CN106096144B - A kind of automobile brand genetic analysis method based on preceding face moulding - Google Patents
A kind of automobile brand genetic analysis method based on preceding face moulding Download PDFInfo
- Publication number
- CN106096144B CN106096144B CN201610420677.XA CN201610420677A CN106096144B CN 106096144 B CN106096144 B CN 106096144B CN 201610420677 A CN201610420677 A CN 201610420677A CN 106096144 B CN106096144 B CN 106096144B
- Authority
- CN
- China
- Prior art keywords
- brand
- automobile
- front face
- picture
- automobile front
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- 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
Abstract
The invention belongs to vehicle configuration analysis fields, provide a kind of automobile brand genetic analysis method based on preceding face moulding.The automobile brand genetic analysis method includes creation automobile front face training library, carries out brand Classification and Identification and carry out brand genetic analysis;Brand genetic analysis further comprises form gene alternate analysis and brand shape-designing trend analysis between form gene analysis, brand in brand;The automobile brand genetic analysis method can analyze the core form gene in automobile brand, can analyze the similarity of form gene between brand, and can also be analyzed the differentiation of brand shape-designing and trend.
Description
Technical field
The invention belongs to vehicle configuration analysis fields.The present invention relates to a kind of automobile brand genes based on preceding face moulding point
Analysis method, and in particular to a method of brand genetic analysis is realized according to automobile brand classification method.
Background technique
Current Automobile Enterprises increasingly pay attention to brand building, and the familial design of the shape of automobile face has become automobile
One of core of brand building.By will include headlight, air-inlet grille and bumper sculpted zone as automobile before
Face, it can be found that the automobile front face of same brand different model has several similar shape-designings, referred to as brand moulding base
Cause.By the analysis to brand form gene, the design feature of face moulding before available brand can be according to sales volume factor analysis
Consumer favorite preceding face moulding.However the analysis of current brand form gene is mostly based on the spy extracted in automobile front face
Line is levied, and only single brand is analyzed, form gene can not be embodied in a manner of constructing complete automobile front face, also not
The similitude of form gene brand can be analyzed.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of automobile brand genetic analysis methods based on preceding face moulding, should
Method can analyze the core form gene in automobile brand, can analyze the similarity of form gene between brand, and may be used also
With to brand shape-designing differentiation and trend analyzed.
Technical solution of the present invention:
A kind of automobile brand genetic analysis method based on preceding face moulding, including creation automobile front face training library, progress product
Board Classification and Identification and progress brand genetic analysis;
(1) creation automobile front face training library: automobile front face training library includes the preceding face picture of a variety of domestic automobile brands,
According to brand to automobile front face training library picture classification, and establish brand label and time label;Count each vehicle of each brand
Year sales volume, calculate the sum of the sales volume of all times year that each vehicle counted, obtain the year sales volume point of each vehicle under same brand
Then cloth quantifies the year sales volume distribution of each vehicle and as shared by vehicle picture number in the automobile front face training each brand in library
Specific gravity parameter carries out the creation in automobile front face training library according to specific gravity parameter, completes the genes amplification of certain brands and vehicle;It is right
In every picture, image content is only comprising headlight, air-inlet grille, the sculpted zone of bumper, referred to as automobile front face figure
Piece.All automobile front face pictures, which will pass through background removal, license plate removal processing, and every picture, will be adjusted to unified high wide
Than.
(2) brand classification and identification: firstly, carrying out gray proces to all images in automobile front face training library and will own
Image Adjusting is to Suitable pixel sizes range 40~60 × 100~150 (high × wide).Then PCANet character representation and SVM are used
The automobile brand classification method that classifier combines carries out leaving-one method (leave-one- to the automobile front face training library that (1) obtains
Out it) tests, it may be assumed that extract an automobile front face picture as test sample, residual graph from automobile front face training library when test every time
Piece is as training sample;Training sample is carried out after classification based training obtains sorter model with PCANet/SVM method, to test
Sample is identified;According to this process, successively all automobile front face pictures in training library are tested.
(3) brand genetic analysis.The test result obtained to (2) is analyzed, and analysis content is as follows:
(a) form gene is analyzed in brand.The test result for obtaining certain brand in (2), before all automobiles in the brand
The decision content of face picture recognition result sorts in descending order, selects corresponding ten automobile front face picture samples of preceding ten values as product
The representative of board gene.Then " face before average vehicle " figure of the automobile brand is generated with " average face " technology calculated in photography
Picture." face before average vehicle " will be somebody's turn to do and be considered as the core gene of automobile brand.
(b) form gene alternate analysis between brand.The test result for obtaining all brands in (2), records and knows in certain brand
Not wrong picture sample and corresponding recognition result, while the picture sample that the brand is identified as in other brands is recorded, so
Afterwards by this two groups of result visualization processing, the recognition result cross chart of certain brand Yu other brands, referred to as form gene are established
Cross chart.Similarly, the form gene cross chart of all brands Yu other brands can be established.
(c) brand shape-designing trend analysis.The test result of certain brand in (2) is obtained, it is right respectively according to time label
The decision content of auto graph recognition result sorts in descending order in certain time, selects corresponding ten automobile front face pictures of preceding ten values
Representative of the sample as the time brand gene.Then the automobile brand is generated with " average face " technology in photography that calculates
This " face before average vehicle " is considered as the automobile brand core gene by the time by " face before average vehicle ".Similarly, it is somebody's turn to do
" face before average vehicle " in brand all times, it can be observed that their variation tendency.Further, in conjunction with (b) module, root
According to the dynamic evolution figure of form gene cross chart at any time between the available automobile brand of time label, to obtain certain automobile product
The development trend of face shape-designing before board.
Beneficial effects of the present invention:
(1) method of the automobile brand core gene analysis based on preceding face moulding is provided.
(2) method of gene similarity analysis between the automobile brand based on preceding face moulding is provided.
(3) method of the automobile brand gene trend analysis based on preceding face moulding is provided.
Detailed description of the invention
Fig. 1 is the techniqueflow chart of the invention patent.
Fig. 2 form gene cross chart between the brand of the invention patent.
In figure: it is 1. the correct results area of certain brand recognition, 2. -7. correctly tied for each self-identifying of other different brands
Fruit region, 8. pointed circle represent 1. brand recognition into other brands as a result, 9.-Pointed symbology other
Brand recognition at 1. brand result.
Specific embodiment
Below in conjunction with attached drawing and technical solution, a specific embodiment of the invention is further illustrated.
Automobile brand identification based on automobile front face, the specific steps of which are as follows:
(1) creation automobile front face training library: automobile front face training library includes the preceding face picture of a variety of domestic automobile brands,
Classified to automobile front face training library picture according to brand, and establishes brand label and time label;Comprising a variety of in each brand
Vehicle, each vehicle include the automobile of a variety of different colours and style again.Each vehicle of each brand is counted from 2000 so far
Year sales volume calculates the sum of all time year sales volumes of each vehicle, obtains the year sales volume distribution of each vehicle under same brand, then quantifies
The year sales volume of each vehicle is distributed and as specific gravity parameter shared by vehicle picture number in the automobile front face training each brand in library, presses
The creation that automobile front face training library is carried out according to specific gravity parameter, to complete the genes amplification of certain brands and vehicle.For every
Picture, image content are only comprising headlight, air-inlet grille, the sculpted zone of bumper, referred to as automobile front face picture.It is all
Automobile front face picture will pass through background removal, license plate removal processing, and every picture will be adjusted to unified depth-width ratio, such as high wide
Than for 1:2.5.
(2) brand classification and identification: firstly, carrying out gray proces to all images in automobile front face training library and will own
Image Adjusting is to Suitable pixel sizes range 40~60 × 100~150 (high × wide).Then PCANet character representation and SVM are used
The automobile brand classification method that classifier combines carries out leaving-one method (leave-one- to the automobile front face training library that (1) obtains
Out it) tests, it may be assumed that extract an automobile front face picture as test sample, residual graph from automobile front face training library when test every time
Piece is as training sample;Training sample is carried out after classification based training obtains sorter model with PCANet/SVM method, to test
Sample is identified;According to this process, successively all automobile front face pictures in training library are tested.
Further, which is made of three basic data processing modules: principal component analysis
(PCA), binary system Hash (Binary Hashing) and blocked histogram (Block-wise Histogram).Wherein lead twice
Constituent analysis constitutes the first two stage of PCANet feature extraction, is main characteristic extraction part;And binary system Hash and point
Block histogram is then the output stage of PCANet, by the output result of principal component analysis twice be mapped as final output feature to
Amount.
Further, in PCANet character representation method, the optimized parameter range of PCANet are as follows: moved in PCA processing
The Pixel Dimensions of window are 3 × 3 or 5 × 5 or 7 × 7 (high × wide);Filter quantity is 8;Moving Window in the histogram treatment stage
The Pixel Dimensions range of mouth is 7~10 × 7~10 (high × wide), and window coincidence factor range is 0.5~0.7.
Further, the training process of SVM classifier are as follows: by the feature vector of the PCANet training sample extracted and
Automobile front face training library brand label information input is to SVM classifier, by SVM classifier training sorter model.
(3) brand genetic analysis.The test result obtained to (2) is analyzed, and analysis content is as follows:
(a) form gene is analyzed in brand.The test result for obtaining certain brand in (2), before all automobiles in the brand
The decision content of face picture recognition result sorts in descending order, selects corresponding ten automobile front face picture samples of preceding ten values as product
The representative of board gene.Then " face before average vehicle " figure of the automobile brand is generated with " average face " technology calculated in photography
Picture." face before average vehicle " will be somebody's turn to do and be considered as the core gene of automobile brand.
(b) form gene alternate analysis between brand.The test result for obtaining all brands in (2), records and knows in certain brand
Not wrong picture sample and corresponding recognition result, while the picture sample that the brand is identified as in other brands is recorded, so
Afterwards by this two groups of result visualization processing, the recognition result cross chart of certain brand Yu other brands, referred to as form gene are established
Cross chart.Similarly, the form gene cross chart of all brands Yu other brands can be established.
As shown in Fig. 2,1. -7. respectively represent the correct results area of each self-identifying of different automobile brands, 1. brand
Identification error sample 8. indicated by circle, be distributed in 2. -7. in the recognition result 8. represented 1. in different brand;Together
Reason, other brand recognitions at picture sample 1. by be distributed in 1. brand 9.-It indicates.Can establish as a result, 1. brand with
The form gene cross chart of other brands.
(c) brand shape-designing trend analysis.The test result of certain brand in (2) is obtained, it is right respectively according to time label
The decision content of auto graph recognition result sorts in descending order in certain time, selects corresponding ten automobile front face pictures of preceding ten values
Representative of the sample as the time brand gene.Then the automobile brand is generated with " average face " technology in photography that calculates
This " face before average vehicle " is considered as the automobile brand core gene by the time by " face before average vehicle ".Similarly, it is somebody's turn to do
" face before average vehicle " in brand all times, it can be observed that their variation tendency.Further, in conjunction with step (b) mould
Block, according to the dynamic evolution figure of form gene cross chart at any time between the available automobile brand of time label, to obtain certain
The development trend of face shape-designing before automobile brand.
A specific embodiment of the invention is described in conjunction with attached drawing above, but these explanations cannot be understood to limit
The scope of the present invention, protection scope of the present invention are limited by appended claims, any in the claims in the present invention base
Change on plinth is all protection scope of the present invention.
Claims (1)
1. a kind of automobile brand genetic analysis method based on preceding face moulding, which is characterized in that the automobile brand genetic analysis side
Method includes creation automobile front face training library, carries out brand Classification and Identification and carry out brand genetic analysis;
(1) creation automobile front face training library: automobile front face training library includes the preceding face picture of a variety of domestic automobile brands, according to
Brand establishes brand label and time label to automobile front face training library picture classification;Count the year of each vehicle of each brand
Sales volume calculates the sum of the sales volume of all times year that each vehicle is counted, obtains the year sales volume distribution of each vehicle under same brand, so
Quantify the year sales volume distribution of each vehicle afterwards and as specific gravity shared by vehicle picture number in the automobile front face training each brand in library
Parameter carries out the creation in automobile front face training library according to specific gravity parameter, completes the genes amplification of certain brands and vehicle;For every
Picture, image content are only comprising headlight, air-inlet grille, the sculpted zone of bumper, referred to as automobile front face picture;Institute
There is automobile front face picture will be by background removal, license plate removal processing, and every picture will be adjusted to unified depth-width ratio;
(2) brand classification and identification: firstly, carrying out gray proces to all images in automobile front face training library and by all images
Being adjusted to Suitable pixel sizes range height × width is 40~60 × 100~150;Then PCANet character representation and svm classifier are used
The automobile brand classification method that device combines carries out leaving-one method test to the automobile front face training library that step (1) obtains, it may be assumed that surveys every time
An automobile front face picture is extracted as test sample from automobile front face training library when examination, and remaining picture is as training sample;With
After PCANet/SVM method obtains sorter model to training sample progress classification based training, test sample is identified;According to
This process successively tests all automobile front face pictures in training library;
(3) brand genetic analysis;The test result obtained to step (2) is analyzed, and analysis content is as follows:
(a) form gene is analyzed in brand;The test result of certain brand in obtaining step (2), before all automobiles in the brand
The decision content of face picture recognition result sorts in descending order, selects corresponding ten automobile front face picture samples of preceding ten values as product
The representative of board gene;Then " face before average vehicle " figure of the automobile brand is generated with " average face " technology calculated in photography
Picture;" face before average vehicle " will be somebody's turn to do and be considered as the core gene of automobile brand;
(b) form gene alternate analysis between brand;The test result of all brands, records and knows in certain brand in obtaining step (2)
Not wrong picture sample and corresponding recognition result, while the picture sample that the brand is identified as in other brands is recorded, so
Afterwards by this two groups of result visualization processing, the recognition result cross chart of certain brand Yu other brands, referred to as form gene are established
Cross chart;Similarly, the form gene cross chart of all brands Yu other brands is established;
(c) brand shape-designing trend analysis;The test result of certain brand, right respectively according to time label in obtaining step (2)
The decision content of auto graph recognition result sorts in descending order in certain time, selects corresponding ten automobile front face pictures of preceding ten values
Representative of the sample as the time brand gene;Then the automobile brand is generated with " average face " technology in photography that calculates
This " face before average vehicle " is considered as the face shaped core gene before the automobile brand in the time by " face before average vehicle ";Together
Reason, obtains " face before average vehicle " in the brand all times, observes their variation tendency;Further, in conjunction with step (b)
Module obtains the dynamic evolution figure of form gene cross chart at any time between automobile brand according to time label, to obtain certain vapour
The development trend of face shape-designing before vehicle brand.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610420677.XA CN106096144B (en) | 2016-06-13 | 2016-06-13 | A kind of automobile brand genetic analysis method based on preceding face moulding |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610420677.XA CN106096144B (en) | 2016-06-13 | 2016-06-13 | A kind of automobile brand genetic analysis method based on preceding face moulding |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106096144A CN106096144A (en) | 2016-11-09 |
CN106096144B true CN106096144B (en) | 2019-04-09 |
Family
ID=57846980
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610420677.XA Active CN106096144B (en) | 2016-06-13 | 2016-06-13 | A kind of automobile brand genetic analysis method based on preceding face moulding |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106096144B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106874418A (en) * | 2017-01-24 | 2017-06-20 | 成都容豪电子信息科技有限公司 | A kind of automobile model data set method for building up for serving deep learning |
CN112258472B (en) * | 2020-10-20 | 2022-09-20 | 大连理工大学 | Automatic scoring method for automobile exterior shape |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103295063A (en) * | 2013-05-17 | 2013-09-11 | 浙江大学 | Method for SUV (Sport Utility Vehicle) product family gene evolution based on genetic algorithm |
CN104156692A (en) * | 2014-07-07 | 2014-11-19 | 叶茂 | Automobile logo sample training and recognition method based on air-inlet grille positioning |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6826300B2 (en) * | 2001-05-31 | 2004-11-30 | George Mason University | Feature based classification |
-
2016
- 2016-06-13 CN CN201610420677.XA patent/CN106096144B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103295063A (en) * | 2013-05-17 | 2013-09-11 | 浙江大学 | Method for SUV (Sport Utility Vehicle) product family gene evolution based on genetic algorithm |
CN104156692A (en) * | 2014-07-07 | 2014-11-19 | 叶茂 | Automobile logo sample training and recognition method based on air-inlet grille positioning |
Also Published As
Publication number | Publication date |
---|---|
CN106096144A (en) | 2016-11-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103903005A (en) | License plate image identification system and method | |
CN111310850A (en) | License plate detection model construction method and system and license plate detection method and system | |
CN105718937B (en) | Multi-class object classification method and system | |
CN102663401B (en) | Image characteristic extracting and describing method | |
CN116188475B (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
CN110532950B (en) | Video feature extraction method and micro-expression identification method based on micro-expression video | |
CN104036284A (en) | Adaboost algorithm based multi-scale pedestrian detection method | |
WO2014087689A1 (en) | Image processing device, image processing system, and program | |
CN107704797B (en) | Real-time detection method, system and equipment based on pedestrians and vehicles in security video | |
CN108073940B (en) | Method for detecting 3D target example object in unstructured environment | |
CN106096144B (en) | A kind of automobile brand genetic analysis method based on preceding face moulding | |
CN114863464B (en) | Second-order identification method for PID drawing picture information | |
CN105975949A (en) | Visual-information-based automobile identification method | |
CN109934129B (en) | Face feature point positioning method, device, computer equipment and storage medium | |
CN111461002B (en) | Sample processing method for thermal imaging pedestrian detection | |
CN111709291B (en) | Takeaway personnel identity recognition method based on fusion information | |
KR101672814B1 (en) | Method for recognizing gender using random forest | |
Badura et al. | Advanced scale-space, invariant, low detailed feature recognition from images-car brand recognition | |
CN102968618A (en) | Static hand gesture recognition method fused with BoF model and spectral clustering algorithm | |
CN111401438B (en) | Image sorting method, device and system | |
CN110866435B (en) | Far infrared pedestrian training method for self-similarity gradient orientation histogram | |
CN105975910A (en) | Technology processing method used for carrying out video identification on moving object and system thereof | |
CN112183333B (en) | Human screen interaction method, system and device based on micro-expressions | |
CN108133210B (en) | Image format identification method and device | |
CN103093273A (en) | Video people counting method based on body weight recognition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |