CN110837772A - Automobile headlamp identification method for night traffic monitoring - Google Patents

Automobile headlamp identification method for night traffic monitoring Download PDF

Info

Publication number
CN110837772A
CN110837772A CN201910914962.0A CN201910914962A CN110837772A CN 110837772 A CN110837772 A CN 110837772A CN 201910914962 A CN201910914962 A CN 201910914962A CN 110837772 A CN110837772 A CN 110837772A
Authority
CN
China
Prior art keywords
bright
headlamps
headlamp
traffic monitoring
night
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.)
Withdrawn
Application number
CN201910914962.0A
Other languages
Chinese (zh)
Inventor
刘传洋
刘景景
孙晖
汪贤才
孙佐
胡昔兵
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chizhou University
Original Assignee
Chizhou University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chizhou University filed Critical Chizhou University
Priority to CN201910914962.0A priority Critical patent/CN110837772A/en
Publication of CN110837772A publication Critical patent/CN110837772A/en
Withdrawn legal-status Critical Current

Links

Images

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
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/08—Detecting or categorising vehicles

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides an automobile headlamp identification method for night traffic monitoring, and relates to the technical field of automobile headlamp identification. The invention relates to a method for identifying automobile headlamps for night traffic monitoring, which is characterized in that image segmentation and extraction are carried out by a multilevel threshold method, and bright interested objects can be properly extracted from the brightest threshold image and decomposed into a group of uniform threshold images; the method has the advantages that the atmosphere scattering attenuation model is adopted to identify the headlamps, the movement tracks of the headlamps are constructed according to the spatial information and the time information, and the shape characteristics of the headlamps are identified through the structure tensor, so that the headlamps and the road surface reflected light are distinguished, the headlamps in the traffic monitoring video can be effectively identified, and the identification rate of the headlamps is effectively improved.

Description

Automobile headlamp identification method for night traffic monitoring
Technical Field
The invention relates to the technical field of automobile headlamp identification, in particular to an automobile headlamp identification method for night traffic monitoring.
Background
The vigorous development of the automobile industry and the increasing number of automobiles bring huge traffic problems, and a great number of traffic accidents are caused, so that great threats are caused to the safety and property of people, and some of the traffic accidents are caused by light problems, so that the problem of driving safety becomes the focus of attention of people more and more. In night traffic videography, automobile headlamps are a remarkable feature for vehicle detection. The main function of the automobile headlamp is to illuminate the road surface, so that a driver can clearly see the road surface condition in front, and discomfort such as glare and dizziness of oncoming pedestrians or automobile drivers is prevented. Headlamp identification is usually performed using spatial information, such as shape, intensity distribution, etc., which varies under different traffic video conditions. The front surface and the inclined surface of the camera are provided with different shot video information; videos shot by the road with and without the lamp are also different. If the above factors are not considered, normal detection of the vehicle headlamps cannot be realized.
Patent 201710081365.5 discloses a method for identifying and tracking a tail lamp of a headlight of an automobile under visual perception, which comprises the following steps: video image acquisition, local space vehicle lamp detection, global space vehicle detection, vehicle lamp track real-time tracking and matrix type LED particle control. Based on an intelligent image recognition and tracking technology, a camera is used for collecting and storing targets at the head and the tail of the automobile in front of the automobile, and the detection of the direction, the speed and the position motion parameters of the front automobile to the automobile is realized by combining a target detection and tracking technology and a feedback correction model; and the visual image perception parameters are used as control input quantity, and the brightness of single or group lamps in the matrix type LED headlamp is adaptively controlled through an image visual field and LED illumination area mapping model.
Patent 201510179612.6 discloses an automatic recognition method for the body color of a motor vehicle in a vehicle video, which obtains a body preliminary region and a vehicle lamp searching region, and obtains the color and confidence of each pixel point in the two regions; preliminarily identifying the color of the vehicle body by using the colors of pixel points in a group of vehicle body identification units with the serial numbers of 12-19; preliminarily identifying the boundary of the vehicle body by using the colors of pixel points in all vehicle body identification units with the serial numbers of 0-11; searching the position of the car lamp by using the color of a pixel point in a group of car lamp searching units with the serial numbers of 15-19; determining an effective area in the initial area of the vehicle body, and extracting five colors to be used as initial color selection; and finally, judging the color according to the initially selected color, judging the color of black, white and gray when the color is determined not to be the color, and finally automatically identifying to obtain the color of the vehicle body.
The invention can realize the multi-target vehicle stable tracking under the complex light condition, adjust the automobile headlamp in real time based on the tracking result, and can be used for the intelligent control service of the headlamp in the auxiliary driving of the automobile; the method has the advantages that various vehicle body colors can be accurately identified under the influence of glass, shadow and reflection light in the daytime, and the identification precision is high. However, since the headlamps have a similarity to the road surface reflected light, the night vehicle detection based on the image processing is disturbed by the road surface light reflection, and the headlamp recognition cannot be performed efficiently.
Disclosure of Invention
Technical problem to be solved
Aiming at the defect that the detection of the vehicle at night is interfered by the reflection of road light and the headlamp cannot be effectively identified, the invention provides the automobile headlamp identification method for night traffic monitoring, which can effectively identify the headlamp in a traffic monitoring video.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme:
a method for identifying automobile headlamps for night traffic monitoring comprises the following steps:
s1, image segmentation: segmenting night video scenes, distinguishing vehicle headlamps and road reflections through color characteristics, and displaying video image regions and backgroundsThe regions are distinguished, the result of the image segmentation is denoted by B,
Figure BDA0002215823790000021
SFto separate the coefficients, VBCAs a variation in illumination intensity, TSIs a separation threshold;
s2, bright object extraction: performing spatial clustering grouping on the images segmented in the step S1, CiBeing a bright object, CCkGroup number of bright objects, CCkIs CiSet of (2), Ncc(CCk) The number of bright objects; bright object CiThe boundary coordinates of the top, bottom, left and right sides of the frame are t (C)i)、b(Ci)、l(Ci)、r(Ci) (ii) a Bright object CiHas a width to height ratio of W (C)i) And H (C)i) (ii) a Two bright objects at a horizontal distance Dh(Ci,Cj)＝max[t(Ci),t(Cj)]-min[b(Ci),b(Cj)](ii) a Vertical distance D of two bright objectsv(Ci,Cj)＝max[l(Ci),l(Cj)]-min[r(Ci),r(Cj)](ii) a Two bright objects in horizontal or vertical direction, if there is overlap, DhOr DvWill be a negative value; pv(Ci,Cj) Indicating an overlap between two bright vertical projections, Pv(Ci,Cj)＝-Dv(Ci,Cj)/min[H(Ci),H(Cj)]，
If the bright object satisfies Dh(Ci,Cj)＜Td×max(H(Ci),H(Cj))，Pv(Ci,Cj)＞Tp，min(H(Ci),H(Cj))/max(H(Ci),H(Cj))＞ThThe bright objects have the same headlamp, tail lamp and street lamp groups; t isd、Tp、ThIs a set threshold value;
s3, vehicle lamp identification: for the grayscale image I pixel (x, y) of the bright object of step S3,
Figure BDA0002215823790000031
respectively an internal adjacent region and an external adjacent region,
Figure BDA0002215823790000032
r is an internal neighborhood
Figure BDA0002215823790000034
Radius of, outside neighbourhood
Figure BDA0002215823790000035
Has a radius of 2 r; according to the law of attenuation by atmospheric scattering,
Figure BDA0002215823790000036
represents scattering of the same light sourceHas a minimum intensity pixel of
Figure BDA0002215823790000038
The maximum intensity pixel of
Figure BDA0002215823790000039
As
Figure BDA00022158237900000310
Is scattered to obtain
Figure BDA00022158237900000311
The scattering coefficient gamma (x, y),
Figure BDA0002215823790000041
wherein
Figure BDA0002215823790000042
Is composed ofAnd
Figure BDA0002215823790000044
the distance between them;
order to
Figure BDA0002215823790000045
Has a minimum intensity pixel of
Figure BDA0002215823790000047
The maximum intensity pixel of
Figure BDA0002215823790000048
Figure BDA0002215823790000049
Is composed ofIs scattered, and
Figure BDA00022158237900000411
and
Figure BDA00022158237900000412
having a scattering coefficient gamma (x, y), resulting in a reflection intensity mapping coefficient RI,
Figure BDA00022158237900000413
s4, headlamp detection: establishing a 3-by-3 structure tensor matrix composed of gradients in three directions (x, y, t), and forming a structure tensor by a gradient covariance matrix (M)cov) Feature vector
Figure BDA00022158237900000414
The form of the characteristic value (lambda) reflects the orientation and shape characteristics of the three-dimensional object, Ix＝Ci*Gx，Iy＝Ci*Gy，It＝Ci*Gt，Ix、Iy、ItThe gradient in the x, y and t directions, Gx、Gy、GtSobel operators in x, y and t directions respectively, convolution operator and covariance matrix
Figure BDA00022158237900000415
From | λ I-McovSolving the eigenvalue (lambda) to obtain | ═ 0
Figure BDA00022158237900000416
The eigenvalues depend on the gradient components;
s5, classification training: and (4) training the shape characteristics of the headlamp obtained in the step (S4) by utilizing a neural network to construct a classification model, wherein the neural network consists of an input layer, a hidden layer and an output layer, three input characteristics come from characteristic values, and one output is used as a recognition result.
According to an embodiment of the present invention, the step S1 separation threshold is 0.85.
According to an embodiment of the invention, the threshold T in step S2d、Tp、ThThe values are 2.5, 0.85 and 0.75 respectively.
According to an embodiment of the present invention, the bright object enclosing bounding box of step S2 is a horizontal rectangle satisfying W (CC)k)/H(CCk)≥τrWherein the threshold value τ isrThe value is 2; the number of bright objects satisfies 2 ≤ Ncc(CCk) Less than or equal to 4, the bright objects are uniformly distributed to meet the requirement
Figure BDA0002215823790000051
Wherein the threshold value taua1、τa20.4 and 2, respectively.
According to an embodiment of the present invention, the step S3 is executed by
Figure BDA0002215823790000052
And
Figure BDA0002215823790000053
the reflection intensity mapping coefficient RI is lower in the area around the headlamp and higher in the reflection area of the headlamp, and the bright object is distinguished as a headlamp or a lamplight reflection through the reflection intensity mapping coefficient RI.
According to an embodiment of the present invention, the step S4 is λ due to the shape stability of the headlamps in the time direction3The value is low, close to 0, and for the other two directions their gradient components are the same because of the circular shape, λ1And λ2Is close to when λ is satisfied1≈λ2Much greater than 0, lambda3And 0, the detected bright object is a headlamp.
(III) advantageous effects
The invention has the beneficial effects that: a method for recognizing automobile headlamps for monitoring traffic at night comprises image segmentation, bright object extraction, automobile lamp recognition, headlamp detection and classification training, wherein the image segmentation and extraction are carried out through a multi-level threshold method, and bright interested objects can be properly extracted from the brightest threshold images and decomposed into a group of uniform threshold images; the method has the advantages that the atmosphere scattering attenuation model is adopted to identify the headlamps, the movement tracks of the headlamps are constructed according to the spatial information and the time information, and the shape characteristics of the headlamps are identified through the structure tensor, so that the headlamps and the road surface reflected light are distinguished, the headlamps in the traffic monitoring video can be effectively identified, and the identification rate of the headlamps is effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a tensor vector diagram of a structure of a motion trail of a headlamp;
FIG. 3 is a normalized contrast operating characteristic curve for three methods.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, a method for identifying headlamps of an automobile for night traffic monitoring includes the following steps:
s1, image segmentation: dividing night video scenes, distinguishing vehicle headlamps and road reflections through color characteristics, distinguishing a video image area from a background area, representing the result of image division by B,
Figure BDA0002215823790000061
SFto separate the coefficients, VBCAs a variation in illumination intensity, TSTo separate the threshold, TSMay be 0.85.
S2, bright object extraction: performing spatial clustering grouping on the images segmented in the step S1, CiBeing a bright object, CCkGroup number of bright objects, CCkIs CiSet of (2), Ncc(CCk) The number of bright objects; bright object CiThe boundary coordinates of the top, bottom, left and right sides of the frame are t (C)i)、b(Ci)、l(Ci)、r(Ci) (ii) a Bright object CiHas a width to height ratio of W (C)i) And H (C)i) (ii) a Two bright objects at a horizontal distance Dh(Ci,Cj)＝max[t(Ci),t(Cj)]-min[b(Ci),b(Cj)](ii) a Vertical distance D of two bright objectsv(Ci,Cj)＝max[l(Ci),l(Cj)]-min[r(Ci),r(Cj)](ii) a Two bright objects in horizontal or vertical direction, if there is overlap, DhOr DvWill be a negative value; pv(Ci,Cj) Indicating an overlap between two bright vertical projections, Pv(Ci,Cj)＝-Dv(Ci,Cj)/min[H(Ci),H(Cj)]，
If the bright object satisfies Dh(Ci,Cj)＜Td×max(H(Ci),H(Cj))，Pv(Ci,Cj)＞Tp，min(H(Ci),H(Cj))/max(H(Ci),H(Cj))＞ThThe bright objects have the same headlamp, tail lamp and street lamp groups; t isd、Tp、ThIs a set threshold value, wherein the threshold value Td、Tp、ThCan be respectively 2.5, 0.85 and 0.75; the closed boundary frame of the bright object is a horizontal rectangle satisfying W (CC)k)/H(CCk)≥τrWherein the threshold value τ isrThe value is 2; the number of bright objects satisfies 2 ≤ Ncc(CCk) Less than or equal to 4, the bright objects are uniformly distributed to meet the requirement
Figure BDA0002215823790000071
Wherein the threshold value taua1、τa20.4 and 2, respectively.
S3, vehicle lamp identification: for the grayscale image I pixel (x, y) of the bright object of step S3,
Figure BDA0002215823790000072
respectively an internal adjacent region and an external adjacent region,
Figure BDA0002215823790000073
Figure BDA0002215823790000074
r is an internal neighborhood
Figure BDA0002215823790000075
Radius of, outside neighbourhoodHas a radius of 2 r; according to the law of attenuation by atmospheric scattering,
Figure BDA0002215823790000077
represents scattering of the same light source
Figure BDA0002215823790000078
Has a minimum intensity pixel of
Figure BDA0002215823790000079
Figure BDA00022158237900000710
The maximum intensity pixel of
Figure BDA00022158237900000711
Figure BDA00022158237900000712
As
Figure BDA00022158237900000713
Is scattered to obtain
Figure BDA00022158237900000714
The scattering coefficient gamma (x, y),
Figure BDA0002215823790000081
wherein
Figure BDA0002215823790000082
Is composed of
Figure BDA0002215823790000083
And
Figure BDA0002215823790000084
the distance between them;
order toHas a minimum intensity pixel of
Figure BDA0002215823790000086
Figure BDA0002215823790000087
The maximum intensity pixel of
Figure BDA0002215823790000089
Is composed of
Figure BDA00022158237900000810
Is scattered, and
Figure BDA00022158237900000811
and
Figure BDA00022158237900000812
having a scattering coefficient gamma (x, y), resulting in a reflection intensity mapping coefficient RI,
Figure BDA00022158237900000813
due to the fact that
Figure BDA00022158237900000814
And
Figure BDA00022158237900000815
the reflection intensity mapping coefficient RI is lower in the area around the headlamp and higher in the reflection area of the headlamp, and the bright object is distinguished as a headlamp or a lamplight reflection through the reflection intensity mapping coefficient RI.
S4, headlamp detection: establishing a 3-by-3 structure tensor matrix composed of gradients in three directions (x, y, t), and forming a structure tensor by a gradient covariance matrix (M)cov) Feature vector
Figure BDA00022158237900000816
The form of the characteristic value (lambda) reflects the orientation and shape characteristics of the three-dimensional object, Ix＝Ci*Gx，Iy＝Ci*Gy，It＝Ci*Gt，Ix、Iy、ItThe gradient in the x, y and t directions, Gx、Gy、GtSobel operators in x, y and t directions respectively, convolution operator and covariance matrixFrom | λ I-McovSolving the eigenvalue (lambda) to obtain | ═ 0
Figure BDA00022158237900000818
The characteristic value depends on the gradient component, lambda being a stable shape of the headlight in the time direction3The value is lower, close to 0. For the other two directions, their gradient components are the same because of the circular shape, λ1And λ2Is close to when λ is satisfied1≈λ2Much greater than 0, lambda3And 0, the detected bright object is a headlamp. The tensor vector diagram of the structure of the motion trail of the headlamp is shown in fig. 2.
S5, classification training: and (4) training the shape characteristics of the headlamp obtained in the step (S4) by utilizing a neural network to construct a classification model, wherein the neural network consists of an input layer, a hidden layer and an output layer, three input characteristics come from characteristic values, and one output is used as a recognition result.
The existing headlamp identification method is generally realized by adopting a rule-based method and a physical model-based method. In order to analyze the performance of the method, a hardware environment under a Windows 10 system adopts an i5-7300 HQ processor and a CPU2.6GHz, 8GB memory, a classifier model is designed in MATLAB software to carry out a comparison experiment, a hidden layer and an output layer adopt tangent square transfer functions, and 10 hidden layers are used for optimizing the performance of the classifier. The surveillance video used a resolution of 960 x 540 and 25 FPSs. Fig. 3 shows normalized contrast operating characteristics for three methods. It can be seen from the figure that the method of the present invention has better recognition capability for the headlamps.
In summary, the embodiment of the present invention provides an automobile headlamp identification method for night traffic monitoring, which includes image segmentation, bright object extraction, headlamp identification, headlamp detection, and classification training, where the image segmentation and extraction are performed by a multi-level thresholding method, and a bright object of interest can be appropriately extracted from the brightest generated threshold image and decomposed into a set of uniform threshold images; the method has the advantages that the atmosphere scattering attenuation model is adopted to identify the headlamps, the movement tracks of the headlamps are constructed according to the spatial information and the time information, and the shape characteristics of the headlamps are identified through the structure tensor, so that the headlamps and the road surface reflected light are distinguished, the headlamps in the traffic monitoring video can be effectively identified, and the identification rate of the headlamps is effectively improved.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (6)

1. A vehicle headlamp identification method for night traffic monitoring is characterized in that: the method comprises the following steps:
s1, image segmentation: dividing night video scenes, distinguishing vehicle headlamps and road reflections through color characteristics, distinguishing a video image area from a background area, representing the result of image division by B,SFto separate the coefficients, VBCAs a variation in illumination intensity, TSIs a separation threshold;
s2, bright object extraction: performing spatial clustering grouping on the images segmented in the step S1, CiBeing a bright object, CCkGroup number of bright objects, CCkIs CiSet of (2), Ncc(CCk) The number of bright objects; bright object CiThe boundary coordinates of the top, bottom, left and right sides of the frame are t (C)i)、b(Ci)、l(Ci)、r(Ci) (ii) a Bright object CiHas a width to height ratio of W (C)i) And H (C)i) (ii) a Two bright objects at a horizontal distance Dh(Ci,Cj)＝max[t(Ci),t(Cj)]-min[b(Ci),b(Cj)](ii) a Vertical distance D of two bright objectsv(Ci,Cj)＝max[l(Ci),l(Cj)]-min[r(Ci),r(Cj)](ii) a Two bright objects in horizontal or vertical direction, if there is overlap, DhOr DvWill be a negative value; pv(Ci,Cj) Indicating an overlap between two bright vertical projections, Pv(Ci,Cj)＝-Dv(Ci,Cj)/min[H(Ci),H(Cj)]，
If the bright object satisfies Dh(Ci,Cj)＜Td×max(H(Ci),H(Cj))，Pv(Ci,Cj)＞Tp，min(H(Ci),H(Cj))/max(H(Ci),H(Cj))＞ThThe bright objects have the same headlamp, tail lamp and street lamp groups; t isd、Tp、ThIs a set threshold value;
s3, vehicle lamp identification: for the grayscale image I pixel (x, y) of the bright object of step S3,
Figure FDA0002215823780000012
respectively an internal adjacent region and an external adjacent region,
Figure FDA0002215823780000022
r is an internal neighborhood
Figure FDA0002215823780000023
Radius of, outside neighbourhood
Figure FDA0002215823780000024
Has a radius of 2 r; according to the law of attenuation by atmospheric scattering,
Figure FDA0002215823780000025
represents scattering of the same light sourceHas a minimum intensity pixel of
Figure FDA0002215823780000027
The maximum intensity pixel ofAs
Figure FDA0002215823780000029
Is scattered to obtain
Figure FDA00022158237800000210
The scattering coefficient gamma (x, y),wherein
Figure FDA00022158237800000212
Is composed of
Figure FDA00022158237800000213
And
Figure FDA00022158237800000214
the distance between them;
order to
Figure FDA00022158237800000215
Has a minimum intensity pixel of
Figure FDA00022158237800000216
The maximum intensity pixel of
Figure FDA00022158237800000217
Figure FDA00022158237800000218
Is composed of
Figure FDA00022158237800000219
Is scattered, and
Figure FDA00022158237800000220
and
Figure FDA00022158237800000221
having a scattering coefficient gamma (x, y), resulting in a reflection intensity mapping coefficient RI,
Figure FDA00022158237800000222
s4, headlamp detection: establishing a 3-by-3 structure tensor matrix composed of gradients in three directions (x, y, t), and forming a structure tensor by a gradient covariance matrix (M)cov) Feature vector
Figure FDA00022158237800000223
The form of the characteristic value (lambda) reflects the orientation and shape characteristics of the three-dimensional object, Ix＝Ci*Gx，Iy＝Ci*Gy，It＝Ci*Gt，Ix、Iy、ItThe gradient in the x, y and t directions, Gx、Gy、GtSobel operators in x, y and t directions respectively, convolution operator and covariance matrixFrom | λ I-McovSolving the eigenvalue (lambda) to obtain | ═ 0
Figure FDA00022158237800000225
The eigenvalues depend on the gradient components;
s5, classification training: and (4) training the shape characteristics of the headlamp obtained in the step (S4) by utilizing a neural network to construct a classification model, wherein the neural network consists of an input layer, a hidden layer and an output layer, three input characteristics come from characteristic values, and one output is used as a recognition result.
2. The method for identifying headlamps of an automobile for night traffic monitoring as claimed in claim 1, wherein said step S1 is performed with a separation threshold of 0.85.
3. The method for recognizing headlights of vehicle for night traffic monitoring as claimed in claim 1, wherein the threshold value T in step S2 is setd、Tp、ThThe values are 2.5, 0.85 and 0.75 respectively.
4. The method for recognizing automotive headlamps for night traffic monitoring as claimed in claim 3, wherein said bright object enclosing bounding box of step S2 is a horizontal rectangle satisfying W (CC)k)/H(CCk)≥τrWherein the threshold value τ isrThe value is 2; the number of bright objects satisfies 2 ≤ Ncc(CCk) Less than or equal to 4, the bright objects are uniformly distributed to meet the requirement
Figure FDA0002215823780000031
Wherein the threshold value taua1、τa20.4 and 2, respectively.
5. The method for recognizing headlights of vehicle for night-time traffic monitoring as claimed in claim 1, wherein the step S3 is performed in order to recognize the headlights
Figure FDA0002215823780000032
Andthe reflection intensity mapping coefficient RI is lower in the area around the headlamp and higher in the reflection area of the headlamp, and the bright object is distinguished as a headlamp or a lamplight reflection through the reflection intensity mapping coefficient RI.
6. The method for recognizing automotive headlamps for night traffic monitoring as set forth in claim 2, wherein said step S4 is λ d due to the shape stability of headlamps in the time direction3The value is low, close to 0, and for the other two directions their gradient components are the same because of the circular shape, λ1And λ2Is close to when λ is satisfied1≈λ2Much greater than 0, lambda3And 0, the detected bright object is a headlamp.
CN201910914962.0A 2019-09-26 2019-09-26 Automobile headlamp identification method for night traffic monitoring Withdrawn CN110837772A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910914962.0A CN110837772A (en) 2019-09-26 2019-09-26 Automobile headlamp identification method for night traffic monitoring

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910914962.0A CN110837772A (en) 2019-09-26 2019-09-26 Automobile headlamp identification method for night traffic monitoring

Publications (1)

Publication Number Publication Date
CN110837772A true CN110837772A (en) 2020-02-25

Family

ID=69575263

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910914962.0A Withdrawn CN110837772A (en) 2019-09-26 2019-09-26 Automobile headlamp identification method for night traffic monitoring

Country Status (1)

Country Link
CN (1) CN110837772A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112101230A (en) * 2020-09-16 2020-12-18 招商局重庆公路工程检测中心有限公司 Method and system for detecting starting of headlamps of vehicles passing through highway tunnel

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112101230A (en) * 2020-09-16 2020-12-18 招商局重庆公路工程检测中心有限公司 Method and system for detecting starting of headlamps of vehicles passing through highway tunnel
CN112101230B (en) * 2020-09-16 2024-05-14 招商局重庆公路工程检测中心有限公司 Method and system for detecting opening of head lamp of road tunnel passing vehicle

Similar Documents

Publication Publication Date Title
CN107766821B (en) Method and system for detecting and tracking full-time vehicle in video based on Kalman filtering and deep learning
CN105981042B (en) Vehicle detection system and method
US8199198B2 (en) Bright spot detection and classification method for a vehicular night-time video imaging system
US7724962B2 (en) Context adaptive approach in vehicle detection under various visibility conditions
US7566851B2 (en) Headlight, taillight and streetlight detection
US8503725B2 (en) Vehicle tracking system and tracking method thereof
US8064643B2 (en) Detecting and recognizing traffic signs
TWI607901B (en) Image inpainting system area and method using the same
Alcantarilla et al. Night time vehicle detection for driving assistance lightbeam controller
Kosaka et al. Vision-based nighttime vehicle detection using CenSurE and SVM
Chen et al. Nighttime vehicle detection for driver assistance and autonomous vehicles
KR101511853B1 (en) Night-time vehicle detection and positioning system and method using multi-exposure single camera
KR20210006276A (en) Image processing method for flicker mitigation
CN106934808A (en) A kind of automobile headlamp taillight recognition and tracking method under visually-perceptible
Wu et al. A Real‐Time Embedded Blind Spot Safety Assistance System
CN107622494B (en) Night vehicle detection and tracking method facing traffic video
Lin et al. Adaptive IPM-based lane filtering for night forward vehicle detection
US20190318490A1 (en) Distance estimation of vehicle headlights
CN111046741A (en) Method and device for identifying lane line
Sakagawa et al. Vision based nighttime vehicle detection using adaptive threshold and multi-class classification
CN110837772A (en) Automobile headlamp identification method for night traffic monitoring
Chen et al. Robust rear light status recognition using symmetrical surfs
Huang et al. Nighttime vehicle detection and tracking base on spatiotemporal analysis using RCCC sensor
CN113581059A (en) Light adjusting method and related device
CN111062231B (en) Vehicle detection method, night vehicle detection method based on light intensity dynamic and system thereof

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20200225