CN106503638A - For the image procossing of colour recognition, vehicle color identification method and system - Google Patents
For the image procossing of colour recognition, vehicle color identification method and system Download PDFInfo
- Publication number
- CN106503638A CN106503638A CN201610893880.9A CN201610893880A CN106503638A CN 106503638 A CN106503638 A CN 106503638A CN 201610893880 A CN201610893880 A CN 201610893880A CN 106503638 A CN106503638 A CN 106503638A
- Authority
- CN
- China
- Prior art keywords
- pixel
- target
- color
- vehicle
- destination object
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of image procossing for colour recognition, vehicle color identification method and system, the above-mentioned image processing method for colour recognition, comprise the steps：Multiple videos are chosen from video acquisition system, some two field pictures are chosen respectively from each video, training set is constructed, according to the target context code book that the training set recognizes video acquisition system；According to the corresponding foreground target pixel of destination object in the target context code book recognition target image；The corresponding pixel of destination object reflector segment in the foreground target pixel is rejected, the corresponding subject pixel point of main part is obtained；According to the colour recognition that the subject pixel point carries out destination object, the color of destination object is determined；Which effectively increases the accuracy of recognized color.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of image procossing for colour recognition, vehicle
Color identification method and system.
Background technology
At present, video recording system etc. extensively should in the video acquisition system that a certain fixing point obtains corresponding place video image
It is used in the live and work of people.Video in above-mentioned video acquisition system is obtained and respective handling, recognized target
The color of object, significant to security or related management work；For example, it is possible in urban road or community gate inhibition
Vehicle monitoring video is obtained in bayonet system in place, is carried out the colour recognition of wherein certain vehicle, and then is determined above-mentioned car
Identity, for work such as vehicle retrieval, traffic administrations.
During tradition is for the colour recognition scheme of target image in video, there is method using HSI color spaces and reference colour
Differ from formula to calculate color value, recycle BP neural network to carry out the color classification of the target images such as vehicle；There is method to employ
The method of color template coupling realizes the classification of target image color；There is method to complete target using H-S two-dimensional color histograms
The classification of color of image；Also the acutance in hsv color space and saturation degree component are separated by method, are then built with this
Grader completes color classification, accordingly to be recognized to target image color.But some targets in video acquisition system
Object, including needing the main part (such as the body portion of vehicle) for carrying out colour recognition, it is also possible to including disturbing colour recognition
Reflector segment (such as the vehicle window part on vehicle)；Above-mentioned colour recognition scheme is easily done by reflector segment in destination object
Disturb, make recognized color accuracy low.
Content of the invention
Be based on this, it is necessary to the low technical problem of recognized color accuracy is made for traditional scheme, there is provided Yi Zhongyong
Image procossing, vehicle color identification method and system in colour recognition.
A kind of image processing method for colour recognition, comprises the steps：
Multiple videos are chosen from video acquisition system, choose some two field pictures, construction training respectively from each video
Collection, according to the target context code book that the training set recognizes video acquisition system；
According to the corresponding foreground target pixel of destination object in the target context code book recognition target image；
The corresponding pixel of destination object reflector segment in the foreground target pixel is rejected, main part is obtained corresponding
Subject pixel point；
According to the colour recognition that the subject pixel point carries out destination object, the color of destination object is determined.
A kind of image processing system for colour recognition, including：
Module is chosen, and for multiple videos are chosen from video acquisition system, chooses some frames respectively from each video
Image, constructs training set, according to the target context code book that the training set recognizes video acquisition system；
Identification module, for according to the corresponding foreground target of destination object in the target context code book recognition target image
Pixel；
Module is rejected, for rejecting the corresponding pixel of destination object reflector segment in the foreground target pixel, is obtained
Arrive subject pixel point；
Determining module, for carrying out the colour recognition of destination object according to the subject pixel point, determines destination object
Color.
Above-mentioned image processing method and system for colour recognition, can be according to the multiple videos in video acquisition system
Determine target context code book, using the corresponding foreground target pixel of destination object in upper target context code book recognition target image
Point, then corresponding for the reflector segment for easily causing colour recognition interference in foreground target pixel pixel is rejected, then
Corresponding colour recognition is carried out to the subject pixel point of energy accurate characterization destination object color, to determine the color of destination object,
Effectively increase the accuracy of recognized color.
A kind of vehicle color identification method, comprises the steps：
Vehicle monitoring video is obtained from bayonet system；In vehicle monitoring video identification target vehicle, with target vehicle
Vehicle body is main part, and the vehicle window with target vehicle is as reflector segment；
According to the above-mentioned image processing method identification body color for destination object colour recognition, according to the car for being recognized
Body color determines the color of target vehicle.
A kind of vehicle color identifying system, including：
Acquisition module, for obtaining vehicle monitoring video from bayonet system；Target vehicle is chosen in vehicle monitoring video,
, as main part, the vehicle window with target vehicle is as reflector segment for vehicle body with target vehicle；
Vehicle color determining module, for according to the above-mentioned image processing system identification car for destination object colour recognition
Body color, according to the color that the body color for being recognized determines target vehicle.
Above-mentioned vehicle color identification method and system, can choose multiple automobile video frequencies from bayonet system and determine background mesh
Coding sheet, using the upper target context code book identification corresponding foreground target pixel of target vehicle, then by foreground target pixel
In easily cause the corresponding pixel of vehicle window of colour recognition interference and rejected, then to can accurate characterization vehicle color vehicle body
Pixel carries out corresponding colour recognition, makes recognized vehicle color that there is higher accuracy.
Description of the drawings
Image processing method flow charts for colour recognition of the Fig. 1 for one embodiment；
Fig. 2 is the position relationship schematic diagram between the main shaft and vehicle of one embodiment；
Fig. 3 is the position relationship schematic diagram between the main shaft and vehicle of one embodiment；
A-B plane subregion schematic diagrames of the Fig. 4 for one embodiment；
Vehicle color identification process schematic diagrames of the Fig. 5 for one embodiment；
Image processing system schematic diagrames for colour recognition of the Fig. 6 for one embodiment.
Specific embodiment
Below in conjunction with the accompanying drawings to the image procossing for colour recognition of the present invention, vehicle color identification method and system
Specific embodiment is described in detail.
With reference to Fig. 1, Fig. 1 show the image processing method flow chart for colour recognition of one embodiment, including such as
Lower step：
S10, chooses multiple videos from video acquisition system, chooses some two field pictures respectively from each video, construction
Training set, according to the target context code book that the training set recognizes video acquisition system；
Above-mentioned video acquisition system can include that bayonet system at the vehicle outpost of the tax office etc. gathers the place in fixed-site and regards
The system of frequency image.Multiple videos can be randomly selected from video acquisition system, then are chosen from each video above-mentioned some
Two field picture, makes selected image and includes the spy of corresponding fixed-site as far as possible according to the training set of above-mentioned image configuration
Levy.Above-mentioned target context code book includes removing in destination object (such as through the vehicle at the corresponding outpost of the tax office) to be identified and fixed-site
Place background outside above-mentioned destination object.Above-mentioned target context code book has its corresponding code word CBComposition, its according to respectively from
Some two field pictures in each video are determined, with higher integrality and accuracy.
S20, according to the corresponding foreground target pixel of destination object in the target context code book recognition target image；
Target image includes destination object and place background, and foreground target pixel is the corresponding picture of above-mentioned destination object
Vegetarian refreshments.
In one embodiment, described according to destination object in the target context code book recognition target image corresponding before
The step of scape target pixel points, can include：
The all code words in the target context code book are recognized, non-background picture is searched in the corresponding pixel of target image
Vegetarian refreshments, obtains foreground target pixel.
The corresponding pixel of above-mentioned target image includes that background pixel point and non-background pixel point, background pixel point refer to place
The corresponding pixel of background, non-background pixel point refer to the corresponding pixel of destination object, as foreground target pixel.
As one embodiment, above-mentioned can be with the step of search non-background pixel point in the corresponding pixel of target image
Including：
The rgb value of any one pixel p in target image is obtained, according to lightness I that the rgb value calculates pixel p, its
In,R component values of the R for vegetarian refreshments p, G component values of the G for vegetarian refreshments p, B component values of the B for vegetarian refreshments p；
The pixel for being unsatisfactory for following conditions A or condition B in target image is judged to non-background pixel point：
Condition A：Color_distance(p,CB)≤ε, wherein, p represents the pixel in target image, CBTarget context
Code word in code book, Color_distance () is represented and is asked for color distance, and ε represents default color threshold；Above-mentioned ε is usual
A less threshold value is set to, such as 0.3 or 0.5 is equivalent；
Condition B： The corresponding minimum lightness of pixel in target image is represented,
Represent the corresponding maximum lightness of pixel in target image.The definition of lightness function brightness is：Wherein,min{}
Expression takes minimum of a value, m1For default first lightness parameter, m2For default second lightness parameter；Above-mentioned first lightness parameter m1
0.5 is could be arranged to, the second lightness parameter m2Could be arranged to 1.5.
All code words have been checked in traversal, will target context code book substitute into condition A and condition B and carry out non-background pixel
After the identification of point, if certain pixel is mismatched with all code words, can determine whether that the pixel is foreground target pixel.
After successively finish all pixels point detection of pending image, foreground target pixel is just can determine.
S30, rejects the corresponding pixel of destination object reflector segment in the foreground target pixel, obtains main part
Corresponding subject pixel point；
Above-mentioned foreground target pixel can include the corresponding subject pixel point of main part, that is, need to carry out colour recognition
Pixel, also include due to the corresponding pixel of reflector segment of reflective interference respective color identification, vehicle window in such as vehicle
Corresponding pixel.
In one embodiment, above-mentioned destination object (such as vehicle) is zhou duicheng tuxing；The light-reflecting portion of the destination object
Divide and be symmetrically distributed in above main part；Under normal circumstances, it is reflector segment above main shaft, is main part below main shaft；
Described reject the corresponding pixel of destination object reflector segment in the foreground target pixel, obtain subject pixel
The step of point, includes：
The main shaft of the corresponding destination object of identification foreground target pixel；
Foreground target pixel below the main shaft is defined as subject pixel point.
Such as the reflector segment such as vehicle window of vehicle generally occurs the high optical phenomenon of mirror-reflection, even by color correction
The interference that these high optical phenomenons bring cannot be also restored to collection image, if carried out accordingly according to such noisy image
Colour recognition, can the recognized result of high degree impact.Corresponding for destination object reflector segment pixel is rejected, can
To ensure the accuracy of recognized color.
It is as one embodiment, above-mentioned after judging the distribution of the foreground target pixel as being horizontal distribution,
The step of center of gravity point of identification foreground target pixel corresponding destination object, can include：
Arranging central moment is：Wherein,
TBRepresent foreground target pixel, (x, y) ∈ TBRepresent that pixel x and y belong to foreground target pixel, p and q are represented down respectively
Mark variable, can distinguish round numbers, and p and q takes different integers, is respectively provided with corresponding λp,q, such as, p=2, q=0, then this
When λp,qFor λ2,0；
According to towards angle computing formulaCalculate towards angleWhereinAbove-mentionedIt is a temporary variable symbol, makes in the parametric procedure that minimizes
Intermediate variable；It is right that above-mentioned steps can pass throughSeek first derivative to calculate towards angle
If describedValue be located at intervalOn, then judge the distribution of the foreground target pixel as
It is horizontal distribution.
In the present embodiment, in order to accurately detect the residing region of the reflector segments such as vehicle window, the prospect mesh is judged first
The distribution of mark pixel is for being horizontal distribution, and then accurately detects the direction of the main shaft of destination object.By taking vehicle as an example,
Above-mentioned main shaft is upper and lower with the vehicle window of vehicle along parallel, main shaft being designated as towards angle in whole imageIf destination object is
Vehicle,It is located at intervalOn, then the position relationship schematic diagram between above-mentioned main shaft and vehicle can with as shown in Fig. 2
Otherwise can be determined that vehicle target is vertical distribution in the target image, the position relationship schematic diagram between main shaft and vehicle can
With as shown in Figure 3.In fig. 2, vehicle is horizontal distribution, and vehicle divide into two parts by cut-off rule (main shaft), positioned at point
The part of secant lower end removes the main part after vehicle window；In figure 3, vehicle is vertical distribution, and vehicle is divided into by cut-off rule
Two parts, remove the vehicle target (main part) after vehicle window positioned at the part of cut-off rule lower end.
S40, according to the colour recognition that the subject pixel point carries out destination object, determines the color of destination object.
In one embodiment, the process bag of the above-mentioned colour recognition for carrying out destination object according to the subject pixel point
Include：
The rgb value of the subject pixel point is carried out standardization calculating, R component value I after calculating of standardizing is obtainedR, G point
Value IGWith B component value IB, wherein, IR=r/ (r+g+b), IG=g/ (r+g+b), IB=b/ (r+g+b), r are subject pixel point
R component value, g for subject pixel point G component values, b for subject pixel point B component value；
According to the IR、IGAnd IBJudge whether subject pixel point belongs to secondary colour；
If the subject pixel point belongs to secondary colour, subject pixel point input secondary colour grader is carried out color
Identification；
If the subject pixel point is not belonging to secondary colour, subject pixel point input monochromatic color grader is carried out face
Color is recognized.
The such as destination object such as vehicle generally includes secondary colour and monochromatic color, first can carry out pixel tone mirror to destination object
Not, judging whether subject pixel point belongs to secondary colour, then subject pixel point is input into corresponding color sorter (such as secondary colour
Grader or monochromatic color grader) identification of respective color is carried out, to carry out destination object colour recognition.
As one embodiment, during can change subject pixel point to LAB color spaces, using A-B planes, using pole
Coordinate pair A-B plane is sampled, in polar diameter direction, with 10 as quantization unit；In polar angle direction, with 90 degree as quantization unit；
A-B planes are divided into 28 sub-regions altogether, as shown in figure 4, a Color Statistical table is correspond to per sub-regions, use mLAB
(k)=# q | q ∈ binkRepresent k-th Color Statistical table in pixel quantity, can build one 28 dimension vectorial mLAB=
[mLAB(1),mLAB(2),...mLAB(28)] ', the vector is the LAB parts of blend color characteristic vector.
For RGB component, 6 dimensional vectors of a record component pixel quantity relativeness will be built, the 6 of the vector
That individual component is recorded respectively is B>G、G>R、G>B、G>R、R>B、R>The quantity of the corresponding pixels of G, uses symbol mRGB=[mRGB
(1),mRGB(2),...mRGB(6)] ' represent.The two vectors are merged characteristic vector m for constituting one 34 dimensionDG, it is input to SVM
Classified in grader, completed the structure of the grader of monochromatic color.
As one embodiment, 3 color components in RGB color are quantified as 8 grades respectively, thus a total of 24
Individual color grade, the characteristic vector that the quantity of this corresponding pixel of 24 grades is respectively constituted one 24 dimension, is designated as mquan,
Additionally, B in 3 passages of record>G、G>R、G>B、G>R、R>B、R>The quantity of the corresponding pixels of G, is designated as：mRGB.Both close
And, obtain characteristic vector m of one 30 dimensionDC.This characteristic vector is input in SVM classifier and is classified, complete to mix
The structure of the grader of color.
As one embodiment, above-mentioned according to the IR、IGAnd IBJudge whether subject pixel point belongs to the process of secondary colour
Can include：
Calculate the Euclidean distance d between subject pixel point and secondary colour central shaft vectorp, wherein, dp=(IR-0.333)2+
(IG-0.333)2+(IB-0.333)2；
Calculate the Euclidean distance d between subject pixel point and hybrid mean value vectormean, wherein, dmean=(r-ur)2+(g-
ug)2+(b-ub)2, urRepresent the average of the R component value of all subject pixel points, ugRepresent the G component values of all subject pixel points
Average, ubRepresent the average of the B component value of all subject pixel points；
D is calculated respectivelymeanMean μmean, dpMean μd, dmeanVariances sigmameanAnd dpVariances sigmad；
According toCalculate subject pixel point and belong to mixing
The probability P (gray | p) of color；
P (gray | p) is judged to the subject pixel point for belonging to secondary colour more than the subject pixel point of probability threshold value.
P (gray | p) is more than probability threshold value, i.e. and P (gray | p) ＞ ThpIn, above-mentioned probability threshold value ThpCould be arranged to 80%
Equivalence, when two training sample databases are built, by averagely obtained to all of mixing color pixel.Due to vehicle
Etc. complexity and the interference of ambient noise of target object in structure, destination object to be detected had generally both contained secondary colour picture
Element, also contains single-frequency color pixel.Thus can be judged by following rule：Mixing color pixel accounting is then recognized more than 80%
Secondary colour is belonging to for the vehicle target color；Otherwise, then it is assumed that the vehicle target color is belonging to monochromatic color.
As one embodiment, by taking vehicle as an example, in conjunction with the vehicle color gathered in bayonet system, can be by vehicle color
Class population is divided into two big class：Secondary colour, monochromatic color, the process are referred to as pixel tone and differentiate.Then big according to two respectively
Class builds training sample database, i.e.,：Secondary colour training sample database, monochromatic color training sample.Sample in two training sample databases is not
There is any common factor.First setting vehicle color classification includes：Black, white, silver color, navy blue, bottle green, yellow, redness, green
Color, blueness.
The above-mentioned process for carrying out colour recognition to vehicle can be with as shown in figure 5, carry out picture to vehicle target pixel first
Plain color is adjusted and is differentiated, to judge whether the subject pixel point of vehicle belongs to secondary colour, if belonging to secondary colour, can be by above-mentioned vehicle
Subject pixel point carry out secondary colour feature extraction, then be input into secondary colour grader (SVM) and carry out colour recognition, above-mentioned secondary colour
Including black, white, silver color, navy blue, bottle green；If belonging to monochromatic color, the subject pixel of above-mentioned vehicle can be clicked through
Row secondary colour feature extraction, then be input into monochromatic color grader (SVM) and carry out colour recognition, above-mentioned monochromatic color include yellow, redness,
Green, blue.
Provided by the present invention for the image processing method of colour recognition, can be regarded according to multiple in video acquisition system
Frequency determines target context code book, using the corresponding foreground target pixel of destination object in upper target context code book recognition target image
Point, then corresponding for the reflector segment for easily causing colour recognition interference in foreground target pixel pixel is rejected, then
Corresponding colour recognition is carried out to the subject pixel point of energy accurate characterization destination object color, to determine the color of destination object,
Effectively increase the accuracy of recognized color.
In one embodiment, the Cheng Qian that crosses of the above-mentioned colour recognition for carrying out destination object according to the subject pixel point goes back
Can include：
It is divided into M pixel subregion, single region to be designated as T the subject pixel pointi', select in M pixel region
A two field picture R is selected as construction frame, it is R to obtain the corresponding image regions of construction frame Ri；
H is seti(p)=αiRi(p)+βi, wherein, p is Ti' in any one pixel, WithIt is T respectivelyi' pixel average and RiPixel average,WithIt is T respectivelyi' pixel variance and RiPixel
Variance, hiP () is the corresponding color mapping function of i-th subregion；
According toCalculate αiAnd βiValue：|Ti' | expression is picture
The pixel capacity in sub-prime region；
H is setiP () is for weight w of pixel pi(p),Operator | | a-b | | tables
Show and seek Euclidean distance between a and b；
According toCalculate normalized weight
According toAnd hiP () determines weighting global color correction function H (p), wherein
The subject pixel point is substituted into and is corrected in weighting global color correction function H (p).
In the present embodiment, for the color acquisition difference for overcoming the factors such as different illumination conditions, shooting angle to cause, can be with
Color mapping and correction are carried out to subject pixel point.To reject the subject pixel point of reflector segment as input image data.Due to
Real scene image construction is extremely complex, therefore, it is necessary to the mapping of color is implemented on subregion ground.The first step, it is possible to use
Input picture T ' (subject pixel point) is divided into M region, single region to be designated as by the Open-Source Tools that Felzenszwalb is provided
Ti′；Additionally, selecting the preferable two field picture R of image quality as construction frame at random, its foreground target pixel is removed, frame R pair is constructed
The image region that answers is Ri.Meanwhile, in mapping function construction process, it is necessary to keep the invariable of respective pixel coordinate.
Second step, completes to map in LAB color spaces in the following way：
hi(p)=αiRi(p)+βi, wherein, p is subregion Ti' in any one pixel, WithIt is input image data T respectivelyi' average and construction frame image data RiAverage,WithIt is input figure respectively
As data Ti' variance and construction frame image data RiVariance, hiP () is the corresponding color mapping function of i-th subregion；
Come the α that gets parms by way of minimizing energy functioniAnd βiValue,
To above formula right-hand member expression respectively to parameter alphaiAnd βiSeek first derivative and solve, can obtain：
Wherein, | Ti' | expression be subregion pixel capacity.
3rd step, constructs global color correction function H (p) by the mapping function of subregion；Define sub-district domain mapping letter
Number hi() is for weight w of pixel pi(p):
Wherein, operator | | a-b | | represents the Euclidean distance for asking a and b.For M
, just there is M local mapping function h in regioni()；In order to ensure the standardization of different colours weight computing result, it is necessary to weight
Function standardized (normalization) process：
Can obtain weighting global color correction function：
So as to the subject pixel point is substituted in weighting global color correction function H (p)
It is corrected.
With reference to shown in Fig. 6, image processing system schematic diagrames for colour recognition of the Fig. 6 for one embodiment, bag
Include：
Module 10 is chosen, and for multiple videos are chosen from video acquisition system, is chosen respectively some from each video
Two field picture, constructs training set, according to the target context code book that the training set recognizes video acquisition system；
Identification module 20, for according to the corresponding prospect mesh of destination object in the target context code book recognition target image
Mark pixel；
Module 30 is rejected, for rejecting the corresponding pixel of destination object reflector segment in the foreground target pixel,
Obtain subject pixel point；
Determining module 40, for carrying out the colour recognition of destination object according to the subject pixel point, determines destination object
Color.
Image processing system provided by the present invention for colour recognition and the figure provided by the present invention for colour recognition
As processing method correspond, the image processing method for colour recognition embodiment illustrate technical characteristic and its
Beneficial effect, suitable for the embodiment for the image processing system of colour recognition, hereby give notice that.
A kind of vehicle color identification method, comprises the steps：
Vehicle monitoring video is obtained from bayonet system；In vehicle monitoring video identification target vehicle, with target vehicle
Vehicle body is main part, and the vehicle window with target vehicle is as reflector segment；
According to the above-mentioned image processing method identification body color for destination object colour recognition, according to the car for being recognized
Body color determines the color of target vehicle.
Above-mentioned bayonet system includes the bayonet system in the place such as urban road or community gate inhibition.Above-mentioned target vehicle is upper
The destination object in the image processing method for destination object colour recognition is stated, is utilized above-mentioned for destination object colour recognition
Image processing method carry out the colour recognition of automobile body part (main part), and then determine corresponding vehicle color, can
To improve the accuracy of recognized vehicle color.
The vehicle color identification method that the present invention is provided, can choose multiple automobile video frequencies from bayonet system and determine background
Target code book, using the upper target context code book identification corresponding foreground target pixel of target vehicle, then by foreground target pixel
The corresponding pixel of vehicle window for easily causing colour recognition interference in point is rejected, then to can accurate characterization vehicle color car
Body pixel carries out corresponding colour recognition, makes recognized vehicle color that there is higher accuracy.
A kind of vehicle color identifying system, including：
Acquisition module, for obtaining vehicle monitoring video from bayonet system；Target vehicle is chosen in vehicle monitoring video,
, as main part, the vehicle window with target vehicle is as reflector segment for vehicle body with target vehicle；
Vehicle color determining module, for according to the above-mentioned image processing system identification car for destination object colour recognition
Body color, according to the color that the body color for being recognized determines target vehicle.
The vehicle color identifying system that the present invention is provided is corresponded with the vehicle color identification method that the present invention is provided,
The technical characteristic and its advantage that the embodiment of the vehicle color identification method is illustrated is applied to vehicle color identification system
In the embodiment of system, hereby give notice that.
Each technical characteristic of embodiment described above arbitrarily can be combined, for making description succinct, not to above-mentioned reality
Apply all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, the scope of this specification record is all considered to be.
Embodiment described above only expresses the several embodiments of the present invention, and its description is more concrete and detailed, but simultaneously
Therefore can not be construed as limiting the scope of the patent.It should be pointed out that for one of ordinary skill in the art
Say, without departing from the inventive concept of the premise, some deformations and improvement can also be made, these belong to the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be defined by claims.
Claims (10)
1. a kind of image processing method for colour recognition, it is characterised in that comprise the steps：
Multiple videos are chosen from video acquisition system, some two field pictures are chosen respectively from each video, construct training set, root
According to the target context code book that the training set recognizes video acquisition system；
According to the corresponding foreground target pixel of destination object in the target context code book recognition target image；
The corresponding pixel of destination object reflector segment in the foreground target pixel is rejected, the corresponding master of main part is obtained
Body image vegetarian refreshments；
According to the colour recognition that the subject pixel point carries out destination object, the color of destination object is determined.
2. the image processing method for colour recognition according to claim 1, it is characterised in that described according to the back of the body
Include the step of destination object corresponding foreground target pixel in scape target code book recognition target image：
The all code words in the target context code book are recognized, in the corresponding pixel of target image, searches non-background pixel
Point, obtains foreground target pixel.
3. the image processing method for colour recognition according to claim 2, it is characterised in that described in target image
The step of searching non-background pixel point in corresponding pixel includes：
Obtain target image in any one pixel p rgb value, according to the rgb value calculate pixel p lightness I, wherein,R component values of the R for vegetarian refreshments p, G component values of the G for vegetarian refreshments p, B component values of the B for vegetarian refreshments p；
The pixel for being unsatisfactory for following conditions A or condition B in target image is judged to non-background pixel point：
Condition A：Color_distance(p,CB)≤ε, wherein, p represents the pixel in target image, CBTarget context code book
In code word, Color_distance () represent ask for color distance, ε represents default color threshold；
Condition B： The corresponding minimum lightness of pixel in target image is represented,Represent
The corresponding maximum lightness of pixel in target image.The definition of lightness function brightness is：Wherein,min{}
Expression takes minimum of a value, m1For default first lightness parameter, m2For default second lightness parameter.
4. the image processing method for colour recognition according to claim 1, it is characterised in that the destination object is
Zhou duicheng tuxing；The reflector segment of the destination object is symmetrically distributed in above main part；
Described reject the corresponding pixel of destination object reflector segment in the foreground target pixel, obtain subject pixel point
Step includes：
The main shaft of the corresponding destination object of identification foreground target pixel；
Foreground target pixel below the main shaft is defined as subject pixel point.
5. the image processing method for colour recognition according to claim 4, it is characterised in that described judge described
The distribution of foreground target pixel is for after being horizontal distribution, recognizing the gravity of the corresponding destination object of foreground target pixel
The step of central point, includes：
Arranging central moment is：Wherein,TB
Represent foreground target pixel, (x, y) ∈ TBRepresent that pixel x and y belong to foreground target pixel；
According to towards angle computing formulaCalculate towards angle
Wherein
If describedValue be located at intervalOn, then judge the distribution of the foreground target pixel as being water
Divide cloth equally.
6. the image processing method for colour recognition according to claim 1, it is characterised in that described according to the master
The process of the colour recognition that volumetric pixel point carries out destination object includes：
The rgb value of the subject pixel point is carried out standardization calculating, R component value I after calculating of standardizing is obtainedR, G component values
IGWith B component value IB, wherein, IR=r/ (r+g+b), IG=g/ (r+g+b), IB=b/ (r+g+b), r R minute for subject pixel point
Value, G component values of the g for subject pixel point, B component values of the b for subject pixel point；
According to the IR、IGAnd IBJudge whether subject pixel point belongs to secondary colour；
If the subject pixel point belongs to secondary colour, subject pixel point input secondary colour grader is carried out color knowledge
Not；
If the subject pixel point is not belonging to secondary colour, subject pixel point input monochromatic color grader is carried out color knowledge
Not.
7. the image processing method for colour recognition according to claim 6, it is characterised in that described according to the IR、
IGAnd IBJudge whether subject pixel point belongs to the process of secondary colour and include：
Calculate the Euclidean distance d between subject pixel point and secondary colour central shaft vectorp, wherein, dp=(IR-0.333)2+(IG-
0.333)2+(IB-0.333)2；
Calculate the Euclidean distance d between subject pixel point and hybrid mean value vectormean, wherein, dmean=(r-ur)2+(g-ug)2+
(b-ub)2, urRepresent the average of the R component value of all subject pixel points, ugRepresent all subject pixel points G component values equal
Value, ubRepresent the average of the B component value of all subject pixel points；
D is calculated respectivelymeanMean μmean, dpMean μd, dmeanVariances sigmameanAnd dpVariances sigmad；
According toCalculate subject pixel point and belong to secondary colour
Probability P (gray | p)；
P (gray | p) is judged to the subject pixel point for belonging to secondary colour more than the subject pixel point of probability threshold value.
8. a kind of image processing system for colour recognition, it is characterised in that include：
Module is chosen, and for multiple videos are chosen from video acquisition system, chooses some two field pictures respectively from each video,
Construction training set, according to the target context code book that the training set recognizes video acquisition system；
Identification module, for according to the corresponding foreground target pixel of destination object in the target context code book recognition target image
Point；
Module is rejected, for rejecting the corresponding pixel of destination object reflector segment in the foreground target pixel, is led
Body image vegetarian refreshments；
Determining module, for carrying out the colour recognition of destination object according to the subject pixel point, determines the color of destination object.
9. a kind of vehicle color identification method, it is characterised in that comprise the steps：
Vehicle monitoring video is obtained from bayonet system；In vehicle monitoring video identification target vehicle, with the vehicle body of target vehicle
For main part, the vehicle window with target vehicle is as reflector segment；
The image processing method identification vehicle body face for destination object colour recognition according to any one of claim 1 to 7
Color, according to the color that the body color for being recognized determines target vehicle.
10. a kind of vehicle color identifying system, it is characterised in that include：
Acquisition module, for obtaining vehicle monitoring video from bayonet system；Target vehicle is chosen in vehicle monitoring video, with mesh
The vehicle body of mark vehicle is main part, and the vehicle window with target vehicle is as reflector segment；
Vehicle color determining module, for the image procossing system for destination object colour recognition according to claim 8
System identification body color, according to the color that the body color for being recognized determines target vehicle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610893880.9A CN106503638B (en) | 2016-10-13 | 2016-10-13 | Image procossing, vehicle color identification method and system for color identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610893880.9A CN106503638B (en) | 2016-10-13 | 2016-10-13 | Image procossing, vehicle color identification method and system for color identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106503638A true CN106503638A (en) | 2017-03-15 |
| CN106503638B CN106503638B (en) | 2019-09-13 |
Family
ID=58293946
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610893880.9A Active CN106503638B (en) | 2016-10-13 | 2016-10-13 | Image procossing, vehicle color identification method and system for color identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106503638B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107154032A (en) * | 2017-04-20 | 2017-09-12 | 腾讯科技（深圳）有限公司 | A kind of image processing method and device |
| CN110519489A (en) * | 2019-06-20 | 2019-11-29 | 杭州海康威视数字技术股份有限公司 | Image-pickup method and device |
| CN110751053A (en) * | 2019-09-26 | 2020-02-04 | 高新兴科技集团股份有限公司 | Vehicle color identification method, device, equipment and storage medium |
| CN111340896A (en) * | 2020-02-21 | 2020-06-26 | 北京迈格威科技有限公司 | Object color identification method and device, computer equipment and storage medium |
| CN111626188A (en) * | 2020-05-26 | 2020-09-04 | 西南大学 | Indoor uncontrollable open fire monitoring method and system |
| CN112016433A (en) * | 2020-08-24 | 2020-12-01 | 高新兴科技集团股份有限公司 | Vehicle color identification method based on deep neural network |
| CN113570597A (en) * | 2021-09-01 | 2021-10-29 | 南通中煌工具有限公司 | Method and system for judging dirt degree of mud-headed vehicle carriage based on artificial intelligence |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6088482A (en) * | 1998-10-22 | 2000-07-11 | Symbol Technologies, Inc. | Techniques for reading two dimensional code, including maxicode |
| CN101436252A (en) * | 2008-12-22 | 2009-05-20 | 北京中星微电子有限公司 | Method and system for recognizing vehicle body color in vehicle video image |
| CN102034080A (en) * | 2009-09-24 | 2011-04-27 | 北京汉王智通科技有限公司 | Vehicle color identification method and device |
| CN103530366A (en) * | 2013-10-12 | 2014-01-22 | 湖北微模式科技发展有限公司 | Vehicle searching method and system based on user-defined features |
| CN105005766A (en) * | 2015-07-01 | 2015-10-28 | 深圳市迈科龙电子有限公司 | Vehicle body color identification method |
-
2016
- 2016-10-13 CN CN201610893880.9A patent/CN106503638B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6088482A (en) * | 1998-10-22 | 2000-07-11 | Symbol Technologies, Inc. | Techniques for reading two dimensional code, including maxicode |
| CN101436252A (en) * | 2008-12-22 | 2009-05-20 | 北京中星微电子有限公司 | Method and system for recognizing vehicle body color in vehicle video image |
| CN102034080A (en) * | 2009-09-24 | 2011-04-27 | 北京汉王智通科技有限公司 | Vehicle color identification method and device |
| CN103530366A (en) * | 2013-10-12 | 2014-01-22 | 湖北微模式科技发展有限公司 | Vehicle searching method and system based on user-defined features |
| CN105005766A (en) * | 2015-07-01 | 2015-10-28 | 深圳市迈科龙电子有限公司 | Vehicle body color identification method |
Non-Patent Citations (2)
| Title |
|---|
| DAVID SETIABUDI ET AL.: "Digital Color Classification for Colorful Cross Stitch Threads Using RGB+Euclidean Distance and LAB+CIE94", 《2016 INTERNATIONAL CONFERENCE ON INFORMATION,COMMUNICATION TECHNOLOGY AND SYSTEM》 * |
| 赵占杰 等: "一种基于码本背景模型的运动目标检测方法", 《北京石油化工学院学报》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107154032A (en) * | 2017-04-20 | 2017-09-12 | 腾讯科技（深圳）有限公司 | A kind of image processing method and device |
| CN110519489A (en) * | 2019-06-20 | 2019-11-29 | 杭州海康威视数字技术股份有限公司 | Image-pickup method and device |
| CN110751053A (en) * | 2019-09-26 | 2020-02-04 | 高新兴科技集团股份有限公司 | Vehicle color identification method, device, equipment and storage medium |
| CN110751053B (en) * | 2019-09-26 | 2022-02-22 | 高新兴科技集团股份有限公司 | Vehicle color identification method, device, equipment and storage medium |
| CN111340896A (en) * | 2020-02-21 | 2020-06-26 | 北京迈格威科技有限公司 | Object color identification method and device, computer equipment and storage medium |
| CN111340896B (en) * | 2020-02-21 | 2023-10-27 | 北京迈格威科技有限公司 | Object color recognition method, device, computer equipment and storage medium |
| CN111626188A (en) * | 2020-05-26 | 2020-09-04 | 西南大学 | Indoor uncontrollable open fire monitoring method and system |
| CN111626188B (en) * | 2020-05-26 | 2022-05-06 | 西南大学 | Indoor uncontrollable open fire monitoring method and system |
| CN112016433A (en) * | 2020-08-24 | 2020-12-01 | 高新兴科技集团股份有限公司 | Vehicle color identification method based on deep neural network |
| CN113570597A (en) * | 2021-09-01 | 2021-10-29 | 南通中煌工具有限公司 | Method and system for judging dirt degree of mud-headed vehicle carriage based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106503638B (en) | 2019-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106503638A (en) | For the image procossing of colour recognition, vehicle color identification method and system | |
| CN107729801B (en) | Vehicle color recognition system based on multitask deep convolution neural network | |
| CN108596277B (en) | Vehicle identity recognition method and device and storage medium | |
| CN109271991B (en) | License plate detection method based on deep learning | |
| CN103971126B (en) | A kind of traffic sign recognition method and device | |
| CN110309781B (en) | House damage remote sensing identification method based on multi-scale spectrum texture self-adaptive fusion | |
| CN111738064B (en) | Haze concentration identification method for haze image | |
| CN102364496B (en) | Method and system for identifying automobile license plates automatically based on image analysis | |
| CN103824091B (en) | A kind of licence plate recognition method for intelligent transportation system | |
| CN110598560A (en) | Night monitoring and identifying method and system based on neural network enhancement | |
| CN107392968B (en) | The image significance detection method of Fusion of Color comparison diagram and Color-spatial distribution figure | |
| CN109409208A (en) | A kind of vehicle characteristics extraction and matching process based on video | |
| CN103035013A (en) | Accurate moving shadow detection method based on multi-feature fusion | |
| CN111292321A (en) | Method for identifying defect image of insulator of power transmission line | |
| CN108491498A (en) | A kind of bayonet image object searching method based on multiple features detection | |
| CN111027475A (en) | Real-time traffic signal lamp identification method based on vision | |
| CN112241692B (en) | Channel foreign matter intelligent detection and classification method based on aerial image super-pixel texture | |
| CN108319958A (en) | A kind of matched driving license of feature based fusion detects and recognition methods | |
| CN104657980A (en) | Improved multi-channel image partitioning algorithm based on Meanshift | |
| CN107590500A (en) | A kind of color recognizing for vehicle id method and device based on color projection classification | |
| CN105678318A (en) | Traffic label matching method and apparatus | |
| CN117011563B (en) | Road damage inspection cross-domain detection method and system based on semi-supervised federal learning | |
| CN109635679B (en) | Real-time target paper positioning and loop line identification method | |
| CN106845458A (en) | A kind of rapid transit label detection method of the learning machine that transfinited based on core | |
| CN109389132A (en) | A kind of mist Concentration Testing method for early warning and system based on image |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |