CN105956521A - Vehicle identification method based on images - Google Patents
Vehicle identification method based on images Download PDFInfo
- Publication number
- CN105956521A CN105956521A CN201610254892.7A CN201610254892A CN105956521A CN 105956521 A CN105956521 A CN 105956521A CN 201610254892 A CN201610254892 A CN 201610254892A CN 105956521 A CN105956521 A CN 105956521A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- image
- point
- matching
- identification
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 10
- 230000009466 transformation Effects 0.000 claims abstract description 9
- 230000008569 process Effects 0.000 claims abstract description 7
- 238000005259 measurement Methods 0.000 claims abstract description 6
- 239000000284 extract Substances 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 230000002146 bilateral effect Effects 0.000 abstract 1
- 230000002708 enhancing effect Effects 0.000 abstract 1
- 239000011159 matrix material Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000005286 illumination Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000000746 purification Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- 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
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a vehicle identification method based on images. The method extracts vehicle images from a video and comprises the following steps of: (1) carrying out a preprocessing process and enhancing image contrast; (2) extracting the front part of a vehicle by using an inter-frame difference method; (3) performing similarity measurement by using a distance ratio method; (4) coarsely removing mismatched points by using bilateral matching; (5) finely removing the mismatched points by using an improved RANSAC algorithm; and (6) performing vehicle matching identification again by using affine transformation. The vehicle identification method may well remove the mismatched points, well increases a recognition rate, is simple, and increases identification accuracy.
Description
Technical field
The present invention relates to a kind of vehicle identification method, particularly to a kind of raising vehicle identification method based on image.
Background technology
The charging mechanism of highway department still uses information inside reading IC-card now, then according to the vehicle entrance read
The license board information in place, and combine the type of vehicle, then collect corresponding toll.But, some truck is driven
Member is in order to escape the toll of great number, and they drive fake-licensed car or change IC-card with other vehicles, and this causes to traffic department
The biggest loss.But currently only rely on artificial at backstage comparison vehicle characteristic information, substantial amounts of video is found out illegal car
, it is not only the work wasted time and energy, and also is difficult to find.We thus propose to increase some of vehicle appearance in IC-card
With this, characteristic information, ensures that IC card information and information of vehicles keep concordance.
The matching technique of distinguished point based, is images match, the study hotspot of splicing in recent years.Classical algorithm has SIFT to calculate
Method, although be relative maturity, stable algorithm, but owing to the operand of SIFT algorithm is big, and do not reach real-time and interior
The requirement deposited, is difficult in engineering use.If use SURF algorithm to extract vehicle local feature, but due to each bayonet socket
Photographic head definition different, illumination condition is the most different, it is contemplated that above all situations, and we use ORB algorithm,
It is greatly improved on the extraction rate of characteristic point, and its characteristic point describes son and uses two-value string to represent, the internal memory taken
Relatively small, Hamming distance improves matching efficiency as similarity measurement simultaneously.But we can not ignore a problem, as
Fruit the most only by Hamming distance as similarity measurement it is possible that the situation of Mismatching point.For error hiding situation, both domestic and external
Person proposes many methods removing Mismatching point.Such as it has been proposed that a kind of gradual sampling concordance (PROSAC) algorithm, it is somebody's turn to do
Algorithm can effectively reduce the burden of calculating.Li Zhi proposes a kind of Feature Points Matching based on match strength, this algorithm Matching supporting
Value eliminates Mismatching point, can preferably remove the error hiding situation of one-to-many.Prolong big east and propose a kind of based on offset minimum binary
(PLS) Scale invariant features transform (SIFT) error hiding bearing calibration, the method utilizes the feature point pairs after SIFT coupling
Positional information, redescribed by PLS, utilize the influence function of definition to be rejected by the big feature point pairs of impact, permissible
Effectively remove error hiding.Lei Yuzhen proposes based on consistent (RANSAC) algorithm of stochastic sampling, and he is the space utilizing index point
Feature invariance realizes the Auto-matching of index point to remove Mismatching point.
Said method contributes to remove the Mismatching point occurred in the matching process of distinguished point based, and RANSAC is only
Certain probability obtains believable model, and probability is directly proportional to iterations, if two incoherent objects carry out coupling meeting
Likely obtain some correct match points, affect final discrimination.
Summary of the invention
The technical problem to be solved in the present invention is the defect overcoming prior art, it is provided that a kind of raising vehicle identification side based on image
Method.
In order to solve above-mentioned technical problem, the invention provides following technical scheme:
A kind of raising vehicle identification method based on image of the present invention, extracts for the image of vehicle in video, carry out based on
The method step of the raising vehicle identification of image is as follows:
1) carry out preprocessing process, strengthen picture contrast;
2) pin calculus of finite differences proposes the front face part of vehicle;
3) distance ratio method carries out similarity measurement;
4) bi-directional matching is used to carry out rough removal Mismatching point;
5) improve RANSAC algorithm and accurately remove Mismatching point;
6) affine transformation carries out vehicle match identification again.
The present invention is reached to provide the benefit that: can well remove Mismatching point, well improve discrimination, and use
Method simple, improve the accuracy identified.
Accompanying drawing explanation
Accompanying drawing is for providing a further understanding of the present invention, and constitutes a part for description, with embodiments of the invention one
Rise and be used for explaining the present invention, be not intended that limitation of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of the present invention.
Detailed description of the invention
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are illustrated, it will be appreciated that preferred embodiment described herein is only
For instruction and explanation of the present invention, it is not intended to limit the present invention.
Embodiment
As it is shown in figure 1, the present invention provides a kind of raising vehicle identification method based on image, enter for the image of vehicle in video
Row extracts, and the method step carrying out raising vehicle identification based on image is as follows:
1) carry out preprocessing process, strengthen picture contrast;
2) pin calculus of finite differences proposes the front face part of vehicle;
3) distance ratio method carries out similarity measurement;
4) bi-directional matching is used to carry out rough removal Mismatching point;
5) improve RANSAC algorithm and accurately remove Mismatching point;
6) affine transformation carries out vehicle match identification again.
The present invention is in the step being identified, and the present invention is vehicle match method based on video, and the principle of employing is;Use and work as
Modern popular matching algorithm, but be adjusted on matching strategy, we use Hamming distance to do the most slightly to mate, then
Use bi-directional matching method to carry out preliminary purification, re-use RANSAC algorithm and precisely purify, filter error hiding, the most permissible
In ensureing, point is the most all correct matching double points.It is simultaneous for the weak point of RANSAC algorithm, such as utilizes RANSAC
Algorithm may continue on " interior point " collection so we have proposed " exterior point " as being " interior point ", use a young waiter in a wineshop or an inn
The affine transformation matrix that multiplication estimation is optimal.Illegal vehicle decision plan is based on " interior point " ratio and the ginseng of affine transformation matrix
Number, significantly reduces False Rate.
Concrete employing below step realizes:
The first step: be partitioned into the front face of vehicle
Assume that we obtain real-time traffic video by camera apparatus on traffic route, be then passed through preprocessing process and obtain
Video image the most clearly, and then we accurately to obtain the front face part of vehicle.But moving Object Segmentation is conventional
Method is had powerful connections calculus of finite differences, frame differential method, optical flow method etc..The method of background difference is easy to by illumination, the impact of shade etc..
Optical flow method needs each pixel in image is calculated a motion vector, and due to big by vehicle amount, real-time hardly results in
Ensure.We find vehicle from multitude of video, and generally when entering bayonet socket, car speed is very slow and is not zero, and vehicle stops
Position is essentially identical, thus we can use frame differential method to be partitioned into front face part.The benefit of employing frame differential method is
The image that the same time gathers, insensitive to the change such as illumination, time, and algorithm is simple, real-time is good.By to a large amount of cards
The research of the current vehicle of mouth, we use the vehicle location that kth frame and kth-5 frame difference can substantially find, then use form
Student movement is calculated, and searches connected region, can navigate to the position of front face.
Second step: extract vehicle characteristics
By above step, we are partitioned into the front face part of vehicle, then use the most popular algorithm to carry out
The feature extraction of front face.
3rd step: the thick matching process of vehicle
We use Hamming distance as Feature Points Matching strategy.First with the vehicle characteristic information extracted and now vehicle
Characteristic information do and once mate, owing to not being the target of Same Scene shooting, illumination, the change such as environment necessarily causes a lot of mistake
Error hiding pair.The method eliminating erroneous matching pair has a variety of, in the set of matches of distance ratio method error hiding most for existence not
Can play good effect, we use bi-directional matching method.Its thought is: be relation one to one according to Feature Points Matching,
First do an arest neighbors with the feature point set of target image characteristics point set with image to be matched to mate, the most again with image to be matched
Feature point set does an arest neighbors and mates with the feature point set of target image, and we select the common factor of twice matched data, so pick
Except non-intersection point, those non-intersection points are exactly the point of error hiding, but yet suffer from Mismatching point in intersection point.
4th step a: determination strategy
For above occur simultaneously present in Mismatching point, we use RANSAC algorithm.RANSAC algorithm is disappearing of a kind of classics
Except erroneous matching algorithm, the advantages such as matching precision is high, strong robustness, it is by the model parameter of estimation optimum, and searching meets should
At most counting of model parameter.But due to the defect of RANSAC algorithm, it only has certain probability to obtain optimal transformation parameter,
In order to reduce the probability of mistake judgement, we select to intercept the most continuously 25 images, repeatedly use above-mentioned algorithm,
Exist and be once judged to normally, i.e. be considered as normal vehicle, be otherwise illegal vehicle.
5th step: secondary determination strategy
Owing to there may be picture quality, the shooting congruent reason of vehicle picture, same car correct match point number is little, root
According to an above-mentioned determination strategy, illegal vehicle can be judged as.For this situation, we send out by observing a large amount of bayonet socket video frequency vehicles
Existing, the photographic head of bayonet socket, be entirely and shoot from the left front of vehicle, and shooting distance very close to, we can be vehicle figure
As little degree rotates and scaling in regarding the face of rigid body as.Utilize this feature on " interior point " collection, use method of least square estimation
One optimum affine transformation matrix, by observing the anglec of rotation and the change of scale of affine matrix, can judge as secondary
Strategy.By comparing yardstick ratio and the meansigma methods calculating affine transformation matrix and whether fall in [0.5,1.5] and affine angle existing
In [-50,50].If all falling within appointment threshold value, then it is judged to normal vehicle;Otherwise, it is illegal vehicle.
The present invention can well remove Mismatching point, well improves discrimination, and the method used is simple, improves and identifies
Accuracy.
Finally it is noted that the foregoing is only the preferred embodiments of the present invention, it is not limited to the present invention, although
Being described in detail the present invention with reference to previous embodiment, for a person skilled in the art, it still can be to front
State the technical scheme described in each embodiment to modify, or wherein portion of techniques feature is carried out equivalent.All at this
Within bright spirit and principle, any modification, equivalent substitution and improvement etc. made, should be included in protection scope of the present invention
Within.
Claims (1)
1. a raising vehicle identification method based on image, it is characterised in that extract for the image of vehicle in video,
The method step carrying out raising vehicle identification based on image is as follows:
1) carry out preprocessing process, strengthen picture contrast;
2) pin calculus of finite differences proposes the front face part of vehicle;
3) distance ratio method carries out similarity measurement;
4) bi-directional matching is used to carry out rough removal Mismatching point;
5) improve RANSAC algorithm and accurately remove Mismatching point;
6) affine transformation carries out vehicle match identification again.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610254892.7A CN105956521B (en) | 2016-04-22 | 2016-04-22 | A kind of raising vehicle identification method based on image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610254892.7A CN105956521B (en) | 2016-04-22 | 2016-04-22 | A kind of raising vehicle identification method based on image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105956521A true CN105956521A (en) | 2016-09-21 |
CN105956521B CN105956521B (en) | 2019-06-18 |
Family
ID=56915778
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610254892.7A Expired - Fee Related CN105956521B (en) | 2016-04-22 | 2016-04-22 | A kind of raising vehicle identification method based on image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105956521B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114743023A (en) * | 2022-06-14 | 2022-07-12 | 安徽大学 | Wheat spider image detection method based on RetinaNet model |
-
2016
- 2016-04-22 CN CN201610254892.7A patent/CN105956521B/en not_active Expired - Fee Related
Non-Patent Citations (2)
Title |
---|
翟亚丹: "基于SIFT特征的车牌识别***的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
钱志伟: "智能交通***中车型识别的研究与应用", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114743023A (en) * | 2022-06-14 | 2022-07-12 | 安徽大学 | Wheat spider image detection method based on RetinaNet model |
Also Published As
Publication number | Publication date |
---|---|
CN105956521B (en) | 2019-06-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103208185B (en) | A kind of vehicle detection at night method and system based on car light identification | |
Yuan et al. | Robust lane detection for complicated road environment based on normal map | |
CN111210477B (en) | Method and system for positioning moving object | |
CN115717894B (en) | Vehicle high-precision positioning method based on GPS and common navigation map | |
CN103971097B (en) | Vehicle license plate recognition method and system based on multiscale stroke models | |
CN103500322B (en) | Automatic lane line identification method based on low latitude Aerial Images | |
CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
CN108596165B (en) | Road traffic marking detection method and system based on unmanned plane low latitude Aerial Images | |
CN104978567B (en) | Vehicle checking method based on scene classification | |
CN101980245B (en) | Adaptive template matching-based passenger flow statistical method | |
CN104200495A (en) | Multi-target tracking method in video surveillance | |
CN105701448B (en) | Three-dimensional face point cloud nose detection method and the data processing equipment for applying it | |
CN103593832A (en) | Method for image mosaic based on feature detection operator of second order difference of Gaussian | |
WO2015096507A1 (en) | Method for recognizing and locating building using constraint of mountain contour region | |
CN104112122A (en) | Vehicle logo automatic identification method based on traffic video | |
CN102043953A (en) | Real-time-robust pedestrian detection method aiming at specific scene | |
CN102289948A (en) | Multi-characteristic fusion multi-vehicle video tracking method under highway scene | |
CN103902985B (en) | High-robustness real-time lane detection algorithm based on ROI | |
CN110866430A (en) | License plate recognition method and device | |
CN105185118A (en) | Vehicle peccancy detection device and method based on video detection | |
CN111178161A (en) | Vehicle tracking method and system based on FCOS | |
CN107729853A (en) | A kind of automatic identifying method suitable for the narrow tuning drive gear formula instrument of transformer station | |
CN113112483B (en) | Rigid contact net defect detection method and system based on similarity measurement | |
CN104408741A (en) | Video global motion estimation method with sequential consistency constraint | |
CN109934108A (en) | The vehicle detection and range-measurement system and implementation method of a kind of multiple target multiple types |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
DD01 | Delivery of document by public notice |
Addressee: Pan Jiwu Document name: Notification to Pay the Fees |
|
DD01 | Delivery of document by public notice | ||
DD01 | Delivery of document by public notice | ||
DD01 | Delivery of document by public notice |
Addressee: Pan Jiwu Document name: payment instructions |
|
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190618 |
|
DD01 | Delivery of document by public notice | ||
DD01 | Delivery of document by public notice |
Addressee: ANHUI University Person in charge of patentsThe principal of patent Document name: Notice of Termination of Patent Rights |