ratio in top, front and back than, rise and exist than these vehicle features The excursion that each vehicle is corresponding, chooses suitable membership function by corresponding excursion, and determines correspondence Threshold value；Afterwards the vehicle eigenvalue extracted is substituted in membership function, use the principle of maximum membership degree to corresponding vehicle It is identified judging, and sets up repeatedly membership function judge according to actual can employing.
Specifically include following steps:
Step one, by the photographic head installed, it is thus achieved that ratio is in the photo of 16 grades of different gray values of scenery gray scale.
Step 2, by the Edge contrast to image, obtains automobile image edge contour clearly；
Step 3, by the profile of automobile, calculates the feature of required extraction: top long than, front and back than, rise ratio.
Step 4, sets up suitable membership function, calculates the feature of identification vehicle to be detected, utilizes maximum membership degree former Then, it is judged that corresponding vehicle.
Further, in step 2, for obtaining edge contour clearly, utilize gradient operator that image is processed, Defined from gradient: if pixel is on border, then its shade of gray direction is perpendicular to border, and mould is rate of gray level Maximum.
Further, in step 3, the top long ratio ratio for hood length Yu whole Vehicle length, front and back ratio is for car On the basis of ceiling perpendicular bisector, the ratio of vehicle front and rear part, rise the ratio than length and its height for hood.
The present invention based on setting up the model recognizing method of membership function, relative to prior art, disdained traditional meaning Vehicle commander in justice, overall width, these absolute references of overall height as the feature of vehicle cab recognition, and have employed top long than, front and back than, rise It is used as the feature of vehicle cab recognition than these relative quantities with ratio feature, is more prone to real relative to neural network recognization Existing, by use top long than, front and back than, rise than these relative amounts not changed with the change of photographic head angle, more ensure The accuracy of vehicle cab recognition, improves recognition efficiency.
And the foundation of membership function, the feature that all kinds of vehicles are extracted is not the most a fixed value, is all at certain model Enclose interior change.The most all kinds of vehicle features determine that in certain interval, but inevitably with other vehicle intersections There is equitant region, and belong to a kind of fuzzy diagnosis in this equitant region decision vehicle.Person in servitude in fuzzy diagnosis Genus its trapezoidal profile of degree function can be good show this feature, and be identified.Said two devices is combined and is formed A kind of new method to vehicle cab recognition have higher discrimination, and easily realize, recognition speed is fast, and solves The inaccuracy of characteristic parameter is caused because of camera position or the randomness of angle.
Accompanying drawing explanation
Fig. 1 be the present invention based on the model recognizing method flow chart setting up membership function；
Fig. 2 for top long than, front and back than, rise and compare schematic diagram；
Fig. 3 vehicle cab recognition procedure chart；
The present invention is described in further detail with embodiment below in conjunction with the accompanying drawings.
Detailed description of the invention
The present embodiment is given a kind of based on setting up the model recognizing method of membership function, first extracts in vehicle's contour Top long than, front and back than, rise the vehicle feature than these ratios, then analyze extraction the long ratio in top, front and back than, rise than these Vehicle feature, in excursion corresponding to each vehicle, chooses suitable membership function by corresponding excursion, and Determine the threshold value of correspondence；Afterwards the vehicle eigenvalue extracted is substituted in membership function, use the principle of maximum membership degree It is identified judging to corresponding vehicle, and sets up repeatedly membership function judge according to actual can employing.
Concrete identification step is as follows:
Step one, by the photographic head installed, it is thus achieved that ratio is in the photo of 16 grades of different gray values of scenery gray scale.
Step 2, by the Edge contrast to image, obtains automobile image edge contour clearly；
Step 3, by the profile of automobile, calculates the feature of required extraction: top long than, front and back than, rise ratio；
Step 4: set up suitable membership function, calculates the feature of identification vehicle to be detected, utilizes maximum membership degree former Then, it is judged that corresponding vehicle.
Further, step 4 judging, corresponding vehicle specifically includes:
1) classifying vehicle, classification has various；Such as car, passenger vehicle, lorry three class；Or car, mini van, The multiple mode classifications such as passenger vehicle, motorbus, lorry five class.
Vehicle might as well be set and be divided into A, B, C, D ... class vehicle.
2) by the top length of all kinds of vehicles substantial amounts of on market is compared T1, rise and compare T2, front and back than T3Carry out data statistics and Theory analysis, can show that all kinds of vehicle is at T1、T2、T3In excursion.
3) com-parison and analysis T1、T2、T3Excursion in all kinds of vehicles, finds out wherein the most significant to vehicle discrimination One feature.
Such as: T in B vehicle1Discrimination is the most notable, say, that T1(a, b) in the range of certain corresponding be vehicle B, (a is b) other vehicles in addition to vehicle B outside scope, is denoted as B1。
4) membership function for the first time is set up, T1As domain, c, d, as the threshold value distinguished, can set up two moulds Stick with paste subset B1And B2, as a example by rising half trapezoidal profile form, set up membership function as follows:
Wherein c≤a ＜ d ＜ b
B 1 ( T 1 ) = 1 - B ( T 1 )
Lower semi-trapezoid distribution form is in like manner.
5) by vehicle T of vehicle to be identified1Feature substitutes into and calculates, if B(T1)＞ B1(T1)Then can determine whether as B vehicle, otherwise then For B1Vehicle, then need to carry out second time degree of membership and judge.
6) with 3), when eliminating B vehicle and T1After feature, at T2,T3In find out feature the most significant to vehicle discrimination, As: T in A vehicle2Characteristic area indexing is the most notable, say, that T2At (a2, b2In the range of), certain correspondence is vehicle A, at (a2, b2) it is other vehicles in addition to vehicle A outside scope, it is denoted as A1。
7) with 4), set up second time membership function, T2As domain, two fuzzy subset A and A can be set up1, with Membership function is set up as follows as a example by the half trapezoidal profile form of liter:
Wherein c2≤a2＜ d2＜ b2
A 1 ( T 2 ) = 1 - A ( T 2 )
Lower semi-trapezoid distribution form is in like manner.
8) with 5), by vehicle feature T of vehicle to be identified2Substitute into and calculate, if A(T1)＞ A1(T1)Then can determine whether as A vehicle, Otherwise it is then A1Vehicle, then need to carry out third time degree of membership and judge.
9) to continue judge, process with above-mentioned in like manner.
Vehicle is divided into by this example passenger vehicle, car, lorry three class vehicle are identified, by statistics mass data, whole Manage out extraction feature value as shown in table 1:
Table 1: vehicle feature value statistical table
Vehicle The long ratio in top Rise ratio Front and back than
Car 0.2～0.5 About 1 1～2.5
Passenger vehicle Approximation 1 More than 1.5 About 1
Lorry 0.1-0.3 About 0.5-1 0.1-0.5
Top is long more notable than as can be seen from Table 1, and passenger vehicle is about 1, and car and lorry, all below 0.5, therefore may be used With according to pushing up long ratio by the maximum membership grade principle of first time Fuzzy Pattern Recognition passenger vehicle district from passenger vehicle, car, lorry Separate, then contrast car and lorry, front and back compare for significantly, can be according to ratio front and back by second time Fuzzy Pattern Recognition Maximum membership grade principle car and lorry are made a distinction.
It is used for domain long for top, is denoted as T1, T1Scope is regarded as belonging to R.Set up two fuzzy subset: A1{ passenger vehicle }, A2 { car, lorry }.Then being front and back used for domain, it is denoted as T3,T3Scope regards R equally as.Set up two fuzzy subset: B1{ sedan-chair Car }, B3{ lorry }, sets up T1{ passenger vehicle }, A2The membership function of { car, lorry }:
A 1 ( T 1 ) = 0 0 < T 1 < 0.75 20 ( T 1 - 0.75 ) 0.75 &le; T 1 < 0.8 1 0.8 &le; T 1 &le; 1 0 1 < T 1 A 2 ( T 1 ) = 1 0 < T 1 < 0.75 20 ( 0.8 - T 1 ) 0.75 &le; T 1 < 0.8 0 0.8 &le; T 1 &le; 1 0 1 < T 1
B 1 ( T 3 ) = 0 0 < T 3 < 0.6 2.5 ( T 3 - 0.6 ) 0.6 &le; T 3 < 1 1 1 &le; T 3 B 2 ( T 3 ) = 1 0 < T 3 < 0.6 2.5 ( 1 - T 3 ) 0.6 &le; T 3 < 1 0 1 &le; T 3
Identification process is: such as, records an a length of 2530mm in vehicle top, a length of 4830mm in the end, a length of 2300mm of forward part, The a length of 2530mm in rear section.Then top is long compares T1=2300/4830 ≈ 0.52, substitutes into respectively, obtains A1(0.52)=0, A2(0.52) =1, principle A based on maximum membership degree2(0.52) ＞ A1(0.52) then can determine whether the one into car or lorry, then judge, T is compared before and after calculating3=2300/2530 ≈ 0.909, substitutes into respectively, obtains B1(0.909)=0.7705, B2(0.909)= 0.2275, based on maximum membership grade principle B1(0.909) ＞ B2(0.909), can determine whether that this car is car.
Above embodiment is only the preferred embodiment of the present invention, the invention is not restricted to above-described embodiment, it is noted that For the person of ordinary skill of the art, without departing from the concept of the premise of the invention, it is also possible to make some changes Shape and improvement, within these all should belong to protection scope of the present invention.

Claims (4)

1. one kind based on the model recognizing method setting up membership function, it is characterised in that first the method extracts vehicle's contour In top long than, front and back than, rise the vehicle feature than these ratios, then analyze extraction the long ratio in top, front and back than, rise ratio These vehicle features, in excursion corresponding to each vehicle, choose suitable membership function by corresponding excursion, And determine the threshold value of correspondence；Afterwards the vehicle eigenvalue extracted is substituted in membership function, use maximum membership degree Corresponding vehicle is identified judging by principle, and sets up repeatedly membership function judge according to actual can employing.
2. the method for claim 1, it is characterised in that specifically include following steps:
Step one, by the photographic head installed, it is thus achieved that ratio is in the photo of 16 grades of different gray values of scenery gray scale；
Step 2, by the Edge contrast to image, obtains automobile image edge contour clearly；
Step 3, by the profile of automobile, calculates the feature of required extraction: top long than, front and back than and rise ratio；
Step 4, sets up suitable membership function, calculates the feature of identification vehicle to be detected, utilizes maximum membership grade principle, Judge corresponding vehicle.
3. method as claimed in claim 2, it is characterised in that in step 2, described acquisition edge contour clearly, utilize Image is processed by gradient operator, gradient define: if pixel is on border, then its shade of gray direction is perpendicular to Border, and mould is the maximum of rate of gray level.
4. method as claimed in claim 2, it is characterised in that in step 3, ratio is grown for hood length with whole in described top The ratio of Vehicle length, described before and after ratio on the basis of vehicle roof perpendicular bisector, the ratio of vehicle front and rear part, described in rise Compare the length and the ratio of its height being hood.
CN201610473212.0A 2016-06-23 2016-06-23 A kind of based on the model recognizing method setting up membership function Pending CN106203464A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610473212.0A CN106203464A (en) 2016-06-23 2016-06-23 A kind of based on the model recognizing method setting up membership function

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610473212.0A CN106203464A (en) 2016-06-23 2016-06-23 A kind of based on the model recognizing method setting up membership function

Publications (1)

Publication Number Publication Date
CN106203464A true CN106203464A (en) 2016-12-07

Family

ID=57461912

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610473212.0A Pending CN106203464A (en) 2016-06-23 2016-06-23 A kind of based on the model recognizing method setting up membership function

Country Status (1)

Country Link
CN (1) CN106203464A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109712429A (en) * 2019-02-27 2019-05-03 深圳集智云创科技开发有限公司 A kind of city intelligent shutdown system based on big data

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101964061A (en) * 2010-09-02 2011-02-02 北京航空航天大学 Binary kernel function support vector machine-based vehicle type recognition method
CN104657748A (en) * 2015-02-06 2015-05-27 中国石油大学(华东) Vehicle type recognition method based on convolutional neural network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101964061A (en) * 2010-09-02 2011-02-02 北京航空航天大学 Binary kernel function support vector machine-based vehicle type recognition method
CN104657748A (en) * 2015-02-06 2015-05-27 中国石油大学(华东) Vehicle type recognition method based on convolutional neural network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
杨俊: "基于模糊理论的车型识别", 《广西轻工业》 *
聂小燕等: "基于多特征融合的汽车车型模糊识别方法", 《微计算机应用》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109712429A (en) * 2019-02-27 2019-05-03 深圳集智云创科技开发有限公司 A kind of city intelligent shutdown system based on big data

Similar Documents

Publication Publication Date Title
CN105160309B (en) Three lanes detection method based on morphological image segmentation and region growing
CN104021574B (en) Pavement disease automatic identifying method
CN103034836B (en) Road sign detection method and road sign checkout equipment
CN108090429B (en) Vehicle type recognition method for graded front face bayonet
CN111563412B (en) Rapid lane line detection method based on parameter space voting and Bessel fitting
CN104778444B (en) The appearance features analysis method of vehicle image under road scene
CN103745598B (en) Based on the model recognizing method of front face feature
CN103971097B (en) Vehicle license plate recognition method and system based on multiscale stroke models
CN105335710A (en) Fine vehicle model identification method based on multi-stage classifier
CN107122776A (en) A kind of road traffic sign detection and recognition methods based on convolutional neural networks
CN105930791A (en) Road traffic sign identification method with multiple-camera integration based on DS evidence theory
CN103440503B (en) The recognition methods of a kind of automobile body color detection
CN109101924A (en) A kind of pavement marking recognition methods based on machine learning
CN105046196A (en) Front vehicle information structured output method base on concatenated convolutional neural networks
CN106529532A (en) License plate identification system based on integral feature channels and gray projection
CN103593981B (en) A kind of model recognizing method based on video
CN106529592A (en) License plate recognition method based on mixed feature and gray projection
CN103020582A (en) Method for computer to identify vehicle type by video image
CN110866430A (en) License plate recognition method and device
CN107274668A (en) A kind of congestion in road modeling method based on vehicle detection
CN104966049A (en) Lorry detection method based on images
CN105718866A (en) Visual target detection and identification method
CN105740886A (en) Machine learning based vehicle logo identification method
CN102831600B (en) A kind of image level dividing method merged based on Weighted Cut
CN108388871A (en) A kind of vehicle checking method returned based on vehicle body

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into 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: 20161207