CN115984828B - High beam opening detection method, device and equipment based on geometric feature descriptors - Google Patents
High beam opening detection method, device and equipment based on geometric feature descriptors Download PDFInfo
- Publication number
- CN115984828B CN115984828B CN202310267314.7A CN202310267314A CN115984828B CN 115984828 B CN115984828 B CN 115984828B CN 202310267314 A CN202310267314 A CN 202310267314A CN 115984828 B CN115984828 B CN 115984828B
- Authority
- CN
- China
- Prior art keywords
- geometric feature
- car light
- preset
- image
- feature descriptors
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention provides a high beam opening detection method, a device and equipment based on a geometric feature descriptor, wherein the method comprises the following steps: acquiring at least one vehicle running photo shot by a camera, and extracting a car light area image of the vehicle from the vehicle running photo; calculating a geometric feature descriptor corresponding to the vehicle lamp area image according to a preset first geometric feature calculation formula; inputting geometric feature descriptors corresponding to the car light region images into a pre-trained car light classification model so as to carry out far-near light classification and identification on the car light region images; and judging whether the vehicle starts a high beam or not according to the classification and identification result of the vehicle lamp classification model. According to the invention, the high beam opening behavior is judged by the geometric feature descriptors corresponding to the car light area images and the pre-trained car light classification model, so that the judgment reliability of the high beam opening behavior is greatly improved.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to a method, an apparatus, and a device for detecting turn-on of a high beam based on a geometric feature descriptor.
Background
Generally, under the condition of severe illumination, a driver can be allowed to turn on a high beam for illumination so as to ensure driving safety. However, in real life, some drivers have a phenomenon of using a high beam maliciously, for example, in a road section with good lighting conditions such as urban areas, there are still many situations that vehicles turn on the high beam, and this situation easily causes visual interference to other drivers, resulting in unnecessary safety accidents. So that the supervision of the turn-on behavior of the high beam needs to be enhanced.
The prior far-reaching headlamp opening behavior recognition scheme based on image recognition is commonly implemented, and the main principle is as follows: the method comprises the steps of shooting an image of the head part of a vehicle, determining a lamplight area based on a target positioning algorithm, and judging whether the vehicle starts a high beam or not by combining a high beam judging strategy. However, the currently set far-reaching headlamp judging strategy is mainly implemented based on the brightness of the lamplight area, namely, when the brightness of the lamplight area is larger than the brightness threshold value, the far-reaching headlamp is judged to be turned on, but is affected by the interference of the uneven brightness of the far-reaching headlamp of different vehicles, in order to avoid erroneous penalty, the brightness threshold value is generally set higher, so that the missed judgment of the turn-on behavior of the far-reaching headlamp is easily caused, and therefore, the currently adopted far-reaching headlamp turn-on behavior recognition scheme based on image recognition obviously has the defect of insufficient reliability.
Disclosure of Invention
Based on the above, the present invention aims to provide a method, a device and a device for detecting the opening of a high beam based on a geometric feature descriptor, so as to solve at least one technical problem in the background art.
According to the embodiment of the invention, the method for detecting the opening of the high beam based on the geometric feature descriptors comprises the following steps:
acquiring at least one vehicle running photo shot by a camera, and extracting a car light area image of a vehicle from the vehicle running photo;
calculating a geometric feature descriptor corresponding to the car light area image according to a preset first geometric feature calculation formula;
inputting geometric feature descriptors corresponding to the car light region images into a pre-trained car light classification model so as to carry out far-near light classification identification on the car light region images;
and judging whether the vehicle starts a high beam or not according to the classification and identification result of the vehicle lamp classification model.
In addition, the method for detecting the opening of the high beam based on the geometric feature descriptors according to the embodiment of the invention may further have the following additional technical features:
further, the geometric feature descriptor includes a high-order moment, and the step of calculating the geometric feature descriptor corresponding to the vehicle lamp area image according to a preset first geometric feature calculation formula includes:
According to a preset high-order moment calculation formula, calculating a high-order moment corresponding to the car light area image, wherein the high-order moment calculation formula is as follows:
in the method, in the process of the invention,m p,q (f) Is a high order moment in two dimensions, whereinp、qIs a non-negative integer number of the number,pand (3) withqThe sum ism p,q (f) Is used for the number of the order of (c),f(x,y) The method is that the image is at the pixel pointx,y) Is a gray value of (a).
Further, the training process of the car light classification model is as follows:
acquiring historical car light area images of known far and near light lamp classification results, and calculating geometric feature descriptors of each historical car light area image by adopting the first geometric feature calculation formula;
marking each historical car light area image according to the geometric feature descriptors of the historical car light area images and the far and near light classification results thereof, so as to construct a model training set;
training a preset network model through the model training set to obtain the car lamp classification model capable of carrying out far and near light lamp classification and identification according to the geometric feature descriptors.
Further, the step of extracting the lamp area image of the vehicle from the vehicle running photograph includes:
graying the vehicle running photo according to a preset image graying algorithm to obtain a graying photo, binarizing the graying photo according to a gray threshold to obtain a binarizing photo, and intercepting all connected area images with gray values larger than the gray threshold from the binarizing photo;
According to a preset second geometric feature calculation formula, calculating geometric feature descriptors of each communication area image, and screening the communication area images according to the geometric feature descriptors of each communication area image to screen out target communication areas with the geometric feature descriptors meeting preset conditions so as to extract the car light area images.
Further, the image graying algorithm is as follows:
f(x,y)=0.3*R(x,y)+0.59*G(x,y)+0.11*B(x,y)
in the method, in the process of the invention,R(x,y)、G(x,y) and B(x,y) Pixel points representing the vehicle running pictures respectivelyx,y) Is set, the gray threshold is 220.
Further, according to a preset second geometric feature calculation formula, calculating a geometric feature descriptor of each communication area image, and screening the communication area images according to the geometric feature descriptor of each communication area image to screen out a target communication area with the geometric feature descriptor meeting a preset condition, so as to extract the car light area image, wherein the step of extracting the car light area image comprises the following steps:
calculating the area, the aspect ratio and the roundness of each communication area image according to a preset second geometric feature calculation formula, wherein the aspect ratio is the ratio of the width to the height of the communication area image;
And removing all the communication area images with the area smaller than an area threshold value, the aspect ratio smaller than an aspect ratio threshold value and/or the roundness smaller than a roundness threshold value from the communication area images, so as to screen out target communication areas with geometric feature descriptors meeting preset conditions, and extract the car light area images.
Further, after the step of calculating the geometric feature descriptor of each connected region image according to the preset second geometric feature calculation formula and screening the connected region image according to the geometric feature descriptor of each connected region image to screen out the target connected region whose geometric feature descriptor meets the preset condition, the method further includes:
judging whether the target communication areas meeting the preset conditions are two;
if the number of the target communication areas is not two, the area threshold value, the aspect ratio threshold value and/or the roundness threshold value are/is adjusted until the number of the target communication areas meeting the preset condition is two;
if the number of the geometric feature descriptors is two, according to a preset first geometric feature calculation formula, the step of calculating the geometric feature descriptors corresponding to the car light area image comprises the following steps:
respectively calculating geometric feature descriptors corresponding to the two car light area images according to a preset first geometric feature calculation formula, and judging whether the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value;
And if the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value, executing the step of inputting the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to carry out far and near light classification recognition on the car light area images.
According to an embodiment of the invention, a high beam opening detection device comprises:
the image extraction module is used for acquiring at least one vehicle running photo shot by the camera and extracting a car light area image of the vehicle from the vehicle running photo;
the feature calculation module is used for calculating a geometric feature descriptor corresponding to the car light area image according to a preset first geometric feature calculation formula;
the classification and identification module is used for inputting geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to perform far-near light classification and identification on the car light area images;
and the result output module is used for judging whether the vehicle starts a high beam or not according to the classification recognition result of the car light classification model.
The invention also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, implements the above-mentioned high beam on detection method based on the geometric feature descriptors.
The invention also provides a high beam opening detection device which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the high beam opening detection method based on the geometric feature descriptor is realized when the processor executes the program.
Compared with the prior art: the high beam opening behavior is distinguished through the geometric feature descriptors corresponding to the vehicle lamp area images, which is completely different from the traditional method of adopting brightness threshold classification, so that the interference influence of uneven brightness of high beam lights of different vehicles can be effectively avoided, and the geometric feature descriptors of the image areas corresponding to any vehicle high beam lights and low beam lights have obvious distinction, so that the high beam opening behavior recognition precision and reliability can be higher, meanwhile, when the high beam opening behavior is distinguished based on the geometric feature descriptors, the high beam opening behavior is distinguished, the method of traditional threshold classification is not adopted, but is realized through a pre-trained vehicle lamp classification model, and the vehicle lamp classification model is obtained by deep learning based on the geometric feature descriptors of the image areas corresponding to a large number of vehicle high beam lights and low beam lights, so that the output result is more accurate, and the judgment reliability of the high beam opening behavior is further improved.
Drawings
Fig. 1 is a flowchart of a high beam on detection method based on a geometric feature descriptor in a first embodiment of the present invention;
fig. 2 is a schematic structural diagram of a high beam on detecting device according to a fourth embodiment of the present invention;
fig. 3 is a schematic structural diagram of a high beam on detection apparatus according to a fifth embodiment of the present invention.
The following detailed description will further illustrate the invention with reference to the above-described drawings.
Detailed Description
In order that the invention may be readily understood, a more complete description of the invention will be rendered by reference to the appended drawings. Several embodiments of the invention are presented in the figures. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
It will be understood that when an element is referred to as being "mounted" on another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. The terms "vertical," "horizontal," "left," "right," and the like are used herein for illustrative purposes only.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
Example 1
Referring to fig. 1, a method for detecting turn-on of a high beam based on a geometric feature descriptor in a first embodiment of the present invention is shown, and the method specifically includes steps S01-S04.
Step S01, at least one vehicle running photo shot by a camera is acquired, and a car light area image of the vehicle is extracted from the vehicle running photo.
In a specific implementation, the step of extracting the lamp area image of the vehicle from the vehicle running photograph may specifically include:
graying the vehicle running photo according to a preset image graying algorithm to obtain a graying photo, binarizing the graying photo according to a gray threshold to obtain a binarizing photo, and intercepting all connected area images with gray values larger than the gray threshold from the binarizing photo;
According to a preset second geometric feature calculation formula, calculating geometric feature descriptors of each communication area image, and screening the communication area images according to the geometric feature descriptors of each communication area image to screen out target communication areas with the geometric feature descriptors meeting preset conditions so as to extract the car light area images.
In other words, the vehicle running photograph is firstly subjected to grey scale and binary processing in sequence, and all the connected region images with grey scale values larger than the grey scale threshold value are cut out from the binary photograph, so that all the connected regions with relatively high brightness are selected. However, since the vehicle turns on the high beam and low beam, a reflected bright area is generated on the road surface in front, and the bright area is also intercepted during the above-mentioned communication area image intercepting process; or when the vehicle turns on the far and near lights, the fog lamp is turned on, and the fog lamp is generally positioned below the far and near lights, so that the fog lamp area can be intercepted in the process of intercepting the image of the communication area; or when the vehicle turns on the turn signal lamp after turning on the high beam and the low beam, some vehicles are also provided with the turn signal lamp on the outer rearview mirror, so that the turn signal lamp area on the outer rearview mirror can be intercepted in the communication area image intercepting process.
The conventional high beam turning-on behavior recognition scheme based on image recognition can not effectively remove the interference due to judgment based on brightness, and the high beam turning-on behavior is continuously judged under the condition that the interference factor exists, so that the reliability problem is obvious. Therefore, in order to solve the above-mentioned problem, the present embodiment further removes these interference factors in a specific manner after the above-mentioned connected region images are captured, specifically, calculates the geometric feature descriptors of each connected region image according to the geometric feature calculation formula, and screens the connected region images according to the geometric feature descriptors of each connected region image, so as to screen the target connected region whose geometric feature descriptors meet the preset condition, so as to extract the vehicle lamp region image.
More specifically, the geometric feature descriptors of the connected region images to be calculated include an area, an aspect ratio and a roundness, and then the geometric feature descriptors of each connected region image are calculated according to a preset second geometric feature calculation formula, and the connected region images are screened according to the geometric feature descriptors of each connected region image, so as to screen out a target connected region whose geometric feature descriptors meet a preset condition, so that the step of extracting the vehicle lamp region image specifically includes:
Calculating the area, the aspect ratio and the roundness of each communication area image according to a preset second geometric feature calculation formula, wherein the aspect ratio is the ratio of the width to the height of the communication area image;
and removing all the communication area images with the area smaller than an area threshold value, the aspect ratio smaller than an aspect ratio threshold value and/or the roundness smaller than a roundness threshold value from the communication area images, so as to screen out target communication areas with geometric feature descriptors meeting preset conditions, and extract the car light area images. That is, the second geometric feature calculation formula includes an area calculation formula, an aspect ratio calculation formula, and a roundness calculation formula.
It should be understood that the geometric feature descriptors of the area, the aspect ratio, the roundness and the like presented by the road surface reflection bright light area image (generally long strip), the fog lamp area image (generally small circular area) or the steering lamp area image (generally long strip) are obviously different from the vehicle lamp area image, so that the vehicle lamp area image actually required can be accurately screened out through the geometric feature descriptors.
In specific implementation, the area of the communication region image can be calculated by a communication region area calculation formula, the aspect ratio of the communication region image can be calculated according to the width and the height of the communication region image, and the roundness of the communication region image can be calculated by a region roundness calculation formula. In addition, the vehicle running photograph may be specifically grayed by using a maximum value method, an average value method or a weighted average value method, and in this embodiment, the vehicle running photograph is grayed by using a weighted average value method in consideration of the specificity of the vehicle running photograph and the requirement after the subsequent graying, and specifically, the image graying algorithm is as follows:
f(x,y)=0.3*R(x,y)+0.59*G(x,y)+0.11*B(x,y)
In the method, in the process of the invention,R(x,y)、G(x,y) and B(x,y) Pixel points representing the vehicle running pictures respectivelyx,y) Is set, the gray threshold is 220.
In some alternative embodiments, because the shooting angle of the camera is generally fixed, and the driving range of the vehicle in the shooting area is basically consistent, the image of the fixed position area can be intercepted by adopting a preset standard intercepting frame from the driving photo of the vehicle, or the vehicle object detection can be performed on the driving photo of the vehicle, then the area where the vehicle is located is intercepted, and then the vehicle lamp area image of the vehicle is extracted from the intercepted image, so that the workload and the efficiency of the subsequent image processing can be greatly reduced.
Step S02, calculating a geometric feature descriptor corresponding to the car light area image according to a preset first geometric feature calculation formula.
In a specific implementation, the geometric feature descriptors corresponding to the vehicle lamp area image to be calculated in the step include, but are not limited to, one or more combinations of area, aspect ratio, roundness, perimeter and high-order moment, that is, the first geometric feature calculation formula includes one or more of area calculation formula, aspect ratio calculation formula, roundness calculation formula, perimeter calculation formula and high-order moment calculation formula. It should be noted that, compared with the low beam region image, the high beam region image has larger area, longer circumference, higher roundness (circularity) and larger high-order moment, so that the low beam region image and the high beam region image have obvious geometrical feature descriptor differences, and can be used as factors for distinguishing the classification and identification of the low beam and the high beam. The low beam region image is a car light region image which is correspondingly extracted when the low beam is started, and the high beam region image is a car light region image which is correspondingly extracted when the high beam is started. That is, the training process of the vehicle lamp classification model may specifically be:
Acquiring historical car light area images of known far and near light lamp classification results, and calculating geometric feature descriptors of each historical car light area image by adopting the first geometric feature calculation formula;
marking each historical car light area image according to the geometric feature descriptors of the historical car light area images and the far and near light classification results thereof, so as to construct a model training set;
training a preset network model through the model training set to obtain the car lamp classification model capable of carrying out far and near light lamp classification and identification according to the geometric feature descriptors.
And S03, inputting the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model to perform far-near light classification recognition on the car light area images.
In specific implementation, historical car light area images after high beam and low beam are turned on by different car types can be collected, and the historical car light area images can be extracted from historical car driving photos according to the same method. And then calculating geometric feature descriptors of each historical car light area image by adopting the first geometric feature calculation formula respectively, manually labeling the historical car light area images according to the calculated geometric feature descriptors, wherein labeling contents comprise geometric feature descriptors and far and near light classification results by using a labelImg and other picture labeling tools, so as to construct a training data set, and then performing full-supervision learning on a convolutional neural network model (Convolutional Neural Networks, CNN) by using the training data set, so that the model learns the difference rule among the geometric feature descriptors of far and near lights, and the car light classification model capable of automatically realizing the far and near light classification according to the geometric feature descriptors is obtained.
And step S04, judging whether the vehicle starts a high beam or not according to the classification recognition result of the vehicle lamp classification model.
In summary, in the far-beam opening detection method based on the geometric feature descriptors in the embodiment of the invention, the far-beam opening behavior is distinguished by the geometric feature descriptors corresponding to the images of the car light areas, which is completely different from the traditional method of classifying by adopting the brightness threshold, so that the interference influence of uneven brightness of the far-beam lights of different vehicles can be effectively avoided, and the geometric feature descriptors of the image areas corresponding to any car far-beam lights have obvious distinction, so that the high-beam opening behavior recognition precision and reliability can be higher, and meanwhile, when the high-beam opening behavior is distinguished based on the geometric feature descriptors, the method of traditional threshold classification is not adopted, but the method is realized by a pre-trained car light classification model, and because the car light classification model is obtained by deep learning the geometric feature descriptors of the image areas corresponding to a large number of car far-beam lights, the output result is more accurate, thereby further improving the judgment reliability of the high-beam opening behavior.
Example two
The second embodiment of the present invention also provides a far-reaching headlamp opening detection method based on a geometric feature descriptor, which is different from the far-reaching headlamp opening detection method based on a geometric feature descriptor in the first embodiment in that:
in this embodiment, the step of using a high-order moment corresponding to the lamp area image to perform classification and identification of the high beam specifically, that is, calculating the geometric feature descriptor corresponding to the lamp area image according to a preset first geometric feature calculation formula includes:
according to a preset high-order moment calculation formula, calculating a high-order moment corresponding to the car light area image, wherein the high-order moment calculation formula is as follows:
in the method, in the process of the invention,m p,q (f) Is a high order moment in two dimensions, whereinp、qIs a non-negative integer number of the number,pand (3) withqThe sum ism p,q (f) Is used for the number of the order of (c),f(x,y) The method is that the image is at the pixel pointx,y) Is a gray value of (a).
The above formula is expressed for a binary function or two-dimensional imagef(x,y) Its advantages are high effectp,q) Moment of orderm p,q Is a calculation formula of (2). Wherein, the method comprises the following steps ofp,q) The order of the moment is represented and,pand (3) withqRespectively represented inxAndyindex in direction. For example, the number of the cells to be processed,p=1,q=0when the first moment is expressed, that is, the function is xAverage value in direction;p=0,q=1when the first moment is expressed, that is, the function isyAverage value in direction. The integral part of the formula represents a function over the entire two-dimensional planef(x,y) Integrating, i.e. toxFrom minus infinity to plus infinity, pairyIntegrating from negative infinity to positive infinity, and obtaining a moment as a result of calculationm p,q Is a value of (2). For example, if one wants to calculatef(x,y) Is of the second moment of (2)m 2,2 Will be in the formulap=2,q=2Substituting and calculating to obtain the following components:
specifically, in a binary image, the above formula can be simplified as:
in some preferred embodiments, the higher order moment is preferably a second, third or fourth order moment, i.ep=q=2Or p=q=3 orp=q=4. In geometric descriptors, high-order momentsMay be used to describe features such as shape, texture, edges, etc. For example, a second moment may be used to represent the area and centroid of a shape, a third moment may be used to describe the rotational torque and symmetry of a shape, and a fourth moment may be used to measure the roundness and edge sharpness of a shape, etc. In edge detection, high-order moments can be used to describe edge shape and intensity in an image. For example, a second moment may be used to measure the length and position of an edge, a third moment may be used to describe the rotational torque and symmetry of an edge, and a fourth moment may be used to measure the curvature and sharpness of an edge, etc. Therefore, by extracting the high-order moment characteristics of the edges in the image, more accurate and reliable edge information can be obtained, so that the performance of image edge detection is improved. In texture analysis, high-order moments may be used to describe the complexity and directionality of textures. For example, a fourth moment may be used to measure directionality and roughness of a texture, and a sixth moment may be used to describe complexity and symmetry of the texture, etc. Therefore, by extracting high-order moment features from textures in the image, richer and more accurate texture features can be obtained, so that the performance of tasks such as image classification, identification, retrieval and the like is improved.
The research shows that the high-order moment corresponding to the low beam region image and the high beam region image is most obvious, the obvious difference is suitable for any vehicle type and any vehicle type, namely, the high-order moment of the low beam region image and the high beam region image corresponding to any vehicle type and different vehicle types is obviously different, namely, the accuracy of the result of judging that the vehicle starts the high beam behavior is very high as long as the high-order moment meets the high beam judging condition, other geometric feature descriptors are not easy to be influenced by the different sizes and forms of the vehicle lamps of different vehicle types, and the difference between the high beam region images of different vehicle types and the low beam region images of different vehicle types, such as the size, the aspect ratio, the roundness, the perimeter and other geometric feature descriptors are large and small, and although the difference is certain, the problem that errors exist due to the small feature difference is considered to exist is considered.
In this embodiment, the training process of the vehicle lamp classification model is:
acquiring historical car light area images of known far and near light classification results, and calculating the high-order moment of each historical car light area image by adopting a high-order moment calculation formula, wherein the historical car light area images comprise far light area images and near light area images of different car types;
Marking each historical car light area image according to the high-order moment of the historical car light area image and the far-near light classification result thereof, so as to construct a model training set;
and training a preset network model through the model training set to obtain the car lamp classification model capable of carrying out far and near light lamp classification recognition according to the high-order moment.
It should be noted that, in other alternative embodiments, it is of course possible to use one or more of other areas, aspect ratios, circularities, and circumferences to perform the high-low beam classification discrimination, where in order to improve reliability, the vehicle type may be first identified, and then the high-low beam classification discrimination is performed under the condition that the vehicle type is known, where of course, the high-high beam classification discrimination may also be performed using the above-mentioned high-order moment. Under the same vehicle type, the low beam region image and the high beam region image have obvious geometrical feature descriptor differences, so that the reliability and the accuracy can reach more than 95 percent by adopting one or more of the area, the aspect ratio, the roundness and the perimeter to classify and judge the high beam and the low beam under the condition of the known vehicle type. In this case, the license plate recognition may be performed on the vehicle running photograph, the vehicle model may be determined according to the license plate information, and then the subsequent vehicle light region image extraction and far-near light classification and discrimination may be further performed, and in addition, the recognition license plate may be associated with the subsequently determined far-light behavior, for example, malicious far-light behavior may be penalized according to the license plate. Meanwhile, when the car light classification model is trained, the obtained historical car light area images of the known far and near light classification result can be clustered, the historical car light area images corresponding to the same car type are gathered into one type, training data sets corresponding to a plurality of car type categories are obtained, then the training data sets corresponding to different car type categories are used for carrying out full-supervision learning on the neural network model respectively, so that the model learns the difference rule among geometric feature descriptors of the far and near light corresponding to different car types, and certainly, the car light classification model corresponding to each car type can be obtained through training respectively aiming at the training data sets corresponding to each car type category, namely, each car type corresponds to one car light classification model, and thus geometric feature descriptors of the known car type can be input into the corresponding car light classification model for recognition.
Example III
The third embodiment of the present invention also provides a far-reaching headlamp opening detection method based on a geometric feature descriptor, which is different from the far-reaching headlamp opening detection method based on a geometric feature descriptor in the first embodiment in that:
according to a second preset geometric feature calculation formula, calculating a geometric feature descriptor of each connected region image, and screening the connected region images according to the geometric feature descriptors of each connected region image to screen out a target connected region with the geometric feature descriptor meeting preset conditions, wherein the method further comprises the following steps:
judging whether the target communication areas meeting the preset conditions are two;
if the number of the target communication areas is not two, the area threshold value, the aspect ratio threshold value and/or the roundness threshold value are/is adjusted until the number of the target communication areas meeting the preset condition is two;
if the number of the geometric feature descriptors is two, according to a preset first geometric feature calculation formula, the step of calculating the geometric feature descriptors corresponding to the car light area image comprises the following steps:
respectively calculating geometric feature descriptors corresponding to the two car light area images according to a preset first geometric feature calculation formula, and judging whether the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value;
And if the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value, executing the step of inputting the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to carry out far and near light classification recognition on the car light area images.
It should be noted that, in general, the vehicle headlamp is disposed in two and symmetrically disposed on both sides of the vehicle hood, and therefore, among the corresponding images, there should be two substantially symmetrical lamp area images, that is, in the case of normal photographing and normal screening, two substantially identical lamp area images should be screened out. However, in practical researches, if the preset conditions are fixed and unchanged, there may occur a phenomenon that one of the lamp area images is screened out, or there are multiple lamp area images (for example, three lamp area images are actually screened out, but one of the lamp area images is obviously a road surface emission area image), which obviously affects the subsequent reliability. When two target communication areas are successfully screened out, the geometric feature descriptors (specifically, high-order moments) of the two target communication areas are subjected to differential analysis, namely, the high-order moment difference value of the two target communication areas is calculated, if the high-order moment difference value is smaller than a preset value, the image interception and screening of the current car light area are reasonable, the follow-up model judgment environment is further entered, if the high-order moment difference value is not smaller than the preset value, the image interception and screening of the current car light area are reasonable, the shooting angle, the environmental influence or the cause of the car (such as damage of a car light on one side of the car) possibly are considered, the fact that the used car running photo cannot be used at this time is considered, the next car running photo is replaced for judgment again, and therefore the reliability of the high beam judgment behavior is greatly improved.
It should be noted that, the above embodiments and the features thereof may be arbitrarily combined without any conflict or special explanation, and the new technical solution obtained by combining the embodiments still belongs to the protection scope of the embodiments of the present invention.
Example IV
In another aspect, referring to fig. 2, a far-reaching headlamp opening detection apparatus according to a fourth embodiment of the present invention is shown, where the far-reaching headlamp opening detection apparatus includes:
an image extraction module 11, configured to obtain at least one vehicle running photograph taken by a camera, and extract a lamp area image of a vehicle from the vehicle running photograph;
the feature calculation module 12 is configured to calculate a geometric feature descriptor corresponding to the vehicle lamp area image according to a preset first geometric feature calculation formula;
the classification and identification module 13 is configured to input geometric feature descriptors corresponding to the vehicle lamp area images into a pre-trained vehicle lamp classification model, so as to perform far-near light classification and identification on the vehicle lamp area images;
and a result output module 14, configured to determine whether the vehicle turns on a high beam according to the classification recognition result of the vehicle lamp classification model.
Further, in some alternative embodiments of the present invention, the geometric feature descriptors include high order moments, and the feature calculation module 12 is further configured to: according to a preset high-order moment calculation formula, calculating a high-order moment corresponding to the car light area image, wherein the high-order moment calculation formula is as follows:
in the method, in the process of the invention,m p,q (f) Is a high order moment in two dimensions, whereinp、qIs a non-negative integer number of the number,pand (3) withqThe sum ism p,q (f) Is used for the number of the order of (c),f(x,y) The method is that the image is at the pixel pointx,y) Is a gray value of (a).
Further, in some alternative embodiments of the present invention, the classification and identification module 13 includes:
the model training unit is used for acquiring historical car light area images of known far and near light lamp classification results and calculating geometric feature descriptors of each historical car light area image by adopting the first geometric feature calculation formula; marking each historical car light area image according to the geometric feature descriptors of the historical car light area images and the far and near light classification results thereof, so as to construct a model training set; training a preset network model through the model training set to obtain the car lamp classification model capable of carrying out far and near light lamp classification and identification according to the geometric feature descriptors.
Further, in some optional embodiments of the present invention, the image extraction module 11 includes:
the image screening unit is used for graying the vehicle running photo according to a preset image graying algorithm to obtain a graying photo, binarizing the graying photo according to a gray threshold value to obtain a binarized photo, and intercepting all connected area images with gray values larger than the gray threshold value from the binarized photo; according to a preset second geometric feature calculation formula, calculating geometric feature descriptors of each communication area image, and screening the communication area images according to the geometric feature descriptors of each communication area image to screen out target communication areas with the geometric feature descriptors meeting preset conditions so as to extract the car light area images.
Further, in some optional embodiments of the present invention, the image graying algorithm is:
f(x,y)=0.3*R(x,y)+0.59*G(x,y)+0.11*B(x,y)
in the method, in the process of the invention,R(x,y)、G(x,y) and B(x,y) Pixel points representing the vehicle running pictures respectivelyx,y) Is set, the gray threshold is 220.
Further, in some optional embodiments of the present invention, the image filtering unit is further configured to calculate an area, an aspect ratio, and a roundness of each of the connected region images according to a preset second geometric feature calculation formula, where the aspect ratio is a ratio of a width to a height of the connected region image; and removing all the communication area images with the area smaller than an area threshold value, the aspect ratio smaller than an aspect ratio threshold value and/or the roundness smaller than a roundness threshold value from the communication area images, so as to screen out target communication areas with geometric feature descriptors meeting preset conditions, and extract the car light area images.
Further, in some optional embodiments of the present invention, the far-reaching headlamp opening detection apparatus further includes:
the screening detection module is used for judging whether the number of the target communication areas meeting the preset conditions is two after screening the target communication areas meeting the preset conditions of the geometric feature descriptors; if the number of the target communication areas is not two, the area threshold value, the aspect ratio threshold value and/or the roundness threshold value are/is adjusted until the number of the target communication areas meeting the preset condition is two; if the number of the vehicle lamp images is two, the feature calculation module 12 is further configured to calculate geometric feature descriptors corresponding to the two vehicle lamp area images according to a preset first geometric feature calculation formula, and determine whether a difference between the geometric feature descriptors corresponding to the two vehicle lamp area images is smaller than a preset value; if the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value, the classification and identification module 13 inputs the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to perform far-and-near light classification and identification on the car light area images.
The functions or operation steps implemented when the above modules and units are executed are substantially the same as those in the above method embodiments, and are not described herein again.
Example five
In another aspect, referring to fig. 3, a high beam on detection apparatus according to a fifth embodiment of the present invention includes a memory 20, a processor 10, and a computer program 30 stored in the memory and capable of running on the processor, where the processor 10 implements the high beam on detection method based on the geometric feature descriptors as described above when executing the computer program 30.
The high beam on detection device may specifically be a cloud server, an image capturing apparatus, or the like, and the processor 10 may in some embodiments be a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, for running program codes or processing data stored in the memory 20, for example, executing an access restriction program, or the like.
The memory 20 includes at least one type of readable storage medium including flash memory, a hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 20 may in some embodiments be an internal memory unit of the high beam on detection device, such as a hard disk of the high beam on detection device. The memory 20 may also be an external storage device of the high beam light on detection apparatus in other embodiments, such as a plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card) or the like, which is provided on the high beam light on detection apparatus. Further, the memory 20 may also include both an internal memory unit and an external memory device of the high beam on detection apparatus. The memory 20 may be used not only for storing application software installed in the high beam on detection apparatus and various types of data, but also for temporarily storing data that has been output or is to be output.
It is noted that the configuration shown in fig. 3 does not constitute a limitation of the high beam on detection apparatus, and in other embodiments, the high beam on detection apparatus may include fewer or more components than shown, or may combine certain components, or may have a different arrangement of components.
The embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, realizes the high beam on detection method based on the geometric feature descriptors.
Those of skill in the art will appreciate that the logic and/or steps represented in the flow diagrams or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable storage medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable storage medium may even be paper or other suitable medium upon which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The foregoing examples illustrate only a few embodiments of the invention and are described in detail herein without thereby limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (7)
1. The method for detecting the opening of the high beam based on the geometric feature descriptors is characterized by comprising the following steps of:
Acquiring at least one vehicle running photo shot by a camera, and extracting a car light area image of a vehicle from the vehicle running photo;
calculating a geometric feature descriptor corresponding to the car light area image according to a preset first geometric feature calculation formula;
inputting geometric feature descriptors corresponding to the car light region images into a pre-trained car light classification model so as to carry out far-near light classification identification on the car light region images;
judging whether the vehicle starts a high beam or not according to the classification and identification result of the vehicle lamp classification model;
wherein the step of extracting the lamp area image of the vehicle from the vehicle running photograph comprises the following steps:
graying the vehicle running photo according to a preset image graying algorithm to obtain a graying photo, binarizing the graying photo according to a gray threshold to obtain a binarizing photo, and intercepting all connected area images with gray values larger than the gray threshold from the binarizing photo;
calculating a geometric feature descriptor of each communication area image according to a preset second geometric feature calculation formula, and screening the communication area images according to the geometric feature descriptor of each communication area image to screen out a target communication area of which the geometric feature descriptor meets preset conditions so as to extract the car light area image;
The step of calculating the geometric feature descriptor of each communication area image according to a preset second geometric feature calculation formula, and screening the communication area images according to the geometric feature descriptor of each communication area image to screen out a target communication area of which the geometric feature descriptor meets preset conditions, so as to extract the car light area image comprises the following steps:
calculating the area, the aspect ratio and the roundness of each communication area image according to a preset second geometric feature calculation formula, wherein the aspect ratio is the ratio of the width to the height of the communication area image;
removing all the communication area images with the area smaller than an area threshold value, the aspect ratio smaller than an aspect ratio threshold value and/or the roundness smaller than a roundness threshold value from the communication area images, so as to screen out target communication areas with geometric feature descriptors meeting preset conditions, and extracting the car light area images;
the step of calculating the geometric feature descriptor of each connected region image according to a preset second geometric feature calculation formula, and screening the connected region image according to the geometric feature descriptor of each connected region image to screen out a target connected region with the geometric feature descriptor meeting preset conditions further comprises the following steps:
Judging whether the target communication areas meeting the preset conditions are two;
if the number of the target communication areas is not two, the area threshold value, the aspect ratio threshold value and/or the roundness threshold value are/is adjusted until the number of the target communication areas meeting the preset condition is two;
if the number of the geometric feature descriptors is two, according to a preset first geometric feature calculation formula, the step of calculating the geometric feature descriptors corresponding to the car light area image comprises the following steps:
respectively calculating geometric feature descriptors corresponding to the two car light area images according to a preset first geometric feature calculation formula, and judging whether the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value;
and if the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value, executing the step of inputting the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to carry out far and near light classification recognition on the car light area images.
2. The method for detecting the opening of the high beam based on the geometric feature descriptors according to claim 1, wherein the geometric feature descriptors include high-order moments, and the step of calculating the geometric feature descriptors corresponding to the vehicle lamp area image according to a preset first geometric feature calculation formula includes:
According to a preset high-order moment calculation formula, calculating a high-order moment corresponding to the car light area image, wherein the high-order moment calculation formula is as follows:
in the method, in the process of the invention,m p,q (f) Is a high order moment in two dimensions, whereinp、qIs a non-negative integer number of the number,pand (3) withqThe sum ism p,q (f) Is used for the number of the order of (c),f(x,y) The method is that the image is at the pixel pointx,y) Is a gray value of (a).
3. The method for detecting the opening of the high beam based on the geometric feature descriptors of claim 1, wherein the training process of the car light classification model is as follows:
acquiring historical car light area images of known far and near light lamp classification results, and calculating geometric feature descriptors of each historical car light area image by adopting the first geometric feature calculation formula;
marking each historical car light area image according to the geometric feature descriptors of the historical car light area images and the far and near light classification results thereof, so as to construct a model training set;
training a preset network model through the model training set to obtain the car lamp classification model capable of carrying out far and near light lamp classification and identification according to the geometric feature descriptors.
4. The method for detecting the opening of the high beam based on the geometric feature descriptors according to claim 1, wherein the image graying algorithm is as follows:
f(x,y)=0.3* R(x,y)+0.59* G(x,y)+0.11* B(x,y)
In the method, in the process of the invention,R(x,y)、G(x,y) and B(x,y) Pixel points representing the vehicle running pictures respectivelyx,y) Is set, the gray threshold is 220.
5. The utility model provides a detection device is opened to far-reaching headlamp, its characterized in that, detection device is opened to far-reaching headlamp includes:
the image extraction module is used for acquiring at least one vehicle running photo shot by the camera and extracting a car light area image of the vehicle from the vehicle running photo;
the feature calculation module is used for calculating a geometric feature descriptor corresponding to the car light area image according to a preset first geometric feature calculation formula;
the classification and identification module is used for inputting geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to perform far-near light classification and identification on the car light area images;
the result output module is used for judging whether the vehicle starts a high beam or not according to the classification and identification result of the vehicle lamp classification model;
wherein, the image extraction module includes:
the image screening unit is used for graying the vehicle running photo according to a preset image graying algorithm to obtain a graying photo, binarizing the graying photo according to a gray threshold value to obtain a binarized photo, and intercepting all connected area images with gray values larger than the gray threshold value from the binarized photo; calculating a geometric feature descriptor of each communication area image according to a preset second geometric feature calculation formula, and screening the communication area images according to the geometric feature descriptor of each communication area image to screen out a target communication area of which the geometric feature descriptor meets preset conditions so as to extract the car light area image;
The image screening unit is further configured to calculate an area, an aspect ratio and a roundness of each of the connected region images according to a preset second geometric feature calculation formula, where the aspect ratio is a ratio of a width to a height of the connected region image; removing all the communication area images with the area smaller than an area threshold value, the aspect ratio smaller than an aspect ratio threshold value and/or the roundness smaller than a roundness threshold value from the communication area images, so as to screen out target communication areas with geometric feature descriptors meeting preset conditions, and extracting the car light area images;
the high beam opening detection device further comprises:
the screening detection module is used for judging whether the number of the target communication areas meeting the preset conditions is two after screening the target communication areas meeting the preset conditions of the geometric feature descriptors; if the number of the target communication areas is not two, the area threshold value, the aspect ratio threshold value and/or the roundness threshold value are/is adjusted until the number of the target communication areas meeting the preset condition is two; if the number of the geometric feature descriptors is two, the feature calculation module is also used for respectively calculating geometric feature descriptors corresponding to the two car light area images according to a preset first geometric feature calculation formula, and judging whether the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value; if the difference between the geometric feature descriptors corresponding to the two car light area images is smaller than a preset value, the classification and identification module inputs the geometric feature descriptors corresponding to the car light area images into a pre-trained car light classification model so as to perform far and near light lamp classification and identification on the car light area images.
6. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements a high beam on detection method based on a geometrical descriptor according to any one of claims 1-4.
7. A high beam on detection device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, said processor implementing the high beam on detection method based on a geometrical descriptor according to any one of claims 1-4 when executing said program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310267314.7A CN115984828B (en) | 2023-03-20 | 2023-03-20 | High beam opening detection method, device and equipment based on geometric feature descriptors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310267314.7A CN115984828B (en) | 2023-03-20 | 2023-03-20 | High beam opening detection method, device and equipment based on geometric feature descriptors |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115984828A CN115984828A (en) | 2023-04-18 |
| CN115984828B true CN115984828B (en) | 2023-06-06 |
Family
ID=85970912
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310267314.7A Active CN115984828B (en) | 2023-03-20 | 2023-03-20 | High beam opening detection method, device and equipment based on geometric feature descriptors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115984828B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108944650A (en) * | 2018-08-14 | 2018-12-07 | 浙江安谐智能科技有限公司 | A kind of car light open state method of discrimination based on long-and-short distant light irradiation principle |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140198213A1 (en) * | 2013-01-15 | 2014-07-17 | Gentex Corporation | Imaging system and method for detecting fog conditions |
| CN103295399B (en) * | 2013-05-14 | 2015-01-21 | 西安理工大学 | On-state judging method of headlights on full beam of night-driving cars based on morphological characteristics |
| CN203580751U (en) * | 2013-10-12 | 2014-05-07 | 郑州日产汽车有限公司 | Motor vehicle daytime running lamp controller |
| CN107563265B (en) * | 2016-06-30 | 2021-08-17 | 杭州海康威视数字技术股份有限公司 | High beam detection method and device |
| CN211827564U (en) * | 2019-04-22 | 2020-10-30 | 桂林金铱星科技发展有限公司 | A recognition device for detecting whether vehicle opens high beam night |
| CN111649918A (en) * | 2020-06-17 | 2020-09-11 | 郑州高识智能科技有限公司 | Method for monitoring turning on of high beam and continuous tracking of vehicle in real time |
| CN111783573B (en) * | 2020-06-17 | 2023-08-25 | 杭州海康威视数字技术股份有限公司 | High beam detection method, device and equipment |
| CN114454809A (en) * | 2020-10-31 | 2022-05-10 | 华为技术有限公司 | Intelligent light switching method, system and related equipment |
| CN112949578B (en) * | 2021-03-30 | 2023-04-07 | 苏州科达科技股份有限公司 | Vehicle lamp state identification method, device, equipment and storage medium |
-
2023
- 2023-03-20 CN CN202310267314.7A patent/CN115984828B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108944650A (en) * | 2018-08-14 | 2018-12-07 | 浙江安谐智能科技有限公司 | A kind of car light open state method of discrimination based on long-and-short distant light irradiation principle |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115984828A (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7724962B2 (en) | Context adaptive approach in vehicle detection under various visibility conditions | |
| US10025998B1 (en) | Object detection using candidate object alignment | |
| Pavlic et al. | Classification of images in fog and fog-free scenes for use in vehicles | |
| TWI497422B (en) | A system and method for recognizing license plate image | |
| Kosaka et al. | Vision-based nighttime vehicle detection using CenSurE and SVM | |
| CN109711264B (en) | Method and device for detecting occupation of bus lane | |
| JP5223675B2 (en) | Vehicle detection device, vehicle detection method, and vehicle detection program | |
| CN106709412B (en) | Traffic sign detection method and device | |
| CN111027535A (en) | License plate recognition method and related equipment | |
| CN111507145A (en) | Method, system and device for detecting barrier at storage position of embedded vehicle-mounted all-round looking system | |
| CN106407951A (en) | Monocular vision-based nighttime front vehicle detection method | |
| WO2023029467A1 (en) | Method and apparatus for vehicle light detection, electronic device, and storage medium | |
| CN110276318A (en) | Nighttime road rains recognition methods, device, computer equipment and storage medium | |
| CN111046741A (en) | Method and device for identifying lane line | |
| Shang et al. | A novel method for vehicle headlights detection using salient region segmentation and PHOG feature | |
| CN110619255B (en) | Target detection method and device | |
| CN112784675B (en) | Target detection method and device, storage medium and terminal | |
| CN115984828B (en) | High beam opening detection method, device and equipment based on geometric feature descriptors | |
| WO2023044656A1 (en) | Vehicle passage warning method and apparatus, and vehicle-mounted terminal | |
| CN104809438A (en) | Method and device for detecting electronic eyes | |
| CN114841874A (en) | Image processing method, device, equipment and storage medium | |
| CN110321973B (en) | Combined vehicle detection method based on vision | |
| JP7264428B2 (en) | Road sign recognition device and its program | |
| CN113581059A (en) | Light adjusting method and related device | |
| Wangsiripitak et al. | Traffic light and crosswalk detection and localization using vehicular camera |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |