CN103020582A - Method for computer to identify vehicle type by video image - Google Patents
Method for computer to identify vehicle type by video image Download PDFInfo
- Publication number
- CN103020582A CN103020582A CN2012103505742A CN201210350574A CN103020582A CN 103020582 A CN103020582 A CN 103020582A CN 2012103505742 A CN2012103505742 A CN 2012103505742A CN 201210350574 A CN201210350574 A CN 201210350574A CN 103020582 A CN103020582 A CN 103020582A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- image
- video image
- target area
- training
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method for a computer to identify a vehicle type by a video image. The method is characterized by comprising the following steps of: (1), acquiring a video image by a computer and extracting a target key frame from the video image, exacting a target region to be identified through wiping out background, and pre-processing the video image; (2), carrying out feature extraction on the pre-processed image, selecting an Hu geometric moment invariant in the image shape characteristic as a vehicle characteristic parameter, and calculating the shape characteristic of the target region; and (3), carrying out sample training of all vehicle types through a machine learning method, and carrying out classified predication on the target region to be identified after the sample training is finished so as to identify the vehicle type. The method can semi-automatically classify the vehicle types, so that the classifying accuracy is improved as much as possible.
Description
Technical field
The invention belongs to the image recognition technology field, be specifically related to a kind of computing machine by the method for video image identification type of vehicle.
Background technology
In recent years, the intelligent transportation system development is (ITS) fast, along with the development of computer vision and mode identification technology, for the more effective application of intelligent transportation system provides opportunity.Computer vision is to utilize computing machine to simulate people's visual performance, and information extraction from the image of objective things is processed and understood, and final being used for actually detected, measures and control.
Mainly comprise from the extracting target from images vehicle and extract characteristic parameter based on the vehicle classification process of image recognition; Input characteristic parameter and obtain two stages of classification results in sorter.
Phase one, characteristic parameter choose the tolerance that mainly concentrates on the vehicle physical dimension, absolute geometry size and relative physical dimension are arranged.The absolute geometry size is the actual size of calculating vehicle according to the angle of the distance between video camera and the vehicle and shooting, and the defective that this method exists is, the distance between video camera and the vehicle must remain unchanged, and this is difficult to realize in actual applications.
To obtaining parameter and existing standard is mated, this method operand is little exactly for subordinate phase, the simplest method, but is only applicable in the less situation of number of parameters, and parameter just can't effectively be classified to vehicle very little.Template matching method compares the characteristic parameter and the standard form that obtain, and this need to expend a lot of computing times and poor fault tolerance.
Another actual problem is, present classification is divided into vehicle large, medium and small or car, passenger vehicle, truck three classes, and all requires in actual applications to charge according to seating capacity and the tonnage of vehicle, and therefore above-mentioned classification results is difficult to be applied among the reality.The present invention therefore.
Summary of the invention
The object of the invention is to provide the method for a kind of computing machine by the video image identification type of vehicle, has solved the prior art Computer and has expended the problems such as a lot of computing times and poor fault tolerance by the video image identification type of vehicle.
In order to solve these problems of the prior art, technical scheme provided by the invention is:
A kind of computing machine is characterized in that said method comprising the steps of by the method for video image identification type of vehicle:
(1) computing machine obtains video image, extracts Target key frames from video image, wipes out by background and extracts target area to be identified, and video image is carried out pre-service;
(2) the complete image of pre-service is carried out feature extraction, select Hu geometric invariant moment in the picture shape feature as the vehicle characteristics parameter, calculate the shape facility of target area;
(3) carry out the sample training of all type of vehicle by machine learning method, after sample training is complete, the target area to be identified Forecasting recognition of classifying is gone out type of vehicle.
Preferably, the HU square computing formula that the shape facility of calculating target area adopts in the described method step (2) is:
h
1＝η
20+η
02；
h
2＝(η
20-η
02)
2+η
11 2；
h
3＝(η
30-3η
12)
2+(3η
21-η
03)
2；
h
4＝(η
30+η
12)
2+(η
21+η
03)
2；
h
5＝(η
30-3η
12)(η
30+η
12)[(η
30+η
12)
2-3(η
21+η
03)
2]
+(3η
21-η
03)(η
21+η
03)[3(η
21+η
03)
2
-(η
21+η
03)
2]；
h
6＝(η
20-η
02)[(η
30+η
12)
2-(η
21+η
03)
2]
+4η
11(η
30+η
12)(η
21+η
03)；
h
7＝(3η
21-η
03)(η
21+η
03)[3(3η
21+η
03)
2-(η
21+η
03)
2]
-(η
30-η
12)(η
21+η
03)[3(η
30+η
12)
2
-(η
21+η
03)
2]；
Wherein
(p+q) rank geometric moment of image f (x, y) is defined as: M
Pq=∫ ∫ x
p* y
q* f (x, y) dxdy (p, q=1 ..., ∞); M wherein
00=∫ ∫ f (x, y) dxdy is by first moment (M
01, M
10) determine image centroid (x
c, y
c): x
c=M
10/ M
00y
c=M
01/ M
00, then true origin is moved to x
cAnd y
cThe place has just obtained for the constant center square of picture displacement, i.e. μ
P, q=∫ ∫ [(x-x
c)
p] * [(y-y
c)
p] * f (x, y) dxdy.
Geometric moment is proposed in 1962 by HU (Visual pattern recognition by moment invariants).(p+q) rank geometric moment of image f (x, y) is defined as: M
Pq=∫ ∫ x
p* y
q* f (x, y) dxdy (p, q=1 ..., ∞) square is used to reflect the distribution situation of stochastic variable in statistics, is generalized in the mechanics, it is used as portraying the mass distribution of space object.
Same reason, if we regard the gray-scale value of image as a two dimension or three-dimensional density fonction, the square method namely can be used for the extraction that art of image analysis also is used as characteristics of image so.The most frequently used, the zeroth order square of object has represented " quality " of image: M
00=∫ ∫ f (x, y) dxdy, first moment (M
01, M
10) for determining image centroid (x
c, y
c): x
c=M
10/ M
00y
c=M
01/ M
00If true origin is moved to x
cAnd y
cThe place has just obtained for the constant center square of picture displacement.Such as μ
P, q=∫ ∫ [(x-x
c)
p] * [(y-y
c)
p] * f (x, y) dxdy; Normalization square and centre distance are also basic identical, except each square will be divided by M
00Certain power:
The HU square is the linear combination of normalized centre distance, and doing like this is in order to obtain the moment function of certain feature of representative image, and these moment functions have unchangeability to some variation such as convergent-divergent, rotation and Mirroring Mapping.The HU square is to calculate from centre distance, and its computing formula is as above schemed h1-h7
Preferably, described method step (3) machine learning method is the method that the artificial neural network of SVM support vector machine and BP combines, and first ready car category training sample pictures feature input SVM support vector machine network and BP neural network is carried out sample training.
The basic procedure of technical solution of the present invention is to obtain image---〉image pre-service---〉to extract feature---〉Images Classification, and the process of carrying out vehicle classification identification based on image recognition in the method mainly comprises two stages: 1, from the extracting target from images vehicle and extract characteristic parameter.2, input characteristic parameter and obtain classification results in sorter.The method can semi-automatically to vehicle classification, improve the accuracy of classification as far as possible.
Vehicle classification is characterized as vehicle ' s contour the most intuitively, the vehicle of same type has similar image outline, color and textural characteristics than image, the shape facility of image can better be distinguished dissimilar vehicle, and in the shape facility, the image that the equipment of considering obtains has the different of far and near height and angle, and therefore selecting the moment characteristics that has rotation and translation invariance in the shape facility work is characteristic parameter.
Because the vehicle classification problem based on image recognition has fuzzy, incomplete, uncertain characteristics, good machine learning algorithm can better adapt to semi-automatic study than template method in the past, be both and consider that the various types of vehicles shape is relatively fixing, the vehicle sample size need not be too much factor, therefore adopt the SVM support vector machine to finish Feature Mapping with the method that the BP neural network combines.Consider at last the needs of practical application, according to as quantity classification being designed to car, SUV and passenger vehicle.
The present invention compared with prior art has following beneficial effect:
Technical solution of the present invention selects Hu geometric invariant moment in the picture shape feature as the vehicle characteristics parameter, compares existing absolute geometry feature and can more accurately identify vehicle characteristics.It belongs to relative geometric properties, does not require the actual size that calculates vehicle, but calculates the ratio between each geometric parameter of vehicle, this can reflect the one-piece construction of vehicle, and the shape size of vehicle do not limited, namely require the installation of camera more flexible, enlarged range of application.And Hu geometric invariant moment technology maturation, good application foundation is arranged.
The sorting technique that technical solution of the present invention selects the artificial neural network of SVM support vector machine and BP to combine, more accurately classification samples.Compare with traditional template matching method, Artificial Neural Network more is good at and is solved that image recognition is this to have problems fuzzy, incomplete, uncertain characteristics.The SVM support vector machine is good at processing small sample and function nonlinear fitting problem simultaneously, compare template matching method and be limited by the parameter scale, the method that the SVM support vector machine combines with artificial neural network still has faster computing velocity and fault-tolerance when the larger sample for the treatment of scale.Test sample after tested, the BP neural network is high for the discrimination of passenger vehicle, reach 88%, and the SVM support vector machine has higher discrimination for car, reaches 85%, and both are in conjunction with the classification that has better solved three class vehicles.
Technical solution of the present invention has been optimized criteria for classification: existing criteria for classification is many to be divided into vehicle large, medium and small or car truck passenger vehicle three classes, actual in having little significance.The present invention according to the passenger stock amount of seats, is divided into passenger vehicle (greater than 10), SUV(greater than 5 and less than 10 with criteria for classification with the criteria for classification refinement), car (less than 5), comparing in the past classification and having more operability.
Description of drawings
The invention will be further described below in conjunction with drawings and Examples:
Fig. 1 is the contour images of passenger vehicle after the pre-service, SUV, car;
Fig. 2 is the pretreated workflow diagram of image;
Fig. 3 is the workflow diagram of image characteristics extraction;
Fig. 4 is the workflow diagram of sorter training.
Embodiment
Below in conjunction with specific embodiment such scheme is described further.Should be understood that these embodiment are not limited to limit the scope of the invention for explanation the present invention.The implementation condition that adopts among the embodiment can be done further adjustment according to the condition of concrete producer, and not marked implementation condition is generally the condition in the normal experiment.
Embodiment
This embodiment Computer carries out in accordance with the following steps by the method for video image identification type of vehicle:
(1) computing machine obtains video image, extracts Target key frames from video image, wipes out by background and extracts target area to be identified, and video image is carried out pre-service;
(2) the complete image of pre-service is carried out feature extraction, select Hu geometric invariant moment in the picture shape feature as the vehicle characteristics parameter, calculate the shape facility of target area;
(3) carry out the sample training of all type of vehicle by machine learning method, after sample training is complete, the target area to be identified Forecasting recognition of classifying is gone out type of vehicle.
Whole technical scheme mainly contains three parts: video image pre-service, feature extraction and sorter training.
1, video image pre-service: from video image, extract Target key frames, use the background technology of wiping out to extract target area to be identified, to the two field picture that contains the target area strengthen, noise reduction, correct, cut apart, the image such as binaryzation processes, and makes it relatively clear and be beneficial to feature extraction and preserve the result who obtains.Fig. 1 is pretreated three kinds of automobile image profiles.
2, feature extraction: utilize hu square computing formula:
h
1＝η
20+η
02；
h
2＝(η
20-η
02)
2+η
11 2；
h
3＝(η
30-3η
12)
2+(3η
21-η
03)
2；
h
4＝(η
30+η
12)
2+(η
21+η
03)
2；
h
5＝(η
30-3η
12)(η
30+η
12)[(η
30+η
12)
2-3(η2
1+η
03)
2]
+(3η
21-η
03)(η
21+η
03)[3(η
21+η
03)
2
-(η
21+η
03)
2]；
h
6＝(η
20-η
02)[(η
30+η
12)
2-(η
21+η
03)
2]
+4η
11(η
30+η
12)(η
21+η
03)；
h
7＝(3η
21-η
03)(η
21+η
03)[3(3η
21+η
03)
2-(η
21+η
03)
2]
-(η
30-η
12)(η
21+η
03)[3(η
30+η
12)
2
-(η
21+η
03)
2]；
Wherein
(p+q) rank geometric moment of image f (x, y) is defined as: M
Pq=∫ ∫ x
p* y
q* f (x, y) dxdy (p, q=1 ..., ∞); M wherein
00=∫ ∫ f (x, y) dxdy is by first moment (M
01, M
10) determine image centroid (x
c, y
c): x
c=M
10/ M
00y
c=M
01/ M
00, then true origin is moved to x
cAnd y
cThe place has just obtained for the constant center square of picture displacement, i.e. μ
P, q=∫ ∫ [(x-x
c)
p] * [(y-y
c)
p] * f (x, y) dxdy.
The complete image of pre-service is carried out the shape facility that the target area is calculated in computing, preserve the result who obtains.
3, ready training sample pictures feature input SVM support vector machine network and BP neural network are carried out sample training, can treat the classification samples prediction of classifying after the sorter training is complete.
As shown in Figure 2, the pretreated program of image comprises and judges whether current video image exists the step of background image.When there is background image in current video image, background image updating; Otherwise, with current video image as foreground image.Background image and foreground image are carried out the phase reducing, obtain movement destination image; Movement destination image is carried out binaryzation, noise reduction, burn into expansion enhancing processing, and the profile that extracts moving target obtains the moving target contour images, then the moving target contour images is carried out subsequent treatment.
As shown in Figure 3, during image characteristics extraction, calculate the Hu square according to the moving target contour images, obtain image features.
The flow process of sorter training imports sorter as shown in Figure 4 with training sample, and classifier parameters carries out initialization, and then whether the training of judgement sample satisfies accuracy requirement.When training sample satisfies accuracy requirement, the storage classifier parameters; Otherwise continue training, whether the accuracy requirement of cycle criterion training sample satisfies the standard of sorter.
The result who obtains is shown in table 1 ~ 5.
Table 1 part sample shape characteristic (after the normalization)
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | |
| 1 | 0.1598 | 0.1290 | 0.0141 | 0.0033 | 1.2205e-04 | 0.0075 | 0.0950 |
| 2 | 0.6275 | 0.5381 | 0.6137 | 0.4254 | 0.2074 | 0.2853 | 0.5357 |
| 3 | 0.7485 | 0.6967 | 0.2959 | 0.1585 | 0.0297 | 0.0888 | 0.2051 |
| 4 | 0.5266 | 0.4221 | 0.4475 | 0.1646 | 0.0323 | 0.0511 | 0.2853 |
| 5 | 0.3389 | 0.2577 | 0.2018 | 0.0616 | 0.0033 | 0.0141 | 0.0579 |
| 6 | 0.6023 | 0.5213 | 0.3209 | 0.0929 | 0.0035 | 0 | 0 |
| 7 | 0.4814 | 0.4105 | 0.1006 | 0.0565 | 0.0042 | 0.0380 | 0.0852 |
| 8 | 0.8246 | 0.7888 | 0.2654 | 0.1364 | 0.0215 | 0.0789 | 0.1868 |
Table 2 neural metwork training finishes the rear section weights
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | |
| 1 | 2.1329 | 0.0276 | -1.6270 | 1.9915 | -0.6882 | 1.1932 | 0.1429 |
| 2 | -0.9149 | 1.1411 | -1.3899 | -0.8599 | 1.8962 | 1.4543 | 1.6001 |
| 3 | 0.4056 | 1.8594 | 1.6207 | -1.4432 | 0.4065 | -0.5678 | 2.0644 |
| 4 | -1.3425 | 2.2326 | -1.1944 | -1.2100 | 0.2417 | 0.3297 | -1.7990 |
| 5 | 1.2410 | 0.2332 | 1.5522 | 0.5731 | 2.0604 | -2.0948 | 0.3399 |
| 6 | -1.2491 | -1.8432 | -1.3082 | -0.1362 | -1.0923 | -2.2751 | -0.1561 |
Table 3 neural metwork training finishes the rear section weights
| 1 | 2 | 3 | 4 | 5 | 6 | |
| 1 | -0.9762 | -0.6756 | -0.3776 | -0.6687 | -0.4741 | 0.3784 |
| 2 | -0.3258 | 0.5886 | 0.0571 | 0.2040 | 0.3082 | 0.4963 |
Two groups of weight w and the v of neural network have trained complete.
Be the bivector of 6*7,
Bivector for 2*6.By
Namely finish some input feature values to the mapping of hidden layer, obtained the result
By
Namely finish the mapping of hidden layer to output layer, obtained last Output rusults
Model after the training of table 4 SVM support vector machine finishes
| Field | Value | Min | Max |
| Parameters | [1;2;3;2.80000;0] | 2 | 2 |
| Nr_class | 2 | 2 | 2 |
| totalCV | 28 | 28 | 28 |
| Rho | -0.5366 | -0.5366 | -0.5366 |
| Label | [1;0] | 0 | 1 |
| ProbA | [] | ||
| ProbB | [] | ||
| nSV | [15;13] | 13 | |
| sv_coef | <28＊1 double> | -1.78 | 1.78 |
| Svs | <28＊7 double> |
Table 5 part predicts the outcome, and numeral 1 represents car, and numeral 0 represents SUV
| 1 | |
| 1 | 0 |
| 2 | 1 |
| 3 | 1 |
| 4 | 1 |
| 5 | 1 |
| 6 | 1 |
| 7 | 1 |
The result shows, test sample book collection degree of accuracy is about 84%.
Above-mentioned example only is explanation technical conceive of the present invention and characteristics, and its purpose is to allow the people who is familiar with technique can understand content of the present invention and according to this enforcement, can not limit protection scope of the present invention with this.All equivalent transformations that Spirit Essence is done according to the present invention or modification all should be encompassed within protection scope of the present invention.
Claims (3)
1. a computing machine is characterized in that said method comprising the steps of by the method for video image identification type of vehicle:
(1) computing machine obtains video image, extracts Target key frames from video image, wipes out by background and extracts target area to be identified, and video image is carried out pre-service;
(2) the complete image of pre-service is carried out feature extraction, select Hu geometric invariant moment in the picture shape feature as the vehicle characteristics parameter, calculate the shape facility of target area;
(3) carry out the sample training of all type of vehicle by machine learning method, after sample training is complete, the target area to be identified Forecasting recognition of classifying is gone out type of vehicle.
2. method according to claim 1 is characterized in that calculating in the described method step (2) the HU square computing formula that the shape facility of target area adopts and is:
h
1＝η
20+η
02；
h
2＝(η
20-η
02)
2+η
11 2；
h
3＝(η
30-3η
12)
2+(3η
21-η
03)
2；
h
4＝(η
30+η
12)
2+(η
21+η
03)
2；
h
5＝(η
30-3η
12)(η
30+η
12)[(η
30+η
12)
2-3(η
21+η
03)
2]
+(3η
21-η
03)(η
21+η
03)[3(η
21+η
03)
2
-(η
21+η
03)
2]；
h
6＝(η
20-η
02)[(η
30+η
12)
2-(η
21+η
03)
2]
+4η
11(η
30+η
12)(η
21+η
03)；
h
7＝(3η
21-η
03)(η
21+η
03)[3(3η
21+η
03)
2-(η
21+η
03)
2]
-(η
30-η
12)(η
21+η
03)[3(η
30+η
12)
2
-(η
21+η
03)
2]；
Wherein
(p+q) rank geometric moment of image f (x, y) is defined as: M
Pq=∫ ∫ x
p* y
q* f (x, y) dxdy (p, q=1 ..., ∞); M wherein
00=∫ ∫ f (x, y) dxdy is by first moment (M
01, M
10) determine image centroid (x
c, y
c): x
c=M
10/ M
00y
c=M
01/ M
00, then true origin is moved to x
cAnd y
cThe place has just obtained for the constant center square of picture displacement, i.e. μ
P, q=∫ ∫ [(x-x
c)
p] * [(y-y
c)
p] * f (x, y) dxdy.
3. method according to claim 1, it is characterized in that described method step (3) machine learning method is the method that the artificial neural network of SVM support vector machine and BP combines, first ready car category training sample pictures feature input SVM support vector machine network and BP neural network are carried out sample training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103505742A CN103020582A (en) | 2012-09-20 | 2012-09-20 | Method for computer to identify vehicle type by video image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103505742A CN103020582A (en) | 2012-09-20 | 2012-09-20 | Method for computer to identify vehicle type by video image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103020582A true CN103020582A (en) | 2013-04-03 |
Family
ID=47969174
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012103505742A Pending CN103020582A (en) | 2012-09-20 | 2012-09-20 | Method for computer to identify vehicle type by video image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103020582A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413439A (en) * | 2013-07-30 | 2013-11-27 | 长安大学 | Method for sorting passenger vehicles and goods vehicles based on videos |
| CN103577756A (en) * | 2013-11-05 | 2014-02-12 | 北京奇虎科技有限公司 | Virus detection method and device based on script type judgment |
| CN104883278A (en) * | 2014-09-28 | 2015-09-02 | 北京匡恩网络科技有限责任公司 | Method for classifying network equipment by utilizing machine learning |
| CN105469095A (en) * | 2015-11-17 | 2016-04-06 | 电子科技大学 | Vehicle model identification method based on pattern set histograms of vehicle model images |
| CN105469062A (en) * | 2015-12-04 | 2016-04-06 | 电子科技大学 | Principal component analysis network framework based vehicle type identification method |
| CN105684431A (en) * | 2015-12-19 | 2016-06-15 | 张北江 | Image identifying method and system in monitoring system |
| CN105678275A (en) * | 2016-01-15 | 2016-06-15 | 苏州大学 | Vehicle identification method |
| CN105975949A (en) * | 2016-05-26 | 2016-09-28 | 大连理工大学 | Visual-information-based automobile identification method |
| CN106504542A (en) * | 2016-09-30 | 2017-03-15 | 北京中星微电子有限公司 | Speed intelligent monitoring method and system |
| CN108205643A (en) * | 2016-12-16 | 2018-06-26 | 同方威视技术股份有限公司 | Image matching method and device |
| CN108460420A (en) * | 2018-03-13 | 2018-08-28 | 江苏实达迪美数据处理有限公司 | A method of classify to certificate image |
| CN108961785A (en) * | 2018-07-10 | 2018-12-07 | 杭州利艾智能科技有限公司 | A kind of system and method adjusting traffic control signal |
| CN109460711A (en) * | 2018-10-11 | 2019-03-12 | 北京佳格天地科技有限公司 | Object identifying method, device, storage medium and terminal |
| CN110991377A (en) * | 2019-12-11 | 2020-04-10 | 辽宁工业大学 | Monocular visual neural network-based front target identification method for automobile safety auxiliary system |
| CN111325752A (en) * | 2018-12-17 | 2020-06-23 | 北京华航无线电测量研究所 | Visual auxiliary method for accurate landing and dynamic pose adjustment of helicopter |
| CN112929561A (en) * | 2021-01-19 | 2021-06-08 | 北京达佳互联信息技术有限公司 | Multimedia data processing method and device, electronic equipment and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101110161A (en) * | 2007-08-31 | 2008-01-23 | 北京科技大学 | System for automatic cab model recognition and automatic vehicle flowrate detection and method thereof |
-
2012
- 2012-09-20 CN CN2012103505742A patent/CN103020582A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101110161A (en) * | 2007-08-31 | 2008-01-23 | 北京科技大学 | System for automatic cab model recognition and automatic vehicle flowrate detection and method thereof |
Non-Patent Citations (1)
| Title |
|---|
| 谢磊: "智能交通系统中的视频处理技术研究", 《中国博士学位论文全文数据库》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413439A (en) * | 2013-07-30 | 2013-11-27 | 长安大学 | Method for sorting passenger vehicles and goods vehicles based on videos |
| CN103413439B (en) * | 2013-07-30 | 2016-01-20 | 长安大学 | A kind of passenger vehicle based on video and lorry sorting technique |
| CN103577756A (en) * | 2013-11-05 | 2014-02-12 | 北京奇虎科技有限公司 | Virus detection method and device based on script type judgment |
| CN104883278A (en) * | 2014-09-28 | 2015-09-02 | 北京匡恩网络科技有限责任公司 | Method for classifying network equipment by utilizing machine learning |
| CN105469095A (en) * | 2015-11-17 | 2016-04-06 | 电子科技大学 | Vehicle model identification method based on pattern set histograms of vehicle model images |
| CN105469062A (en) * | 2015-12-04 | 2016-04-06 | 电子科技大学 | Principal component analysis network framework based vehicle type identification method |
| WO2017101127A1 (en) * | 2015-12-19 | 2017-06-22 | 张北江 | Method and system for image recognition in surveillance system |
| CN105684431A (en) * | 2015-12-19 | 2016-06-15 | 张北江 | Image identifying method and system in monitoring system |
| CN105678275A (en) * | 2016-01-15 | 2016-06-15 | 苏州大学 | Vehicle identification method |
| CN105975949A (en) * | 2016-05-26 | 2016-09-28 | 大连理工大学 | Visual-information-based automobile identification method |
| CN106504542A (en) * | 2016-09-30 | 2017-03-15 | 北京中星微电子有限公司 | Speed intelligent monitoring method and system |
| CN108205643A (en) * | 2016-12-16 | 2018-06-26 | 同方威视技术股份有限公司 | Image matching method and device |
| CN108460420A (en) * | 2018-03-13 | 2018-08-28 | 江苏实达迪美数据处理有限公司 | A method of classify to certificate image |
| CN108961785A (en) * | 2018-07-10 | 2018-12-07 | 杭州利艾智能科技有限公司 | A kind of system and method adjusting traffic control signal |
| CN108961785B (en) * | 2018-07-10 | 2020-10-02 | 杭州利艾智能科技有限公司 | System and method for adjusting traffic control signal |
| CN109460711A (en) * | 2018-10-11 | 2019-03-12 | 北京佳格天地科技有限公司 | Object identifying method, device, storage medium and terminal |
| CN111325752A (en) * | 2018-12-17 | 2020-06-23 | 北京华航无线电测量研究所 | Visual auxiliary method for accurate landing and dynamic pose adjustment of helicopter |
| CN111325752B (en) * | 2018-12-17 | 2023-06-13 | 北京华航无线电测量研究所 | Helicopter accurate landing and dynamic pose adjustment vision auxiliary method |
| CN110991377A (en) * | 2019-12-11 | 2020-04-10 | 辽宁工业大学 | Monocular visual neural network-based front target identification method for automobile safety auxiliary system |
| CN110991377B (en) * | 2019-12-11 | 2023-09-19 | 辽宁工业大学 | Front mesh identification method of automobile safety auxiliary system based on monocular vision neural network |
| CN112929561A (en) * | 2021-01-19 | 2021-06-08 | 北京达佳互联信息技术有限公司 | Multimedia data processing method and device, electronic equipment and storage medium |
| CN112929561B (en) * | 2021-01-19 | 2023-04-28 | 北京达佳互联信息技术有限公司 | Multimedia data processing method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103020582A (en) | Method for computer to identify vehicle type by video image | |
| CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
| CN109117826B (en) | Multi-feature fusion vehicle identification method | |
| CN108830188B (en) | Vehicle detection method based on deep learning | |
| Jolly et al. | Vehicle segmentation and classification using deformable templates | |
| CN106127747B (en) | Car surface damage classifying method and device based on deep learning | |
| CN103258213B (en) | A kind of for the dynamic vehicle model recognizing method in intelligent transportation system | |
| CN100492399C (en) | Method for making human face posture estimation utilizing dimension reduction method | |
| CN101944174B (en) | Identification method of characters of licence plate | |
| CN107767387B (en) | Contour detection method based on variable receptive field scale global modulation | |
| CN109359684A (en) | Fine granularity model recognizing method based on Weakly supervised positioning and subclass similarity measurement | |
| CN104598885B (en) | The detection of word label and localization method in street view image | |
| CN111862093A (en) | Corrosion grade information processing method and system based on image recognition | |
| CN111275044A (en) | Weak supervision target detection method based on sample selection and self-adaptive hard case mining | |
| CN108960055B (en) | Lane line detection method based on local line segment mode characteristics | |
| CN112183237A (en) | Automatic white blood cell classification method based on color space adaptive threshold segmentation | |
| CN117011274A (en) | Automatic glass bottle detection system and method thereof | |
| CN117152625A (en) | Remote sensing small target identification method, system, equipment and medium based on CoordConv and Yolov5 | |
| CN104573722A (en) | Three-dimensional face race classifying device and method based on three-dimensional point cloud | |
| CN108090492B (en) | Contour detection method based on scale clue suppression | |
| CN111461002B (en) | Sample processing method for thermal imaging pedestrian detection | |
| CN113239865A (en) | Deep learning-based lane line detection method | |
| Alyahya et al. | Saudi license plate recognition system using artificial neural network classifier | |
| CN105160666B (en) | SAR image change detection based on Non-Stationary Analysis and condition random field | |
| Bedruz et al. | Vehicle classification using AKAZE and feature matching approach and artificial neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20170301 |
|
| AD01 | Patent right deemed abandoned |