CN105404857A - Infrared-based night intelligent vehicle front pedestrian detection method - Google Patents
Infrared-based night intelligent vehicle front pedestrian detection method Download PDFInfo
- Publication number
- CN105404857A CN105404857A CN201510738594.0A CN201510738594A CN105404857A CN 105404857 A CN105404857 A CN 105404857A CN 201510738594 A CN201510738594 A CN 201510738594A CN 105404857 A CN105404857 A CN 105404857A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- image
- sample
- infrared
- sorter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides an infrared-based night intelligent vehicle front pedestrian detection method. The method comprises the following steps that de-noising preprocessing is performed on input infrared image sequences by utilizing a smoothing filtering method and a morphological processing technology; pedestrian preselection areas in the input image sequences are captured by a vertical projection method based on pixel gradients, and areas of interest are extracted from the pedestrian preselection areas according to pedestrian geometric features; infrared image multistage binary mode feature description areas of interest are extracted; a pedestrian classifier model is offline trained by utilizing a support vector machine algorithm; and the areas of interest are judged to be target pedestrians or backgrounds online by utilizing the classifier model. Compared with methods in the prior art, robustness of the pedestrian classifier is effectively improved by the method so that the method can detect vertical pedestrians in the still, walking or running and other movement modes and the method is also suitable for detecting human bodies riding bicycles or motorcycles, and a system realized based on the method can be applied to barrier detection for an unmanned intelligent vehicle and a wheeled robot, etc.
Description
Technical field
The invention belongs to computer vision and pattern-recognition, intelligent transportation field, be specially a kind of based on infrared intelligent vehicle at night front pedestrian detection method.
Background technology
Along with the development of artificial intelligence technology in recent years, various intelligent vehicle control loop emerges in an endless stream, the attention of the scholar that is more and more subject in recent years being correlated with both at home and abroad.Along with increasing rapidly of vehicle guaranteeding organic quantity, bring great convenience to the life of people and trip, but the road traffic accident brought thus all causes huge loss to the people's lives and properties of every country and national economy every year.Especially the accident occurred under night or overcast sky particularly serious (permitted to rise, yellow iron army, Tian Yonghong. pedestrian detection technology summary [J] in vehicle-mounted vision system. Chinese image graphics journal, 2013).
The fact shows, pedestrian detection is the gordian technique in vehicle DAS (Driver Assistant System) and unmanned intelligent vehicle field, it can a Quick Test Vehicle front pedestrian, carry out safe early warning in time and keep away barrier, reduce or avoid vehicle and pedestrian collision's accident, there is potential economic worth and application prospect widely.
Current pedestrian detection mainly contains two kinds of approach: one is based on visible images, and another kind is based on infrared image.Compared with visual light imaging, infrared imaging has significant advantage: because infrared image is thermal imaging, when at night, there is light is very dark the ability through dark and smog, and not by the impact of visible ray, realize remote, the round-the-clock observation to interesting target.Pedestrian detection technology at night based on infrared image can be implemented in pedestrian detection under night or severe weather conditions and identification, has very large advantage compared with the pedestrian detection based on visible images.
Be exactly, after the infrared image that collected by infrared equipment carries out series of preprocessing, use algorithm the pedestrian in image accurately detected and identify, realize the classification of pedestrian and non-pedestrian based on the pedestrian detection technology of infrared image.Present intelligent vehicle detection system for pedestrian detection Algorithm robustness and requirement of real-time higher, under infrared image, pedestrian detecting system generally all comprises area-of-interest (Regionofinterest, ROI) and extracts and target identification two greatly module.
The extraction algorithm of area-of-interest is divided into the method based on temperature information, the method based on characteristics of human body and the method based on salient region substantially.Based on the segmentation of temperature information (as RonanO ' Malley, MartinGlavin, EdwardJones, AnEfficientRegionofInterestGenerationTechniqueforFar-Inf raredPedestrianDetection.DigestofTechnicalPapers-Interna tionalConferenceonConsumerElectronics (ICCE ' 08), LasVegas, NV, USA, 2008) be carry out according to the heat difference between human body target temperature and background object, but the shortcoming of this kind of method is easily split by the high temp objects of non-pedestrian.Based on the method for characteristics of human body (as BarnichO, VanDroogenbroeckM.ViBe, Auniversalbackgroundsubtractionalgorithmforvideosequence s [J] .ImageProcessing, IEEETransactionson, 2011) according to the characteristic feature Modling model obtaining human body target in infrared image, thus the segmentation of accomplished pedestrian target.The adaptive ability of this method is comparatively strong, but is not easy to define robust, fast feature.Based on the method for salient region (as AkulaA, KhannaN, GhoshR, etal.Adaptivecontour-basedstatisticalbackgroundsubtracti onmethodformovingtargetdetectionininfraredvideosequences [J] .InfraredPhysics & Technology, 2014) consider the salient region that the Image Segmentation Methods Based on Features such as brightness, color, direction obtain human body target.The shortcoming of this kind of method be circumstance complication and pedestrian is blocked when easily leak segmentation.
According to the difference utilizing information, target identification can be divided into identification two kinds of methods of based drive identification and Shape-based interpolation, and the recognition methods of Shape-based interpolation comprises parameter model, template matches and statistical classification three kinds.Based drive recognition methods is (as FanX, MittalS, PrasadT, etal.PedestrianDetectionandTrackingUsingDeformablePartMo delsandKalmanFiltering [J] .JournalofCommunicationandComputer, 2013) refer to that gait (Gait) feature when being moved by analyst identifies pedestrian.According to the feature of infrared image, the general recognition methods detection pedestrian target using Shape-based interpolation.The recognition methods of Shape-based interpolation is (as DanielOlmedaa, CristianoPremebida, UrbanoNunes, JoseMariaArmingol, ArturodelaEscaleraa.Pedestriandetectioninfarinfraredimag es [J] .IntegratedComputer-AidedEngineering, 2013) refer to be identified target by the gray scale of evaluating objects, edge and texture information.Based on the method (YuG of parameter model, SapiroG, MallatS.Solvinginverseproblemswithpiecewiselinearestimat ors:fromGaussianmixturemodelstostructuredsparsity [J] .ImageProcessing, IEEETransactionson, 2012) knowledge according to organization of human body is referred to, construct a clear and definite 2D or 3D parameter model, carry out solving model by the low-level image feature extracting image, thus identify pedestrian.The advantage of this method has clear and definite model, can process occlusion issue, and can infer the attitude of human body.Shortcoming is the more difficult structure of model, and model solution is more complicated also.Based on the method for template matches by storing some gray scales or contour mould represents pedestrian, when identification, only need the distance of measuring template and input window just can identify pedestrian.Based on the algorithm of template matches (as SunH, HuaC, LuoY.Amulti-stageclassifierbasedalgorithmofpedestriandet ectioninnightwithanearinfraredcamerainamovingcar [C], IEEEFirstSymposiumon.2004) advantage calculates simply, shortcoming is the complicacy due to pedestrian's attitude, is difficult to construct enough templates to process different attitudes.The method of Corpus--based Method classification is (as ChengTY, HermanC.Motiontrackingininfraredimagingforquantitativeme dicaldiagnosticapplications [J] .InfraredPhysics & Technology, 2014) sorter is obtained by machine learning from a series of training data learning, represent pedestrian with this sorter, then utilize this sorter to identify input window.The method compares robust, but needs a lot of training data, and is difficult to solve attitude and the problem of blocking.
In sum, although current infrared vehicle-mounted pedestrian detection method is just effective, still in the urgent need to further improvement in Detection accuracy, detection efficiency and robustness.
Summary of the invention
The object of the invention is to the intelligent vehicle front night infrared image pedestrian detection method proposing a kind of real-time robust, to improving existing infrared pedestrian detection method accuracy rate, real-time, the unmanned intelligent vehicle researched and developed at home by this detection method is tested, and test result shows that the effect of testing meets the real-time and robustness requirement that intelligent vehicle independently travels.
For achieving the above object, the technical solution used in the present invention is a kind of intelligent vehicle front night infrared image pedestrian detection method of real-time robust, and it comprises the steps:
(1) smooth filtering method, Morphological scale-space technology is first utilized to carry out noise suppression preprocessing to input infrared image sequence;
(2) by catching the pedestrian's preselected area in input image sequence based on pixel gradient vertical projection method, from pedestrian's preselected area, area-of-interest (regionsofinterest, ROIs) is extracted according to pedestrian's geometric properties;
(3) extract infrared image multistage binary pattern feature (Multi-levelbinarypattern, MBP) and describe area-of-interest;
(4) support vector machine (supportvectormachine, SVM) algorithm off-line training pedestrian sorter model is utilized;
(5) area-of-interest is as target pedestrian or background to utilize this sorter model to judge online.
Further, step (1) described smooth filtering method is that the infrared image sequence got infrared photography instrument carries out medium filtering to remove noise spot isolated in image; Morphological scale-space technology described in step (1) is by noise pixel less in morphological erosion computing filtering image, and utilizes morphological dilations computing to fill up Weak link region (weaklyconnectedregions).
Further, step (2) described pedestrian preselected area utilizes the thermal-radiating feature of infrared imaging, causes pedestrian region and peripheral region gray-value variation in image obviously, to determine pedestrian's preselected area by pixel gradient projecting method; Step (2) described pedestrian geometric properties is partitioned into general pedestrian region according to the constraint of pedestrian's the ratio of width to height.
Further, step (4) described off-line training pedestrian sorter specifically refers to: collect containing pedestrian picture sample and only containing the picture composition of sample training sample space of having powerful connections, extract the multistage binary pattern feature of training set, the training mechanism adopting iteration to collect difficult sample learns training set; Described study adopts the support vector machine learning algorithm based on histogram intersection core (histogramintersectionkernel, HIK), obtains support vector machine pedestrian sorter; Picture sample is obtained by hand cutting, is highly all more than or equal to 12 pixels, and sample size is adjusted to 64 pixel × 128 pixels.
Compared with existing far infrared vehicle-mounted pedestrian at night detection technique, tool of the present invention has the following advantages and effect: the noise in step (1), the smothing filtering of input picture and Morphological scale-space being greatly reduced to image, makes the process in region of interesting extraction stage more accurate; Method described in step (2) is used for rapid extraction pedestrian area-of-interest, improves the real-time of algorithm; The described multistage binary pattern feature of step (3) can describe far infrared pedestrian target more subtly, significantly improves the accuracy rate of far infrared pedestrian detection; In step (4), the off-line repetitive exercise mechanism of pedestrian's sorter is conducive to collecting the difficult negative sample being difficult to be obtained by manual mode, effectively improves the robustness of pedestrian's sorter; And this repetitive exercise mechanism is applicable to but is not limited to the training process of infrared vehicle-mounted pedestrian detection at night sorter.The present invention can not only detect the upright pedestrian be under static, walking or the mode of motion such as running, also be applicable to detecting by bike or the human body of motorcycle, and the system according to said method realized can be applied to the detection of obstacles of unmanned intelligent vehicle, wheeled robot etc.
Accompanying drawing explanation
Fig. 1 is that the principle process of intelligent vehicle front night infrared image pedestrian detection method in embodiment implements schematic diagram.
Fig. 2 a is the enforcement illustration of a frame original input picture in embodiment.
Image after the denoising that Fig. 2 b is input picture shown in Fig. 2 a in embodiment.
Fig. 3 is training sample segment template pedestrian image.
Fig. 4 a is training sample part pedestrian image.
Fig. 4 b is partial detection schematic diagram.
Embodiment
Detailed embodiment operating process is provided below in conjunction with accompanying drawing.Accompanying drawing illustrated embodiment is premised on technical solution of the present invention, and the embodiment described in accompanying drawing belongs to but is not limited to scope.
As shown in Figure 1, the inventive method main body comprises two parts to the overall flow of the inventive method: the off-line training of pedestrian's sorter and the on-line checkingi of pedestrian.Its concrete implementation step is as follows:
(1) infrared picture data that thermal imaging system exports is read.Fig. 2 a is the enforcement illustration of a frame original input picture in embodiment.
(2) carry out the pre-service such as medium filtering, Morphological scale-space filtering noise to image, the size of window is 3 × 3.Image after the denoising that Fig. 2 b is input picture shown in Fig. 2 a in embodiment.
(3) to the result images data of step (2), calculate its gradient image, adopt centrosymmetric first order difference mask in the present embodiment, during gradient component as calculated level direction, mask used is [1,0,1].Gained gradient image is projected to vertical direction, obtains gradient vertical drop shadow curve.The pixel count x that in this curve statistical gained gradient image, in vertical direction, each row exists
i, i=1,2 ..., l, wherein, l is the pixel count that input picture comprises in the horizontal direction, is the columns that gained gradient vertical drop shadow curve has.Then each x in gradient projection curve is searched for
ivalue is greater than threshold value T and line segment continuous print crest, and the position at record crest place the original image region of extracting its correspondence are as pedestrian's preselected area.Last region of removing ungratified requirement according to the length breadth ratio R of pedestrian.R is obtained by formula (1).
r＝W/H(1)
Wherein, W is area-of-interest width, and H is area-of-interest height, and r is the ratio of width to height of area-of-interest.In infrared image, the ratio of R is about 0.4 as calculated.
(4) obtain training sample by manually to cut out infrared video, specifically have the positive sample comprising the state pedestrians such as static, walking or running and the negative sample only comprising background, always have 2400 width pictures.All sample standard deviations are greater than 12 pixels, adopt bilinear interpolation that picture sample size is adjusted to 64 pixel × 128 pixels.Fig. 4 a is training sample part pedestrian image.
(5) multi-scale image pyramid decomposition is carried out to training sample, the present embodiment mesoscale is 2, extract the Central Symmetry local binary patterns (Center-SymmetricLocalBinaryPattern in each scale layer, CS-LBP) feature, finally does to the CS-LBP feature of different resolution layer the multistage local mode character representation that weighted sum obtains image.Shown in (3).
The computing formula of Central Symmetry local binary patterns is such as formula shown in (2).
Wherein, R is radius of a circle, and N is pixel number circumferentially.N
iand n
i+ (N/2)corresponding with center pixel to centrosymmetric pixel value pair.
F
MBP＝Σσ
lF
l,CS-LBP(3)
Wherein, σ
lweights, F
l, CS-LBPit is the CS-LBP Feature Descriptor of l layer.
(6), extract the multistage local mode feature of sample from training set after, the support vector machine learning algorithm training classifier respectively based on histogram intersection core is adopted.The sample searched for and divided by mistake concentrated by current pedestrian's sorter at original video, it can be used as newly-increased training sample and again train pedestrian's sorter, thus the decision hyperplane of adjustment support vector machine pedestrian sorter; When gained pedestrian sorter termination of iterations training process when verifying that the currency increment of the global classification accuracy rate on collection is less than predetermined threshold, and export pedestrian's sorter.
(7) extract the MBP feature of the pedestrian's area-of-interest obtained through step (3), and input the SVM classifier trained and carry out classifying and draw the net result of pedestrian detection.Fig. 4 b is partial detection schematic diagram.
Claims (2)
1., based on infrared intelligent vehicle at a night front pedestrian detection method, it is characterized in that:
It comprises the steps,
(1) smooth filtering method, Morphological scale-space technology is first utilized to carry out noise suppression preprocessing to input infrared image sequence;
(2) by catching the pedestrian's preselected area in input image sequence based on pixel gradient vertical projection method, from pedestrian's preselected area, area-of-interest is extracted according to pedestrian's geometric properties;
(3) infrared image multistage binary pattern feature interpretation area-of-interest is extracted;
(4) algorithm of support vector machine off-line training pedestrian sorter model is utilized;
(5) area-of-interest is as target pedestrian or background to utilize this sorter model to judge online;
Further, step (1) described smooth filtering method is that the infrared image sequence got infrared photography instrument carries out medium filtering to remove noise spot isolated in image; Morphological scale-space technology described in step (1) is by noise pixel less in morphological erosion computing filtering image, and utilizes morphological dilations computing to fill up Weak link region;
Further, step (2) described pedestrian preselected area utilizes the thermal-radiating feature of infrared imaging, causes pedestrian region and peripheral region gray-value variation in image obviously, to determine pedestrian's preselected area by pixel gradient projecting method; Step (2) described pedestrian geometric properties is partitioned into general pedestrian region according to the constraint of pedestrian's the ratio of width to height;
Further, step (4) described off-line training pedestrian sorter specifically refers to: collect containing pedestrian picture sample and only containing the picture composition of sample training sample space of having powerful connections, extract the multistage binary pattern feature of training set, the training mechanism adopting iteration to collect difficult sample learns training set; Described study adopts the support vector machine learning algorithm based on histogram intersection core, obtains support vector machine pedestrian sorter; Picture sample is obtained by hand cutting, is highly all more than or equal to 12 pixels, and sample size is adjusted to 64 pixel × 128 pixels.
2. according to claim 1 a kind of based on infrared intelligent vehicle at night front pedestrian detection method, it is characterized in that: this method main body comprises two parts: the off-line training of pedestrian's sorter and the on-line checkingi of pedestrian; Its concrete implementation step is as follows:
(1) infrared picture data that thermal imaging system exports is read;
(2) carry out the pre-service such as medium filtering, Morphological scale-space filtering noise to image, the size of window is 3 × 3;
(3) to the result images data of step (2), its gradient image is calculated; Adopt centrosymmetric first order difference mask, during the gradient component in calculated level direction, mask used is [1,0,1]; Gained gradient image is projected to vertical direction, obtains gradient vertical drop shadow curve; The pixel count x that in this curve statistical gained gradient image, in vertical direction, each row exists
i, i=1,2 ..., l, wherein, l is the pixel count that input picture comprises in the horizontal direction, is the columns that gained gradient vertical drop shadow curve has; Then each x in gradient projection curve is searched for
ivalue is greater than threshold value T and line segment continuous print crest, and the position at record crest place the original image region of extracting its correspondence are as pedestrian's preselected area; Last region of removing ungratified requirement according to the length breadth ratio R of pedestrian; R is obtained by formula (1);
r＝W/H(1)
Wherein, W is area-of-interest width, and H is area-of-interest height, and r is the ratio of width to height of area-of-interest; In infrared image, the ratio of R is about 0.4 as calculated;
(4) obtain training sample by manually to cut out infrared video, specifically have the positive sample comprising the state pedestrians such as static, walking or running and the negative sample only comprising background, always have 2400 width pictures; All sample standard deviations are greater than 12 pixels, adopt bilinear interpolation that picture sample size is adjusted to 64 pixel × 128 pixels;
(5) multi-scale image pyramid decomposition is carried out to training sample, yardstick is 2, extract the Central Symmetry local binary patterns feature in each scale layer, finally the multistage local mode character representation that weighted sum obtains image is done to the CS-LBP feature of different resolution layer; Shown in (3);
The computing formula of Central Symmetry local binary patterns is such as formula shown in (2);
Wherein, R is radius of a circle, and N is pixel number circumferentially; n
iand n
i+ (N/2)corresponding with center pixel to centrosymmetric pixel value pair;
F
MBP＝Σσ
lF
l,CS-LBP(3)
Wherein, σ
lweights, F
l, CS-LBPit is the CS-LBP Feature Descriptor of l layer;
(6), extract the multistage local mode feature of sample from training set after, the support vector machine learning algorithm training classifier respectively based on histogram intersection core is adopted; The sample searched for and divided by mistake concentrated by current pedestrian's sorter at original video, it can be used as newly-increased training sample and again train pedestrian's sorter, thus the decision hyperplane of adjustment support vector machine pedestrian sorter; When gained pedestrian sorter termination of iterations training process when verifying that the currency increment of the global classification accuracy rate on collection is less than predetermined threshold, and export pedestrian's sorter;
(7) extract the MBP feature of the pedestrian's area-of-interest obtained through step (3), and input the SVM classifier trained and carry out classifying and draw the net result of pedestrian detection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510738594.0A CN105404857A (en) | 2015-11-04 | 2015-11-04 | Infrared-based night intelligent vehicle front pedestrian detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510738594.0A CN105404857A (en) | 2015-11-04 | 2015-11-04 | Infrared-based night intelligent vehicle front pedestrian detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105404857A true CN105404857A (en) | 2016-03-16 |
Family
ID=55470334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510738594.0A Pending CN105404857A (en) | 2015-11-04 | 2015-11-04 | Infrared-based night intelligent vehicle front pedestrian detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105404857A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106600628A (en) * | 2016-12-13 | 2017-04-26 | 广州紫川电子科技有限公司 | Target object identification method and device based on infrared thermal imaging system |
| CN107635099A (en) * | 2017-10-09 | 2018-01-26 | 深圳市天视通电子科技有限公司 | A kind of double optical-fiber network video cameras of human body sensing and safety defense monitoring system |
| CN108549864A (en) * | 2018-04-12 | 2018-09-18 | 广州飒特红外股份有限公司 | Area-of-interest filter method towards vehicle-mounted thermal imaging pedestrian detection and device |
| CN108564030A (en) * | 2018-04-12 | 2018-09-21 | 广州飒特红外股份有限公司 | Classifier training method and apparatus towards vehicle-mounted thermal imaging pedestrian detection |
| CN108805027A (en) * | 2018-05-03 | 2018-11-13 | 电子科技大学 | Face identification method under the conditions of low resolution |
| CN109086678A (en) * | 2018-07-09 | 2018-12-25 | 天津大学 | A kind of pedestrian detection method extracting image multi-stage characteristics based on depth supervised learning |
| CN109522813A (en) * | 2018-10-24 | 2019-03-26 | 北京航空航天大学 | A kind of improvement Random Walk Algorithm based on pedestrian's notable feature |
| CN109643480A (en) * | 2016-07-22 | 2019-04-16 | 路晟（上海）科技有限公司 | Security system and method |
| CN109726621A (en) * | 2017-10-31 | 2019-05-07 | 比亚迪股份有限公司 | Pedestrian detection method, device and equipment |
| CN109782775A (en) * | 2019-03-13 | 2019-05-21 | 刘乐 | A kind of automobile obstacle avoidance system based on thermal image |
| CN109815784A (en) * | 2018-11-29 | 2019-05-28 | 广州紫川物联网科技有限公司 | A kind of intelligent method for classifying based on thermal infrared imager, system and storage medium |
| CN109895780A (en) * | 2017-12-07 | 2019-06-18 | 北京京东尚科信息技术有限公司 | A kind of method and apparatus realizing unmanned equipment autonomously and getting rid of poverty |
| CN110148301A (en) * | 2019-06-21 | 2019-08-20 | 北京精英系统科技有限公司 | A method of detection electric vehicle and bicycle |
| CN110390650A (en) * | 2019-07-23 | 2019-10-29 | 中南大学 | OCT image denoising method based on intensive connection and generation confrontation network |
| CN110753238A (en) * | 2019-10-29 | 2020-02-04 | 北京字节跳动网络技术有限公司 | Video processing method, device, terminal and storage medium |
| CN110866481A (en) * | 2019-11-07 | 2020-03-06 | 北京小龙潜行科技有限公司 | Sow oestrus detection method and device |
| CN111242076A (en) * | 2020-01-20 | 2020-06-05 | 江铃汽车股份有限公司 | Pedestrian detection method and system |
| CN114119577A (en) * | 2021-11-30 | 2022-03-01 | 石家庄铁道大学 | High-speed rail tunnel leaky cable buckle detection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198332A (en) * | 2012-12-14 | 2013-07-10 | 华南理工大学 | Real-time robust far infrared vehicle-mounted pedestrian detection method |
| CN103902976A (en) * | 2014-03-31 | 2014-07-02 | 浙江大学 | Pedestrian detection method based on infrared image |
| US8861842B2 (en) * | 2010-02-05 | 2014-10-14 | Sri International | Method and apparatus for real-time pedestrian detection for urban driving |
| CN104239852A (en) * | 2014-08-25 | 2014-12-24 | 中国人民解放军第二炮兵工程大学 | Infrared pedestrian detecting method based on motion platform |
-
2015
- 2015-11-04 CN CN201510738594.0A patent/CN105404857A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8861842B2 (en) * | 2010-02-05 | 2014-10-14 | Sri International | Method and apparatus for real-time pedestrian detection for urban driving |
| CN103198332A (en) * | 2012-12-14 | 2013-07-10 | 华南理工大学 | Real-time robust far infrared vehicle-mounted pedestrian detection method |
| CN103902976A (en) * | 2014-03-31 | 2014-07-02 | 浙江大学 | Pedestrian detection method based on infrared image |
| CN104239852A (en) * | 2014-08-25 | 2014-12-24 | 中国人民解放军第二炮兵工程大学 | Infrared pedestrian detecting method based on motion platform |
Non-Patent Citations (1)
| Title |
|---|
| 黄非非: "基于LBP的人脸识别研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109643480A (en) * | 2016-07-22 | 2019-04-16 | 路晟（上海）科技有限公司 | Security system and method |
| CN106600628B (en) * | 2016-12-13 | 2020-12-22 | 广州紫川电子科技有限公司 | Target object identification method and device based on thermal infrared imager |
| CN106600628A (en) * | 2016-12-13 | 2017-04-26 | 广州紫川电子科技有限公司 | Target object identification method and device based on infrared thermal imaging system |
| CN107635099A (en) * | 2017-10-09 | 2018-01-26 | 深圳市天视通电子科技有限公司 | A kind of double optical-fiber network video cameras of human body sensing and safety defense monitoring system |
| CN107635099B (en) * | 2017-10-09 | 2020-08-18 | 深圳市天视通电子科技有限公司 | Human body induction double-optical network camera and security monitoring system |
| CN109726621A (en) * | 2017-10-31 | 2019-05-07 | 比亚迪股份有限公司 | Pedestrian detection method, device and equipment |
| CN109726621B (en) * | 2017-10-31 | 2021-05-14 | 比亚迪股份有限公司 | Pedestrian detection method, device and equipment |
| CN109895780A (en) * | 2017-12-07 | 2019-06-18 | 北京京东尚科信息技术有限公司 | A kind of method and apparatus realizing unmanned equipment autonomously and getting rid of poverty |
| CN109895780B (en) * | 2017-12-07 | 2021-03-30 | 北京京东乾石科技有限公司 | Method and device for realizing autonomous escaping of unmanned equipment |
| CN108564030A (en) * | 2018-04-12 | 2018-09-21 | 广州飒特红外股份有限公司 | Classifier training method and apparatus towards vehicle-mounted thermal imaging pedestrian detection |
| WO2019196131A1 (en) * | 2018-04-12 | 2019-10-17 | 广州飒特红外股份有限公司 | Method and apparatus for filtering regions of interest for vehicle-mounted thermal imaging pedestrian detection |
| CN108549864A (en) * | 2018-04-12 | 2018-09-18 | 广州飒特红外股份有限公司 | Area-of-interest filter method towards vehicle-mounted thermal imaging pedestrian detection and device |
| CN108549864B (en) * | 2018-04-12 | 2020-04-10 | 广州飒特红外股份有限公司 | Vehicle-mounted thermal imaging pedestrian detection-oriented region-of-interest filtering method and device |
| CN108805027A (en) * | 2018-05-03 | 2018-11-13 | 电子科技大学 | Face identification method under the conditions of low resolution |
| CN109086678B (en) * | 2018-07-09 | 2022-02-25 | 天津大学 | Pedestrian detection method for extracting image multilevel features based on deep supervised learning |
| CN109086678A (en) * | 2018-07-09 | 2018-12-25 | 天津大学 | A kind of pedestrian detection method extracting image multi-stage characteristics based on depth supervised learning |
| CN109522813B (en) * | 2018-10-24 | 2023-06-06 | 北京航空航天大学 | Improved random walk algorithm based on pedestrian salient features |
| CN109522813A (en) * | 2018-10-24 | 2019-03-26 | 北京航空航天大学 | A kind of improvement Random Walk Algorithm based on pedestrian's notable feature |
| CN109815784A (en) * | 2018-11-29 | 2019-05-28 | 广州紫川物联网科技有限公司 | A kind of intelligent method for classifying based on thermal infrared imager, system and storage medium |
| CN109782775B (en) * | 2019-03-13 | 2020-04-14 | 刘乐 | Automobile obstacle avoidance system based on thermal image |
| CN109782775A (en) * | 2019-03-13 | 2019-05-21 | 刘乐 | A kind of automobile obstacle avoidance system based on thermal image |
| CN110148301A (en) * | 2019-06-21 | 2019-08-20 | 北京精英系统科技有限公司 | A method of detection electric vehicle and bicycle |
| CN110390650B (en) * | 2019-07-23 | 2022-02-11 | 中南大学 | OCT image denoising method based on dense connection and generation countermeasure network |
| CN110390650A (en) * | 2019-07-23 | 2019-10-29 | 中南大学 | OCT image denoising method based on intensive connection and generation confrontation network |
| CN110753238A (en) * | 2019-10-29 | 2020-02-04 | 北京字节跳动网络技术有限公司 | Video processing method, device, terminal and storage medium |
| CN110753238B (en) * | 2019-10-29 | 2022-05-06 | 北京字节跳动网络技术有限公司 | Video processing method, device, terminal and storage medium |
| CN110866481A (en) * | 2019-11-07 | 2020-03-06 | 北京小龙潜行科技有限公司 | Sow oestrus detection method and device |
| CN111242076A (en) * | 2020-01-20 | 2020-06-05 | 江铃汽车股份有限公司 | Pedestrian detection method and system |
| CN114119577A (en) * | 2021-11-30 | 2022-03-01 | 石家庄铁道大学 | High-speed rail tunnel leaky cable buckle detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105404857A (en) | Infrared-based night intelligent vehicle front pedestrian detection method | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN103902976B (en) | A kind of pedestrian detection method based on infrared image | |
| CN104951784B (en) | A kind of vehicle is unlicensed and license plate shading real-time detection method | |
| CN110175576A (en) | A kind of driving vehicle visible detection method of combination laser point cloud data | |
| CN106127137A (en) | A kind of target detection recognizer based on 3D trajectory analysis | |
| Cui et al. | Vehicle localisation using a single camera | |
| CN102214309B (en) | Special human body recognition method based on head and shoulder model | |
| CN104091171B (en) | Vehicle-mounted far infrared pedestrian detecting system and method based on local feature | |
| Derpanis et al. | Classification of traffic video based on a spatiotemporal orientation analysis | |
| CN104978567B (en) | Vehicle checking method based on scene classification | |
| CN106682586A (en) | Method for real-time lane line detection based on vision under complex lighting conditions | |
| Romdhane et al. | An improved traffic signs recognition and tracking method for driver assistance system | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN105551264A (en) | Speed detection method based on license plate characteristic matching | |
| CN103310194A (en) | Method for detecting head and shoulders of pedestrian in video based on overhead pixel gradient direction | |
| EP2813973B1 (en) | Method and system for processing video image | |
| CN103971524B (en) | A kind of traffic flow detection method based on machine vision | |
| CN105741324A (en) | Moving object detection identification and tracking method on moving platform | |
| Kim et al. | Autonomous vehicle detection system using visible and infrared camera | |
| CN107480585A (en) | Object detection method based on DPM algorithms | |
| CN105512618A (en) | Video tracking method | |
| CN107315998A (en) | Vehicle class division method and system based on lane line | |
| CN111079675A (en) | Driving behavior analysis method based on target detection and target tracking | |
| CN114049572A (en) | Detection method for identifying small target |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160316 |
|
| RJ01 | Rejection of invention patent application after publication |