CN106503748A - A kind of based on S SIFT features and the vehicle targets of SVM training aids - Google Patents
A kind of based on S SIFT features and the vehicle targets of SVM training aids Download PDFInfo
- Publication number
- CN106503748A CN106503748A CN201610972305.8A CN201610972305A CN106503748A CN 106503748 A CN106503748 A CN 106503748A CN 201610972305 A CN201610972305 A CN 201610972305A CN 106503748 A CN106503748 A CN 106503748A
- Authority
- CN
- China
- Prior art keywords
- image
- window
- vehicle
- feature
- car
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to computer vision field, refer in particular to a kind of based on S SIFT features and the vehicle targets of SVM training aids.The algorithm is comprised the following steps：1. License Plate, the car face region intercepting and 3. for 2. including vehicle information are set up the S SIFT features model of vehicle area image and carry out vehicle cab recognition with reference to SVM training aids.The method proposed in the present invention can actually be embedded in FPGA realizations, apply to, in the camera or video camera of vehicle cab recognition function with real-time output image function, to effectively improve the accuracy and reliability of system so as to meet real-time demand.
Description
Technical field
The present invention relates to computer vision field, refers in particular to a kind of based on S-SIFT (sparse scale invariant
Feature transform) feature and SVM training aids vehicle targets.
Background technology
With developing rapidly for the industries such as modern transportation, security protection, vehicle automatic identification technology is increasingly subject to people's
Pay attention to, be one of the important subject of computer vision and mode identification technology in intelligent transportation field in recent years.Vehicle
Automatic recognition system can be used for the vehicle management in the places such as toll station, parking lot, crossroad, it can also be used to modernize little
The vehicles while passing management of area or industrial park, for public safety, community security protection, road traffic and parking lot vehicle management are all
There is important facilitation.
Vehicle cab recognition generally comprises the research of three aspects, and domestic and international experts and scholars have also carried out substantial amounts of work, mainly
Including：The positioning of car plate and identification, the detection and identification of logo, and the classification of vehicle size.Wherein, according to Chinese herbaceous peony face image
To recognize that the research method of concrete vehicle is the hot research direction of recent years.
In reality, the usual background complexity of picture that gathers in actual parking lot and cell, uneven illumination, resolution ratio is low,
Vehicle is old, vehicle is dirty etc., and the angle of vehicles while passing is as a rule all inconsistent, and above-mentioned situation is all brought very to vehicle cab recognition
Big difficulty, prior art not yet have suitable method effectively overcome above-mentioned technical problem.
Content of the invention
The technical problem to be solved in the present invention is：These the specific difficult points existed for existing model recognition system are asked
Topic, propose a kind of based on S-SIFT features and the vehicle targets of SVM training aids, the vehicle so as to improve monitoring system is known
Other order of accuarcy, and make which meet real-time demand.
For solving above-mentioned technical problem, the invention provides a kind of vehicle based on S-SIFT features and SVM training aids is known
Other algorithm, the algorithm specifically include following steps：
Step S1：License Plate, it include：
Step S1.1：Training car plate sample characteristics are extracted and feature organization, including taking out arbitrarily normal GB car manually
Board, the license plate image for taking out is integrated channel characteristics extract and based on Adaboost algorithm train detector；
Step S1.2：The detection positioning of car plate, including being scanned to target image to obtain just positioning licence plate image, and
The just positioning licence plate image is carried out after non-maxima suppression algorithm process, then result is carried out based on Hough transformation
Slant correction obtains the license plate image after secondary positioning；
Step S2：Car face area image is intercepted, and which is included in after orienting accurate car plate position, according to the length of car plate
Width, chooses certain ratio and enters to drive a vehicle the intercepting of face area image；
Step S3：Set up the S-SIFT characteristic models of vehicle area image and vehicle cab recognition is carried out with reference to SVM training aids.
Used as the further improvement of technical solution of the present invention, the described pair of license plate image for taking out is integrated channel characteristics
Extracting includes setting up LUV passages, gradient magnitude passage and histogram of gradients passage first respectively, and according to LUV passages, gradient width
Value passage and histogram of gradients passage obtain the license plate image feature of respective channel；；
Described based on Adaboost algorithm training detector include：
Training stage, strong classifier is gone out as detector to the integrating channel features training that extracts by the use of Adaboost；
In the differentiation stage, the integrating channel feature for detecting positioning licence plate window is calculated, is given a mark with strong classifier, institute
Marking is stated for differentiating the Confidence of car plate position, one section of that frame of video mid-score highest or a few two field pictures is finally stored.
Used as the further improvement of technical solution of the present invention, step S1.2 is specifically included：1) using sliding window method to warp
The target image for crossing enhancing process is scanned, and scanning truncated picture every time is integrated channel characteristics calculating, with
The strong detector that AdaBoost Algorithm for Training goes out is compared, and chooses similarity highest image-region as first positioning licence plate figure
Picture；
2) the first positioning image that detector is exported is carried out the first positioning result after non-maxima suppression process to be based on
The slant correction of Hough transformation, and to slant correction after image be integrated again channel characteristics extract after be input into strong detector
Secondary positioning is carried out, the license plate image after secondary positioning is obtained.
Used as the further improvement of technical solution of the present invention, the non-maxima suppression algorithm steps are as follows：
(1) initial detecting window is sorted from high to low according to detection score；
(2) using the 1st initial detecting window as current suppression window；
(3) non-maxima suppression：The home window for currently suppressing window low all detection score ratios is used as suppressed window
Mouthful, calculate the current overlapping area ratio for suppressing window and suppressed window：The union of the common factor/area of area, and reject weight
Close window of the area ratio higher than given threshold；
(4) terminate if only last initial detecting window is remained, otherwise according to the order for sequencing, take next not by
The window of suppression goes to step (3) as window is suppressed.
Used as the further improvement of technical solution of the present invention, step S3 sets up the S-SIFT features of car face area image
Model is simultaneously carried out vehicle cab recognition with reference to SVM training aids and specifically includes following steps：
S3.1 sets up the S-SIFT characteristic models of car face area image, and which is included using S-SIFT algorithm process car faces region
Image SIFT feature describes son and obtains sparse coding, and using pond method statistic sparse coding result, obtains car face administrative division map
The summary statistics feature of picture；
S3.2 car face area image vehicle tagsorts and identification, after which is included in extraction S-SIFT features, using SVM
Training aids is trained classification；After training classification, by the input SVM instructions of the car face area image comprising vehicle characteristic information for intercepting
Practice in device, and export the vehicle information of identification.
Compared with prior art, the invention has the advantages that：
1st, the bayonet socket camera vehicle characteristic image for having intercepted is combined, the low layer SIFT feature vector of target image is extracted,
Trained acquisition encoder dictionary and sparse SIFT feature, obtain deeper level characteristics of image again, are become with adapting to different visual angles, illumination
Change, shade, the complex scene such as block, further improve discrimination；
2nd, sparse SIFT feature classification is realized using linear SVM, reduces time complexity, it is ensured that real-time.
3rd, algorithm reliability proposed by the present invention is high, debates that resolution is good, and robustness is good, while step calculates simple, can protect
High efficiency is held, real-time can also meet demand.
Description of the drawings
Fig. 1 is total algorithm flow chart described in the present embodiment；
Fig. 2 is three kinds of integration feature channel images of license plate image described in the present embodiment.
Fig. 3 is four direction gradient operator schematic diagram described in the present embodiment.
Fig. 4 is pixel direction schematic diagram described in the present embodiment.
Fig. 5 is the license plate image of Hough transformation slant correction described in the present embodiment.
Fig. 6 is each examples of parameters of the car face image scope intercepted described in the present embodiment.
Specific embodiment：
The present invention is described in further details in conjunction with accompanying drawing, embodiment of the present invention particular content is based on S- for a kind of
SIFT feature and the vehicle targets of SVM training aids, the vehicle targets are comprised the following steps：
S1 License Plates；
S1.1 training car plate sample characteristics are extracted and feature organization；
S1.1.1 takes out arbitrarily normal GB car plate manually；
S1.1.2 is integrated channel characteristics extraction to the license plate image for taking out；
Integrating channel feature is proposed in 2009 by Doll á r P et al., is generally used for pedestrian detection earliest, is to comment at present
Estimate the preferable detective operators of effect, the basic thought of integrating channel feature is by various linear and non-thread are carried out to tablet pattern
Property conversion, a lot of common features of image, such as local summation, histogram, Haar and their mutation can be by products
Component is fast and efficiently calculating.An input picture matrix I is given, its corresponding passage refers to original input picture
Certain output response.For gray-scale map, its corresponding access matrix C=I, i.e. artwork itself；
For coloured picture, each of which Color Channel all corresponds to a passage, and other similar passages can pass through various linear
Be calculated with non-linear method, make certain path computation function of Ω representative images, then corresponding channel C=Ω (I).
In the calculation, different conversion can form different channel types, choose 3 kinds of different passages and make in the present invention
For integrating channel feature, to ensure its accuracy.Wherein LUV Color Channels can describe car plate brightness well and colourity becomes
Change, gradient magnitude passage reflects the profile of car plate well, histogram of gradients passage is then comprehensive right from different gradient directions
The change of car plate position and attitude is described, and 3 kinds of passage transform effects are as shown in Figure 2.
The foundation of S1.1.2.1LUV passages
In image procossing, LUV color spaces (full name CIE1976 (L*, U*, V*)) are better than rgb color space.LUV colors
The purpose of color space is to set up the color space unified with the vision of people, possesses between uniformity and uniformity and each color component
Uncorrelated.In LUV color spaces, L represents that brightness, U, V represent colourity.General pattern color is all RGB color, leads to
Cross equation below to may switch in LUV color spaces.
L, U, V passage being finally calculated in LUV color spaces.
S1.1.2.2 gradient magnitude passages
Gradient magnitude is a kind of description method for Image Edge-Detection.In piece image, each pixel has eight neighbours
Domain and four edge direction detections.In order to detect that in pixel X-direction, Y-direction, Z-direction edge, the present invention are implemented
Example using in the window respectively the first-order partial derivative finite difference average of calculating X-direction Y-direction, Z-direction determining pixel
The method of gradient magnitude.The gradient operator of four direction is respectively shown in Fig. 3.It is 3 × 3 window centers that wherein I [i, j] is coordinate
The gray value of pixel, the gradient magnitude of pixel centered on M [i, j], its computing formula are as follows, on corresponding four direction
Computing formula be：
The gradient magnitude figure that entire image is finally obtained by above-mentioned formula.
S1.1.2.3 histogram of gradients passages
Histogram of gradients thought source in gradient orientation histogram (Histograms of Oriented Gradients,
HOG) to be Dalal in 2005 et al. be used for pedestrian by it recognizes and gains the name.HOG as a kind of local feature description son, to direction,
Yardstick, illumination-insensitive.HOG was successfully applied to recognition of face by Deniz et al. later, had obtained reasonable effect.Gradient
Histogram feature extraction process is as follows：
Step 1 takes 3 × 3 neighborhood of pixels centered on image I [i, j] as sampling window.
Step 2 calculates the gradient direction θ [i, j] and gradient magnitude M [i, j] of the pixel [i, j].
θ [i, j]=arctan (I [i, j+1]-I [i, j-1])/I [i+1, j]-I [i-1, j]
As shown in figure 4, arrow represents the direction of the pixel [i, j].
Gradient direction is divided into 6 directions by step 3, will 180 ° be divided into 6 parts, 30 ° of equispaced.According to oval circle
Gauss weighting scope all of gradient direction angle identical pixel gradient magnitude in the neighborhood of pixels is added.
Gradient magnitude on 6 directions of the last statistics of step 4 adds up and obtains the gradient width on 6 directions of entire image
Value figure.
The image including 10 passages such as LUV passages, gradient magnitude passage, histogram of gradients passages for finally obtaining is as schemed
Shown in 2.
S1.1.3 trains detector based on Adaboost algorithm
Training stage, strong classifier is gone out to the integrating channel features training that extracts using Adaboost, differentiating stage, meter
The integrating channel feature for detecting positioning licence plate window is calculated, is given a mark with strong classifier and (is differentiated the confidence of car plate position
Degree), finally store one section of that frame of video mid-score highest or a few two field pictures.
AdaBoost algorithms are proposed in 1996 by Schapire, Freund et al., its essence is the classification of Weak Classifier
Learning process, is one kind of ensemble machine learning method, with the few structure for Weak Classifier of computational efficiency height, regulation parameter
Compatibility is made by force and to sample priori and the low advantage of data format requirement, therefore, is widely popularized.
In AdaBoost algorithms, each feature corresponds to a Weak Classifier, but is not that each feature can describe prospect well
The characteristics of target.How optimal characteristics are picked out from big measure feature and be fabricated to Weak Classifier, then integrated by Weak Classifier,
High-precision strong classifier is finally obtained, is AdaBoost Algorithm for Training processes key issue to be solved.
The definition of Weak Classifier is：
Wherein, fjRepresent a feature, pjRepresent inequality direction, θjRepresent threshold value.
AdaBoost Algorithm for Training processes are as follows：
(1) n sample image, x are giveniIt is input sample image, yiIt is class formative, wherein yi=0 is expressed as negative sample
This, yi=1 is expressed as positive sample.
(2) weight is initialized：
Wherein m and l is respectively the quantity of incorrect car plate sample and correct car plate sample, n=m+l.
(3) For t=1,2,3 ..., T
1. normalized weight：Wherein ωtFor statistical distribution.
2. integrating channel feature j is randomly choosed：
Random selection passage index bink(k=1,2 ..., 10)；
Random selection rectangular area RectjAnd calculate pixel value sum；
3. pair each feature j, trains a Weak Classifier hj, calculate corresponding ωtError rate：
εj=∑iωi|hj(xi)-yi|
4. minimal error rate ε is selectedtWeak Classifier ht.
5. weight is updated：Wherein, work as xiWhen correctly being classified, ei=0, conversely, ei=1；
(4) final strong classifier is h (x):
Wherein,
S1.2. the detection of car plate is positioned；
S1.2.1 sliding window methods are scanned to target image, obtain just positioning licence plate image；
Fixed proportion of the embodiment of the present invention according to domestic car plate, sets the sliding window of a fixed size, from acquisition video
Image apex is proceeded by and is scanned one by one, and in order to improve scanning accuracy, it is 4 pixels generally to arrange sliding window step-length, will every time
Scanning truncated picture is integrated channel characteristics calculating, compares with the strong detector that AdaBoost Algorithm for Training goes out, obtains
(i.e. similarity highest) image-region, i.e. preliminary judgement to highest scoring is car plate position, intercepts the figure of the highest scoring
As region for just positioning image and exports the strong detector.
The first positioning image that detector is exported is carried out the first positioning result after non-maxima suppression process by S1.2.2 to be carried out
License plate image after secondary positioning is obtained based on the slant correction of Hough transformation；
Non-maxima suppression apply in object detection quite varied, its main purpose be in order to eliminate unnecessary interference because
Element, finds the position of optimal object detection.Non-maxima suppression is the last handling process of detection, is one of key link.
Heuristic window blending algorithm is fine to non-coincidence target detection effect, but for vehicle license plate detection discomfort
Close.Initial detecting window is divided into several misaligned subsets, then calculates each subset by heuristic window blending algorithm
Center, finally each subset only retain a detection window, it is clear that heuristic window blending algorithm easily causes a large amount of missing inspections.
Dalal etc. proposes average drifting non-maxima suppression method, and which not only calculates complexity, needs detection window exists
3-dimensional space (abscissa, ordinate, yardstick) represents that detection fraction is changed, the uncertain matrix of calculating, iteration optimization, but also
The parameter for needing adjustment to be much associated with the step-length of detector etc., therefore, less use at present.
Currently, most target detection commonly uses the non-maxima suppression algorithm based on Greedy strategy, because it is simple
Single efficient, key step is as follows：
(1) initial detecting window is sorted from high to low according to detection fraction；
(2) using the 1st initial detecting window as current suppression window；
(3) non-maxima suppression.The home window for currently suppressing window low all detection score ratios is used as suppressed window
Mouthful.Calculate the current overlapping area ratio for suppressing window and suppressed window：The union of the common factor/area of area.Reject and overlap
Window of the area ratio higher than given threshold；
(4) terminate if only last initial detecting window is remained, otherwise according to the order for sequencing, take next not by
The window of suppression goes to step (3) as window is suppressed.
The embodiment of the present invention equally uses the simple efficient non-maxima suppression algorithm based on Greedy strategy, and will
License plate image after non-maxima suppression process carries out the slant correction based on Hough transformation.
Hough transformation is a kind of strong feature extracting method, it using topography's information effectively accumulate all can
Can model instance foundation, this causes it easily can obtain extra information from external data, again can observantly from
Only effective information is presented in the example of some.Hough transformation is commonly utilized in shape in computer vision, position, geometry
In the judgement of transformation parameter.Since proposing from Hough transformation, which is widely used.In recent years, experts and scholars were to suddenly
The theory property of husband's conversion has carried out further discussion again with application process.Hough transformation is used as a kind of effectively identification straight line
Algorithm, with good anti-interference and robustness.
Mapping of the Hough transformation method comprising a feature from image space to the set at parameter space midpoint.Each
Point in individual parameter space characterizes an example of model in image space, and characteristics of image is mapped to ginseng using a function
In the middle of number space, this function produces all of parameter group for being capable of the compatibility characteristics of image that observes and the model that assumes
Close.Each characteristics of image will produce a different plane in the parameter space of multidimensional, but be produced by all characteristics of image
One section of the raw example for belonging to same model can all be intersected in the point for describing common example, Hough transformation basic
It is to produce these planes and recognize intersecting therewith parameter point.
License plate image after the slant correction based on Hough transformation is the image after secondary system positioning, Hough transformation
The license plate image example of slant correction is as shown in Figure 5.
License plate image after the secondary positioning that S1.2.3 will be exported is input into strong detector, draws final positioning licence plate result.
By through non-maxima suppression process and slant correction based on Hough transformation after the output of figure car plate picture after again
Strong detector is input into after being integrated channel characteristics extraction carries out secondary positioning, including：Strong with what AdaBoost Algorithm for Training went out
Detector is compared, and obtains (i.e. similarity highest) image-region of highest scoring, that is, be judged to car plate position, and intercepting should
The image-region of highest scoring is secondary positioning image output detector, obtains final positioning result.
The intercepting of S2 car face area images；
After accurate car plate position is oriented, generally according to the length and width of car plate, choose certain ratio and enter to drive a vehicle face figure
The intercepting of picture, so that the positive face of bayonet socket camera shoots vehicle region as an example, generally, respectively intercepts 1.3 times of cars with the right and left of car plate
Length of the board length for car face region, and vehicle picture height is 0.8 times of car plate length on car plate, is 0.3 times of car under car plate
Board length, parameter is according to can need real-time adjustment.The each examples of parameters of the car face image scope of intercepting is as shown in Figure 6.
S3 sets up the S-SIFT characteristic models of vehicle area image and carries out vehicle cab recognition with reference to SVM training aids；
For the car face area image including vehicle information for having intercepted, need to be identified exporting final car
Type result, the embodiment of the present invention are theoretical based on deep learning, it is proposed that a kind of model recognizing method based on sparse SIFT feature.
The model recognizing method extracts the low layer SIFT feature vector of target image, then trained acquisition encoder dictionary and sparse first
SIFT feature, obtains deeper level characteristics of image, to adapt to different visual angles, illumination variation, shade, the complex scene such as block, enters
One step improves discrimination；Finally sparse SIFT feature classification is realized with linear SVM, reduce time complexity, it is ensured that real
Shi Xing.Model recognizing method is comprised the following steps that：
S3.1 sets up S-SIFT characteristic models
S3.1.1S-SIFT characteristics algorithms：
S-SIFT characteristics algorithms are on the basis of image SIFT feature, further train super complete dictionary base, in L1 models
The sparse SIFT of the lower coding of number constraint, it is possible to achieve higher level vehicle image is abstract.
Define matrix X and include M S-SIFT Feature Descriptor of the image in D dimensional feature spaces, X=(x1..., xM)T, then X
Can be expressed as：
X=WC
In formula：W is the coefficient of sparse coding, C=(c1..., cK)TIt is K base vector.The sparse coding for solving X can be with table
Levy, for following formula, optimization problem is solved to W and C：
In formula：| | | | and | | represent L2 norms and L1 norms respectively.From L1 norm constraint properties, penalty term |
Wm| ensure that the openness of coding result, sparse coefficient β control | Wm| weight, i.e., openness.Base vector was complete (K>
D), therefore use CgL2 norm constraints avoid trivial solution.
Although the formula of solution | | Ck| |≤1,When W and C simultaneously change, object function is not convex optimization problem,
But when fixing W and C respectively, object function deteriorates to the convex function with regard to C and W respectively.During fixed W, object function is deteriorated to
Least square problem with regard to C：
Lagrange duality algorithm rapid solving can be used.Fixed C, object function are deteriorated to individually to each WmAsk most
The linear regression problem of excellent solution：
Can be solved with characteristic symbol searching algorithm.
The embodiment of the present invention test in choose D=128, β=0.15, K from 8,32,128,512,1024 totally 5 kinds coding
Dimension.M depends on image size.By taking the image of 256 × 256 pixels as an example, SIFT tile sizes are defined as 16 × 16
Pixel, step-length are 6, then laterally make (256 16)/6=40 coupling, and longitudinal direction is madeSecondary coupling, M=40 × 40,
I.e. 1600, with 512 dimension S-SIFT algorithm process SIFT feature, the sparse coding of final output is the vector of 1600 512 dimensions.
S3.1.2 ponds
Pond is the process for counting sparse coding result, and which simulates the physiological mechanism of human eye vision cortex, it is possible to reduce defeated
Incoming vector dimension, advantageously reduces the time complexity of training grader.With 256 × 256 images described in the embodiment of the present invention it is
Example, its sparse SIFT coding dimension are 1600 × 512=819200, train grader of the input vector dimension more than 800,000
Difficulty is very big, and over-fitting easily occurs.Therefore, using pond method, the summary statistics for obtaining piece image are special for the present embodiment
Levy, not only reduce the difficulty of training grader, and avoid Expired Drugs.
Pond method common at present has average pondization and maximum Chi Hua etc., and computational methods are:
Average pond：
Maximum pond：pj=max | w1j| ..., | wMj|}
In formula：WmIt is sparse coding vector；P is pond result；wijRepresent j-th element of i-th sparse coding vector.
The simple Linear SVM grader of the feature of Chi Huahou can just reach preferable classifying quality, and time complexity is only O (n).
S3.2 car face area image vehicle tagsorts
The parameter training of S3.2.1SVM graders
Vehicle tagsort is primarily referred to as by the car face area image including vehicle information to be identified and through study
Training vehicle feature carries out contrast by a certain algorithm to be identified.Conventional grader mainly includes minimum distance classification
Device, K- nearest neighbor classifiers, Bayes classifier, decision tree, Adaboost cascade classifiers, artificial neural network and support to
Amount machine (SVM).Train as needed classification vehicle picture characteristics and different classifications device the characteristics of, the present invention is main using supporting
Vector machine is classified.The core concept of SVMs is using an Optimal Separating Hyperplane when the curved surface that makes decision, comes most
Change greatly the Edge Distance of positive class and negative class.
In the present invention, Q={ (x are definedi, yi), i=1 ..., n, wherein Q are n input data point sets；xiRepresent input
Variable；yiRepresent desired value, y in two class problemsi∈ { 1, -1 }.Classification function is defined as：
In formula,Represent the mapping from the input space to high-dimensional feature space.According to sequential minimal optimization algorithm
(sequential minimal optimization, SMO) can be as follows in the hope of decision function
In formula：aiRepresent Lagrange multiplier；k〈xi, x>Kernel function is represented, is mapped to after higher dimensional space for quick calculating
Two vectorial inner products.The linear core of common kernel function, Gaussian kernel, polynomial kernel.Using Non-linear Kernel SVM classifier, its
Training time complexity is O (n2～n3), classification time complexity is O (n),；And adopt linear kernel then can answer the training time
Miscellaneous degree is reduced to O (n), and classification time complexity is still O (n).In actual applications, generally adopt linear kernel function to improve instruction
Practice efficiency, it is ensured that system real time.
In sum, after feature is extracted, classification is trained using SVM.After training classification, will intercept comprising car
In the car face area image input SVM training aids of type characteristic information, and export the vehicle information of identification.
The method proposed in the present invention can actually be embedded in FPGA realizations, apply to the car with real-time output image function
In the camera of type identification function or camera supervised system.
Those skilled in the art will be clear that the scope of the present invention is not restricted to example discussed above, it is possible to which which is carried out
Some changes and modification, without deviating from the scope of the present invention that appended claims are limited.Although oneself is through in accompanying drawing and explanation
Illustrate and describe the present invention in book in detail, but such explanation and description are only explanations or schematic, and nonrestrictive.
The present invention is not limited to the disclosed embodiments.
Claims (5)
1. vehicle targets of a kind of S-SIFT features and SVM training aids, it is characterised in that comprise the following steps：
Step S1：License Plate, it include：
Step S1.1：Training car plate sample characteristics are extracted and feature organization, including taking out arbitrarily normal GB car plate, right manually
The license plate image for taking out is integrated channel characteristics and extracts and train detector based on Adaboost algorithm；
Step S1.2：The detection positioning of car plate, including being scanned to target image to obtain just positioning licence plate image, and to institute
After positioning licence plate image at the beginning of stating carries out non-maxima suppression algorithm process, then result is carried out the inclination based on Hough transformation
Correction obtains the license plate image after secondary positioning；
Step S2：Car face area image is intercepted, and which is included in after orienting accurate car plate position, according to the length and width of car plate, choosing
Take certain ratio to enter to drive a vehicle the intercepting of face area image；
Step S3：Set up the S-SIFT characteristic models of vehicle area image and vehicle cab recognition is carried out with reference to SVM training aids.
2. according to claim 1 a kind of based on the vehicle targets for integrating feature passage and SVM training aids, its feature
It is, the described pair of license plate image for taking out is integrated channel characteristics and extracts and include setting up LUV passages, gradient width first respectively
Value passage and histogram of gradients passage, and respective channel is obtained according to LUV passages, gradient magnitude passage and histogram of gradients passage
License plate image feature；
Described based on Adaboost algorithm training detector include：
Training stage, strong classifier is gone out as detector to the integrating channel features training that extracts by the use of Adaboost；
In the differentiation stage, calculate and detect the integrating channel feature of positioning licence plate window, given a mark with strong classifier, described beat
It is divided into the Confidence for differentiating car plate position, finally stores one section of that frame of video mid-score highest or a few two field pictures.
3. according to claim 2 a kind of based on the vehicle targets for integrating feature passage and SVM training aids, its feature
It is, step S1.2 is specifically included：
1) it is scanned to passing through the target image for strengthening process using sliding window method, scanning truncated picture every time is integrated
Channel characteristics are calculated, and are compared with the strong detector that AdaBoost Algorithm for Training goes out, and choose similarity highest image-region
As first positioning licence plate image；
2) the first positioning image that detector is exported is carried out the first positioning result after non-maxima suppression process is carried out based on Hough
The slant correction of conversion, and to slant correction after image be integrated after channel characteristics are extracted again and be input into strong detector and carry out
Secondary positioning, obtains the license plate image after secondary positioning.
4. according to claim 1 a kind of based on the vehicle targets for integrating feature passage and SVM training aids, its feature
It is, the non-maxima suppression algorithm steps are as follows：
(1) initial detecting window is sorted from high to low according to detection score；
(2) using the 1st initial detecting window as current suppression window；
(3) non-maxima suppression：The home window for currently suppressing window low all detection score ratios is used as suppressed window, meter
Calculate the current overlapping area ratio for suppressing window and suppressed window：The union of the common factor/area of area, and reject overlapping area
Window of the ratio higher than given threshold；
(4) terminate if only last initial detecting window is remained, otherwise according to the order for sequencing, take the next one and be not suppressed
Window as suppress window, go to step (3).
5. according to claim 1 a kind of based on the vehicle targets for integrating feature passage and SVM training aids, its feature
It is, step S3 is set up the S-SIFT characteristic models of car face area image and carries out vehicle cab recognition tool with reference to SVM training aids
Body is comprised the following steps：
S3.1 sets up the S-SIFT characteristic models of car face area image, and which is included using S-SIFT algorithm process car face area images
SIFT feature describes son and obtains sparse coding, and using pond method statistic sparse coding result, obtains car face area image
Summary statistics feature；
S3.2 car face area image vehicle tagsorts and identification, after which is included in extraction S-SIFT features, are trained using SVM
Device is trained classification；After training classification, by the input SVM training aids of the car face area image comprising vehicle characteristic information for intercepting
In, and export the vehicle information of identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610972305.8A CN106503748A (en) | 2016-11-07 | 2016-11-07 | A kind of based on S SIFT features and the vehicle targets of SVM training aids |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610972305.8A CN106503748A (en) | 2016-11-07 | 2016-11-07 | A kind of based on S SIFT features and the vehicle targets of SVM training aids |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106503748A true CN106503748A (en) | 2017-03-15 |
Family
ID=58323190
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610972305.8A Pending CN106503748A (en) | 2016-11-07 | 2016-11-07 | A kind of based on S SIFT features and the vehicle targets of SVM training aids |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106503748A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107085074A (en) * | 2017-04-19 | 2017-08-22 | 中国科学技术大学 | A kind of method for monitoring motor-vehicle tail-gas of classifying |
| CN107092876A (en) * | 2017-04-12 | 2017-08-25 | 湖南源信光电科技股份有限公司 | The low-light (level) model recognizing method combined based on Retinex with S SIFT features |
| CN107886539A (en) * | 2017-10-19 | 2018-04-06 | 昆明理工大学 | High class gear visible detection method under a kind of industrial scene |
| CN108388919A (en) * | 2018-02-28 | 2018-08-10 | 大唐高鸿信息通信研究院（义乌）有限公司 | The identification of vehicle-mounted short haul connection net security feature and method for early warning |
| CN108537151A (en) * | 2018-03-27 | 2018-09-14 | 上海小蚁科技有限公司 | A kind of non-maxima suppression arithmetic unit and system |
| CN109840809A (en) * | 2019-03-12 | 2019-06-04 | 中国联合网络通信集团有限公司 | Advertisement placement method and device |
| CN111178357A (en) * | 2019-12-31 | 2020-05-19 | 松立控股集团股份有限公司 | License plate recognition method, system, device and storage medium |
| CN112528856A (en) * | 2020-12-10 | 2021-03-19 | 天津大学 | Repeated video detection method based on characteristic frame |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520841A (en) * | 2009-03-10 | 2009-09-02 | 北京航空航天大学 | Real-time and anti-interference method for positioning license plate in high-definition TV video |
| CN103544489A (en) * | 2013-11-12 | 2014-01-29 | 公安部第三研究所 | Device and method for locating automobile logo |
| CN103903018A (en) * | 2014-04-02 | 2014-07-02 | 浙江师范大学 | Method and system for positioning license plate in complex scene |
| CN105279512A (en) * | 2015-10-22 | 2016-01-27 | 东方网力科技股份有限公司 | Tilt vehicle license plate recognition method and device |
| CN105335743A (en) * | 2015-10-28 | 2016-02-17 | 重庆邮电大学 | Vehicle license plate recognition method |
| CN106056101A (en) * | 2016-06-29 | 2016-10-26 | 哈尔滨理工大学 | Non-maximum suppression method for face detection |
-
2016
- 2016-11-07 CN CN201610972305.8A patent/CN106503748A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520841A (en) * | 2009-03-10 | 2009-09-02 | 北京航空航天大学 | Real-time and anti-interference method for positioning license plate in high-definition TV video |
| CN103544489A (en) * | 2013-11-12 | 2014-01-29 | 公安部第三研究所 | Device and method for locating automobile logo |
| CN103903018A (en) * | 2014-04-02 | 2014-07-02 | 浙江师范大学 | Method and system for positioning license plate in complex scene |
| CN105279512A (en) * | 2015-10-22 | 2016-01-27 | 东方网力科技股份有限公司 | Tilt vehicle license plate recognition method and device |
| CN105335743A (en) * | 2015-10-28 | 2016-02-17 | 重庆邮电大学 | Vehicle license plate recognition method |
| CN106056101A (en) * | 2016-06-29 | 2016-10-26 | 哈尔滨理工大学 | Non-maximum suppression method for face detection |
Non-Patent Citations (7)
| Title |
|---|
| 张振强等: ""一种新的复杂背景下快速车牌定位方法"", 《科技通报》 * |
| 张鹏等: ""采用稀疏SIFT特征的车型识别方法"", 《西安交通大学学报》 * |
| 杨建 等: ""监控视频中正面人脸快速判别方法"", 《计算机工程与应用》 * |
| 杨涛等: ""一种基于HSV颜色空间和SIFT特征的车牌提取算法"", 《计算机应用研究》 * |
| 王少伟等: ""一种改进的RGB hough车牌校正定位算法"", 《光学技术》 * |
| 赵阳等: ""基于车脸特征的车型识别技术及其在公安领域的应用"", 《警察技术》 * |
| 陈金辉等: ""行人检测中非极大值抑制算法的改进"", 《华东理工大学学报（自然科学版）》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092876A (en) * | 2017-04-12 | 2017-08-25 | 湖南源信光电科技股份有限公司 | The low-light (level) model recognizing method combined based on Retinex with S SIFT features |
| CN107085074B (en) * | 2017-04-19 | 2019-07-23 | 中国科学技术大学 | A method of classification monitoring motor-vehicle tail-gas |
| CN107085074A (en) * | 2017-04-19 | 2017-08-22 | 中国科学技术大学 | A kind of method for monitoring motor-vehicle tail-gas of classifying |
| CN107886539B (en) * | 2017-10-19 | 2021-05-14 | 昆明理工大学 | High-precision gear visual detection method in industrial scene |
| CN107886539A (en) * | 2017-10-19 | 2018-04-06 | 昆明理工大学 | High class gear visible detection method under a kind of industrial scene |
| CN108388919A (en) * | 2018-02-28 | 2018-08-10 | 大唐高鸿信息通信研究院（义乌）有限公司 | The identification of vehicle-mounted short haul connection net security feature and method for early warning |
| CN108388919B (en) * | 2018-02-28 | 2021-08-10 | 大唐高鸿信息通信(义乌)有限公司 | Vehicle-mounted short-distance communication network safety feature identification and early warning method |
| CN108537151A (en) * | 2018-03-27 | 2018-09-14 | 上海小蚁科技有限公司 | A kind of non-maxima suppression arithmetic unit and system |
| US11151777B2 (en) | 2018-03-27 | 2021-10-19 | Shanghai Xiaoyi Technology Co., Ltd. | Non-maximum suppression operation device and system |
| CN109840809A (en) * | 2019-03-12 | 2019-06-04 | 中国联合网络通信集团有限公司 | Advertisement placement method and device |
| CN111178357A (en) * | 2019-12-31 | 2020-05-19 | 松立控股集团股份有限公司 | License plate recognition method, system, device and storage medium |
| CN111178357B (en) * | 2019-12-31 | 2023-09-29 | 松立控股集团股份有限公司 | License plate recognition method, system, device and storage medium |
| CN112528856A (en) * | 2020-12-10 | 2021-03-19 | 天津大学 | Repeated video detection method based on characteristic frame |
| CN112528856B (en) * | 2020-12-10 | 2022-04-15 | 天津大学 | Repeated video detection method based on characteristic frame |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106503748A (en) | A kind of based on S SIFT features and the vehicle targets of SVM training aids | |
| CN105930791B (en) | The pavement marking recognition methods of multi-cam fusion based on DS evidence theory | |
| CN106529532A (en) | License plate identification system based on integral feature channels and gray projection | |
| CN107103317A (en) | Fuzzy license plate image recognition algorithm based on image co-registration and blind deconvolution | |
| CN103258432B (en) | Traffic accident automatic identification processing method and system based on videos | |
| CN102799901B (en) | Method for multi-angle face detection | |
| CN102609686B (en) | Pedestrian detection method | |
| CN102214291B (en) | Method for quickly and accurately detecting and tracking human face based on video sequence | |
| CN101630363B (en) | Rapid detection method of face in color image under complex background | |
| CN103761531B (en) | The sparse coding license plate character recognition method of Shape-based interpolation contour feature | |
| CN106529461A (en) | Vehicle model identifying algorithm based on integral characteristic channel and SVM training device | |
| CN104778453B (en) | A kind of night pedestrian detection method based on infrared pedestrian's brightness statistics feature | |
| CN102332092B (en) | Flame detection method based on video analysis | |
| CN105447503B (en) | Pedestrian detection method based on rarefaction representation LBP and HOG fusion | |
| CN106096602A (en) | A kind of Chinese licence plate recognition method based on convolutional neural networks | |
| CN102194108B (en) | Smile face expression recognition method based on clustering linear discriminant analysis of feature selection | |
| WO2022121039A1 (en) | Bankcard tilt correction-based detection method and apparatus, readable storage medium, and terminal | |
| CN107092876A (en) | The low-light (level) model recognizing method combined based on Retinex with S SIFT features | |
| CN106682586A (en) | Method for real-time lane line detection based on vision under complex lighting conditions | |
| CN102096823A (en) | Face detection method based on Gaussian model and minimum mean-square deviation | |
| CN103971126A (en) | Method and device for identifying traffic signs | |
| CN109255375A (en) | Panoramic picture method for checking object based on deep learning | |
| CN103473571A (en) | Human detection method | |
| CN106529592A (en) | License plate recognition method based on mixed feature and gray projection | |
| CN103605953A (en) | Vehicle interest target detection method based on sliding window search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170315 |
|
| RJ01 | Rejection of invention patent application after publication |