CN107240167A - A kind of drive recorder pedestrian monitoring system - Google Patents

A kind of drive recorder pedestrian monitoring system Download PDF

Info

Publication number
CN107240167A
CN107240167A CN201710230018.4A CN201710230018A CN107240167A CN 107240167 A CN107240167 A CN 107240167A CN 201710230018 A CN201710230018 A CN 201710230018A CN 107240167 A CN107240167 A CN 107240167A
Authority
CN
China
Prior art keywords
pedestrian
collision coefficient
collision
coefficient
dummy
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710230018.4A
Other languages
Chinese (zh)
Inventor
罗健刚
熊爱林
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.)
Jiangsu Car Vision Electronics Co Ltd
Original Assignee
Jiangsu Car Vision Electronics Co Ltd
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 Jiangsu Car Vision Electronics Co Ltd filed Critical Jiangsu Car Vision Electronics Co Ltd
Priority to CN201710230018.4A priority Critical patent/CN107240167A/en
Publication of CN107240167A publication Critical patent/CN107240167A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07CTIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
    • G07C5/00Registering or indicating the working of vehicles
    • G07C5/08Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
    • G07C5/0841Registering performance data
    • G07C5/085Registering performance data using electronic data carriers
    • G07C5/0866Registering performance data using electronic data carriers the electronic data carrier being a digital video recorder in combination with video camera
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/00624Recognising scenes, i.e. recognition of a whole field of perception; recognising scene-specific objects
    • G06K9/00771Recognising scenes under surveillance, e.g. with Markovian modelling of scene activity
    • G06K9/00778Recognition or static of dynamic crowd images, e.g. recognition of crowd congestion
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00Details of television systems
    • H04N5/222Studio circuitry; Studio devices; Studio equipment ; Cameras comprising an electronic image sensor, e.g. digital cameras, video cameras, TV cameras, video cameras, camcorders, webcams, camera modules for embedding in other devices, e.g. mobile phones, computers or vehicles
    • H04N5/225Television cameras ; Cameras comprising an electronic image sensor, e.g. digital cameras, video cameras, camcorders, webcams, camera modules specially adapted for being embedded in other devices, e.g. mobile phones, computers or vehicles
    • H04N5/2257Mechanical and electrical details of cameras or camera modules for embedding in other devices
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00Details of television systems
    • H04N5/222Studio circuitry; Studio devices; Studio equipment ; Cameras comprising an electronic image sensor, e.g. digital cameras, video cameras, TV cameras, video cameras, camcorders, webcams, camera modules for embedding in other devices, e.g. mobile phones, computers or vehicles
    • H04N5/225Television cameras ; Cameras comprising an electronic image sensor, e.g. digital cameras, video cameras, camcorders, webcams, camera modules specially adapted for being embedded in other devices, e.g. mobile phones, computers or vehicles
    • H04N5/2258Cameras using two or more image sensors, e.g. a CMOS sensor for video and a CCD for still image
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N9/00Details of colour television systems
    • H04N9/04Picture signal generators
    • H04N9/09Picture signal generators with more than one pick-up device

Abstract

The present invention relates to a kind of drive recorder pedestrian monitoring system, including acquisition module, the acquisition module includes motion sensing control device, infrared sensor and depth gauge, the motion sensing control device includes RGB cameras, the RGB cameras are used for the color video frequency image generation RGB image of shooting visual angle scope one skilled in the art, depth image of the infrared sensor to obtain pedestrian, the depth image and RGB image that the depth gauge is obtained based on the infrared sensor generate the depth image stream of pedestrian, the present invention can be monitored for pedestrian, in the original writing function of drive recorder, add the monitoring system to pedestrian, the people for touching porcelain tendency can more accurately be differentiated, so as to be prevented in advance.

Description

A kind of drive recorder pedestrian monitoring system
Technical field
The present invention relates to drive recorder field, more particularly to a kind of drive recorder pedestrian monitoring system.
Background technology
Current Peng Ci parties, suspicion road safety hidden danger such as side of colliding have expedited the emergence of drive recorder market, but current row Car recorder is main based on image taking, videograph, without purpose.
Such as Chinese invention patent application of Application No. 201610054061.5, discloses a kind of drive recorder, its Including:Image collection analytic unit, main control chip, information output interface and command receiver;Image collection analytic unit, information Output interface and command receiver are connected with main control chip;Image collection analytic unit is used to transmit the image information of collection To main control chip, main control chip is used to handle conversion image information, and information output interface is used to pass the image information after conversion It is defeated by vehicle mounted guidance;Command receiver is used to receive the command signal that vehicle mounted guidance is sent, and the command signal is transferred into master Chip is controlled, main control chip produces the corresponding control signal of the command signal, and control signal is used to control drive recorder.Using upper Drive recorder is stated, video image is included on vehicle mounted guidance screen, the display effect of video can be improved;And in driving During, human pilot by vehicle mounted guidance with regard to the drive recorder can be operated so that the operation of drive recorder becomes more Plus conveniently, the problem of which solving easy to operate.
In another example the Chinese invention patent of Application No. 201310576506.2, discloses a kind of drive recorder, it is wrapped Central processing unit, memory, power supply circuit and camera are included, the memory, power supply circuit and camera are handled with center Device signal is connected, in addition to acceleration induction device and stepper motor, and the acceleration induction device and stepper motor are and centre Manage device signal to be connected, the stepper motor is also connected with power supply circuit, and the camera direction is adjustable, the stepper motor it is defeated Shaft is connected with the direction regulator site of the camera, can provide and meet the good driving shadow of human eye perspective custom, imaging image quality As information, it is easy to quantify to judge accident conditions, beneficial to ex-post analysis, the problem of solving imaging.
The content of the invention
In order to solve the above problems there is provided a kind of drive recorder pedestrian monitoring system, the present invention can enter for pedestrian Row monitoring, in the original writing function of drive recorder, adds the monitoring system to pedestrian, can more accurately differentiate and touch The people of porcelain tendency, so as to be prevented in advance.
Concrete scheme is:A kind of drive recorder pedestrian monitoring system, including acquisition module, processing module,
The acquisition module includes motion sensing control device, infrared sensor and depth gauge, and the motion sensing control device is taken the photograph including RGB As head, the RGB cameras are used for the color video frequency image generation RGB image of shooting visual angle scope one skilled in the art, the infrared biography Depth image of the sensor to obtain pedestrian, depth image and the RGB figure that the depth gauge is obtained based on the infrared sensor As the depth image stream of generation pedestrian;
The processing module includes graphics processing unit and memory cell, and described image processing unit is based on the depth map Picture stream rebuilds pedestrian's 3-D view and the pedestrian dummy storehouse with being stored in memory cell is contrasted;Described image processing unit bag Include a feature extraction unit and Characteristic Contrast unit, the feature extraction unit be used to extract the feature in pedestrian's 3-D view and Contrasted by Characteristic Contrast unit with the model in pedestrian dummy storehouse, the pedestrian dummy storehouse in the memory cell is provided with spy Storehouse is levied, posture and build of the feature database based on pedestrian dummy are respectively equipped with pose parameter θ and shape parameter β, the pedestrian Any of which pedestrian dummy in model library is respectively provided with corresponding a pose parameter θ and shape parameter β, and pedestrian dummy appearance Potential parameter θ and shape parameter β respectively constitute the first collision coefficient α and the second collision coefficient γ and the first collision coefficient α and second Collision coefficient γ sums constitute a total collision coefficient α γ.
Above-mentioned monitoring system, wherein, in addition to alarm module and power module, the alarm module include display screen and Voice, the alarm module sends alarm according to the result of contrast;The power module respectively with the acquisition module, place Manage module and alarm module electrical connection.
Above-mentioned monitoring system, wherein, the step of described image processing unit rebuilds pedestrian's 3-D view is:Using multiframe Sum-average arithmetic and two-sided filter will obtain the noise remove produced during depth image stream, using the point cloud after denoising as can Varying model fitting algorithm is inputted;Detailed process is:Assuming that depth image stream point cloud T includes point T (q), wherein, q ∈ [1:Q], Q Number of vertices, secondly, the shape parameter of pedestrian dummy is β in pedestrian dummy storehouse, and pose parameter is θ, with shape parameter β and The point that pose parameter θ pedestrian dummy is designated as on M β, θ, M β, θ is denoted as M β, θ (p), wherein p ∈ [1:P], P is pedestrian dummy Number of vertices, based on the summit M β of pedestrian dummy in summit T (q) in above-mentioned depth image stream and pedestrian dummy storehouse, θ (p) is searched It is corresponding, pedestrian's 3-D view is then generated by way of Mesh Fitting, so that it is determined that the shape parameter β and appearance of the pedestrian Potential parameter θ.
Above-mentioned monitoring system, wherein, the shape parameter β includes working as pedestrian's epigastric angle less than 90 degree, its second collision Coefficient gamma is 0.1-1.5;When 90 degree of pedestrian's epigastric angle, its second collision coefficient γ is 1.6-2.5;When pedestrian's epigastric angle is more than 90 Degree, its second collision coefficient γ is 2.6-3.5.
Above-mentioned monitoring system, wherein, the pose parameter θ is included when pedestrian is walking postures, its first collision coefficient α For 0.1-1.5;It is posture of trotting as pedestrian, its first collision coefficient α is 1.6-2.5;As pedestrian to run posture, it first is touched Factor alpha is hit for 2.6-3.5.
Above-mentioned monitoring system, wherein, in the case of different pedestrians, described image processing unit determines the body of the pedestrian Shape parameter β and pose parameter θ, and corresponding first collision coefficient α and the second collision coefficient γ are generated, by the first collision of generation The the first collision coefficient α and the second collision coefficient of factor alpha and the second collision coefficient γ respectively with pedestrian dummy in the model library γ is contrasted, and by the first collision coefficient α and the second collision coefficient γ sums and a certain pedestrian dummy in the model library Collision coefficient α γ contrasted and according to above-mentioned comparing result determine whether collision risk.
Above-mentioned monitoring system, wherein, in the pedestrian dummy of collection, if the first collision coefficient α and the second collision coefficient γ is not less than 1.6 and total collision coefficient is not less than 5, then is collision model;If the first collision coefficient α or the second collision coefficient γ are not Then it is warning model less than 1.6 and total collision coefficient is less than 5;If the first collision coefficient α and the second collision coefficient γ is less than 1.6, It is then security model.
Above-mentioned monitoring system, wherein, if by being collision model with pedestrian dummy contrast in the model library, it is described The pedestrian contour is shown as dangerous color and audio alert in the display screen of alarm module, if by with pedestrian's mould in the model library Type contrast for warning model, then the pedestrian contour is shown as warning color in the display screen of the alarm module, if by with it is described Pedestrian dummy contrast is security model, then not alarm indication in model library.
Advantages of the present invention:It can be monitored, in the original writing function of drive recorder, add for pedestrian To the monitoring system of pedestrian, the people for touching porcelain tendency can be more accurately differentiated, so as to be prevented in advance.
Brief description of the drawings
By reading the detailed description made with reference to the following drawings to non-limiting example, the present invention and its feature, outside Shape and advantage will become more apparent upon.Identical mark indicates identical part in whole accompanying drawings.Not deliberately proportionally Draw accompanying drawing, it is preferred that emphasis is the purport of the present invention is shown.
Fig. 1 provides reference view during pedestrian's walking for the present invention.
Fig. 2 is the reference view in pedestrian dummy storehouse in the present invention.
Fig. 3 is the schematic flow sheet for the reconstruction pedestrian's 3-D view being related in the present invention.
Embodiment
In the following description, a large amount of concrete details are given to provide more thorough understanding of the invention.So And, it is obvious to the skilled person that the present invention can be able to without one or more of these details Implement.In other examples, in order to avoid obscuring with the present invention, do not enter for some technical characteristics well known in the art Row description.
In order to thoroughly understand the present invention, detailed step and detailed structure will be proposed in following description, so as to Explain technical scheme.Presently preferred embodiments of the present invention is described in detail as follows, but in addition to these detailed descriptions, this Invention can also have other embodiment.
A kind of drive recorder pedestrian monitoring system that the present invention is provided, including acquisition module, further, acquisition module bag Motion sensing control device, infrared sensor and depth gauge are included, wherein infrared sensor includes infrared transmitter and infrared remote receiver, body-sensing Controller includes RGB cameras, and RGB cameras are used for the color video frequency image generation RGB image of shooting visual angle scope one skilled in the art, Depth image of the infrared sensor to obtain pedestrian, its principle is:Infrared transmitter and infrared remote receiver actively project near red Outer line spectrum, is irradiated to rough object or penetrates after frosted glass, can form random reflected spot, referred to as speckle, and then The depth image of pedestrian can be read by infrared camera, further, infrared spectrum is analyzed by depth gauge, creates visual model Interior human depth's image is enclosed, the depth image and RGB image that depth gauge is obtained based on infrared sensor generate the depth of pedestrian Image stream, further preferably, the illumination condition regardless of surrounding environment perceive environment by way of black and white spectrum, infrared Sensor and depth gauge generate depth image stream with the speed of 30 frame per second;Processing module, processing module includes image procossing list Member and memory cell, graphics processing unit based on depth image stream rebuild pedestrian's 3-D view and with the row that is stored in memory cell People's model library is contrasted;Also include alarm module and power module, further, alarm module includes display screen and voice, its Middle alarm module sends alarm according to the result of contrast;Power module respectively with acquisition module, processing module and alarm mould Block is electrically connected.Shown in reference picture 1, Fig. 2, the color video frequency image that wherein a large amount of RGB cameras of unit records are shot also is deposited The pedestrian dummy storehouse of various builds is stored up, for example, pedestrian dummy can be static or dynamic.Each pedestrian Model further comprises multiple dangerous gesture models (issuable a large amount of dangerous posture moulds when i.e. pedestrian walks on road Type), such as the dangerous gesture model trot, run, because running, this posture easily has influence on traffic safety.
The present invention one is preferably and in non-limiting embodiment, and graphics processing unit includes a feature extraction unit and feature Comparison unit, wherein feature extraction unit are used to extract the feature in pedestrian's 3-D view and by Characteristic Contrast unit and pedestrian Model in model library is contrasted, the present invention feature for point a cloud, that is, based on the contrast of point cloud chart picture come judge row Whether the action of people has danger, further, and the pedestrian dummy storehouse in memory cell is provided with feature database, and wherein feature database is based on The posture and build of pedestrian dummy are respectively equipped with any of which row in pose parameter θ and shape parameter β, and pedestrian dummy storehouse People's model is respectively provided with corresponding a pose parameter θ and shape parameter β, and the pose parameter θ and shape parameter β of pedestrian dummy divide The first collision coefficient α and the second collision coefficient γ are not constituted and the first collision coefficient α and the second collision coefficient γ sums constitute one Individual total collision coefficient α γ, in the present invention, during drive recorder is recorded, the action to each pedestrian of record is carried out Distinguish, specifically by the image of collection pedestrian, then carry out the three dimensional point cloud after three-dimensional reconstruction is rebuild, and will The cloud data of pedestrian dummy in the three dimensional point cloud and model library of reconstruction is contrasted, can to judge whether the pedestrian has The risk of porcelain can be touched, and be prevented.
The present invention will collection pedestrian image by processing after with pedestrian dummy storehouse pedestrian dummy image carry out pair Than, can be with to judge whether the action of the pedestrian is dangerous play, if be likely to occur the risk for touching porcelain, and prevented More accurately differentiating has the people for touching porcelain tendency.
The present invention one is preferably and in non-limiting embodiment, shown in reference picture 3, and it is three-dimensional that graphics processing unit rebuilds pedestrian The step of image is:Gone using multiframe sum-average arithmetic with two-sided filter by the noise produced during depth image stream is obtained Remove, inputted the point cloud after denoising as variable model fitting algorithm;Detailed process is:Assuming that depth image stream point cloud T bags The T containing point (q), wherein, q ∈ [1:Q], Q is number of vertices, secondly, and the shape parameter of pedestrian dummy is β, appearance in pedestrian dummy storehouse Potential parameter is θ, and the pedestrian dummy with shape parameter β and pose parameter θ is designated as Mβ, θ, Mβ, θOn point be denoted as Mβ, θ (p), wherein p ∈[1:P], P is the number of vertices of pedestrian dummy, based on pedestrian in summit T (q) in above-mentioned depth image stream and pedestrian dummy storehouse The summit M of modelβ, θ (p)Corresponding point is searched, pedestrian's 3-D view is then generated by way of Mesh Fitting, so that it is determined that should The shape parameter β and pose parameter θ of pedestrian, that is to say, that the pedestrian image first to collection is handled in the present invention The cloud data of the pedestrian, is then contrasted the cloud data of pedestrian dummy in obtained cloud data and model library, by In model library the shape parameter and pose parameter of pedestrian dummy be it is known, so as to obtain the pedestrian shape parameter and Pose parameter, and then judge whether the pedestrian is likely to occur the wind for touching porcelain according to the shape parameter and pose parameter of the pedestrian Danger, and prevented.
The present invention one is preferably and in non-limiting embodiment, and shape parameter β includes working as pedestrian's epigastric angle less than 90 degree, is Asthenic type is also known as asthenic, and its second collision coefficient γ is 0.1-1.5, such as when pedestrian is child, its second collision coefficient γ is 1.5, and when pedestrian is old man, its second collision coefficient γ is 0.1, when pedestrian is young people, its second collision coefficient γ is 1;It is that orthotonic type is also known as well-balanced type when 90 degree of pedestrian's epigastric angle, its second collision coefficient γ is 1.6-2.5, similarly basis Pedestrian's all ages and classes, its second collision coefficient γ is also different;It is that sthenic type is also known as short when pedestrian's epigastric angle is more than 90 degree Fat, its second collision coefficient γ is 2.6-3.5, similarly according to pedestrian's all ages and classes, and its second collision coefficient γ is also different 's.
The present invention one is preferably and in non-limiting embodiment, and pose parameter θ is included when pedestrian is walking postures, and it first Collision coefficient α is 0.1-1.5;It is posture of trotting as pedestrian, its first collision coefficient α is 1.6-2.5;When pedestrian is to running style Gesture, its first collision coefficient α is 2.6-3.5, in addition to when pedestrian is old man or child, its corresponding first collision coefficient α is differed, and is made a distinction with reference to the principle in above-mentioned.For example, when pedestrian is child, in the case of walking postures, its First collision coefficient α is 1.5, when pedestrian is old man, and in the case of walking postures, its first collision coefficient α is 0.1, successively Analogize, then for example:When pedestrian is child, in the case of posture of trotting, its first collision coefficient α is 2.5, when pedestrian is old man When, in the case of posture of trotting, its first collision coefficient α is 1.6, in the above cases, in the pedestrian dummy of collection, Be exactly drive recorder collection pedestrian dummy in, if the first collision coefficient α and the second collision coefficient γ is not less than 1.6 and always touching Coefficient is hit not less than 5, then is collision model;If the first collision coefficient α or the second collision coefficient γ is not less than 1.6 and total collision is Number is less than 5, then is warning model;If the first collision coefficient α and the second collision coefficient γ is less than 1.6, for security model.
The present invention one is preferably and in non-limiting embodiment, in the case of different pedestrians, and graphics processing unit is determined The shape parameter β and pose parameter θ of the pedestrian, and corresponding first collision coefficient α and the second collision coefficient γ are generated, it will generate The first collision coefficient α and second respectively with pedestrian dummy in model library of the first collision coefficient α and the second collision coefficient γ touch Coefficient gamma is hit to be contrasted, and by the first collision coefficient α and the second collision coefficient γ sums and certain a line in the model library The collision coefficient α γ of people's model are contrasted and the risk of collision are determined whether according to above-mentioned comparing result, that is to say, that The shape parameter β and pose parameter θ of the pedestrian determined in the present invention by graphics processing unit, the first collision obtained from Factor alpha and the second collision coefficient γ, can pedestrian dummies different from model library the first collision coefficient α and second collision system Number γ is contrasted or contrasted with the first collision coefficient α and the second collision coefficient γ of same pedestrian dummy, including Several situations below;
(1):The first collision coefficient α of the pedestrian can be carried out with the first collision coefficient α in A pedestrian dummies in model library Contrast, so that it is determined that whether the posture of the pedestrian has danger, determination mode is:If the first collision coefficient α is in A pedestrian's mould Corresponding posture is dangerous posture in type, then judges that the posture has dangerous possibility.
(2):The first collision coefficient α of the pedestrian can be carried out with the first collision coefficient α in B pedestrian dummies in model library Contrast, so that it is determined that whether the posture has danger, determination mode is:If the first collision coefficient α is right in B pedestrian dummies The posture answered is safe posture, then judges that the posture has the possibility of safety.
(3):The second collision coefficient γ of the pedestrian can enter with the second collision coefficient γ in A pedestrian dummies in model library Row contrast, so that it is determined that whether the build of the pedestrian has danger, determination mode is:If the second collision coefficient γ is in A rows Corresponding build is dangerous build in people's model, then judges that pedestrian has dangerous possibility under the build state.
(4):The second collision coefficient γ of the pedestrian can enter with the second collision coefficient γ in B pedestrian dummies in model library Row contrast, so that it is determined that whether the build of the pedestrian has danger, determination mode is:If the second collision coefficient γ is in B rows Corresponding build is safe build in people's model, then judges that pedestrian has the possibility of safety under the build state.
(5):The the first collision coefficient α and the second collision coefficient γ sums of the pedestrian and A pedestrian dummies or B rows in model library Total collision coefficient α γ in people's model are contrasted, so that it is determined that whether the build of the pedestrian has danger, one of which side Formula determination mode is:If the first collision coefficient α and the second collision coefficient γ sums are right in A pedestrian dummies or B pedestrian dummies The coefficient answered is safety coefficient, that is to say, that corresponding security model, then judges that pedestrian is safe in this case, do not touch Porcelain risk.
(6):The the first collision coefficient α and the second collision coefficient γ sums of the pedestrian and A pedestrian dummies or B rows in model library Total collision coefficient α γ in people's model are contrasted, so that it is determined that whether the build of the pedestrian has danger, it is another to determine Mode is:If the first collision coefficient α and the second collision coefficient γ sums corresponding system in A pedestrian dummies or B pedestrian dummies Number is danger coefficient, that is to say, that corresponding collision model, then judges that pedestrian has in this case and touch porcelain risk, it is necessary to keep away Allow, prevent from touching porcelain.
To sum up several situations, if same pedestrian meets (1) and (3) or meet the situation of (6) simultaneously, had both met first and had touched Factor alpha and the second collision coefficient γ are hit not less than 1.6 and total collision coefficient is not less than 5 collision model parameter, then is needed to row The porcelain risk of touching of people is avoided, and the risk for touching porcelain is very high;If meeting one kind in (1) or (3), the first collision coefficient was both met α or the second collision coefficient γ is not less than 1.6 and total collision coefficient is less than 5 warning model parameter, then needs pre- anti-collision porcelain risk; It if meeting (2) and (4) simultaneously or meeting the situation of (5), need not be prevented, both meet the first collision coefficient α and second Collision coefficient γ is less than 1.6 security model parameter, does not touch porcelain risk, it is to be noted that the A rows being related in the present invention The implication that people's model or B pedestrian dummies are not particularly limited, is only intended to the differentiation to different pedestrian dummies, can refer to model library Any of pedestrian dummy.
Further preferably, if by being collision model, the display screen of alarm module with pedestrian dummy contrast in model library In the pedestrian contour be shown as dangerous color and audio alert, can be for example red, if by with pedestrian's mould in the model library Type contrast is warning model, then the pedestrian contour is shown as warning color in the display screen of alarm module, for example, can be yellow, if By being security model with pedestrian dummy contrast in the model library, then what not alarm indication, wherein profile display color were realized Process is:Graphics processing unit extracts the edge of the depth image of correspondence pedestrian, by the edge transition color, then is superimposed to the depth Spend on the corresponding RGB image of image, be allowed to show that pedestrian contour is respective color, and the action movement of human body can be traced and moves It is dynamic.
Presently preferred embodiments of the present invention is described above.It is to be appreciated that the invention is not limited in above-mentioned Particular implementation, wherein the equipment and structure be not described in detail to the greatest extent are construed as giving reality with the common mode in this area Apply;Any those skilled in the art, without departing from the scope of the technical proposal of the invention, all using the disclosure above Methods and techniques content make many possible variations and modification to technical solution of the present invention, or be revised as equivalent variations etc. Embodiment is imitated, this has no effect on the substantive content of the present invention.Therefore, every content without departing from technical solution of the present invention, foundation The technical spirit of the present invention still falls within the present invention to any simple modifications, equivalents, and modifications made for any of the above embodiments In the range of technical scheme protection.

Claims (8)

1. a kind of drive recorder pedestrian monitoring system, including acquisition module, processing module, it is characterised in that
The acquisition module includes motion sensing control device, infrared sensor and depth gauge, and the motion sensing control device is imaged including RGB Head, the RGB cameras are used for the color video frequency image generation RGB image of shooting visual angle scope one skilled in the art, the infrared sensing Depth image of the device to obtain pedestrian, depth image and RGB image that the depth gauge is obtained based on the infrared sensor Generate the depth image stream of pedestrian;
The processing module includes graphics processing unit and memory cell, and described image processing unit is based on the depth image stream Rebuild pedestrian's 3-D view and the pedestrian dummy storehouse with being stored in memory cell is contrasted;Described image processing unit includes one Feature extraction unit and Characteristic Contrast unit, the feature extraction unit are used to extract the feature in pedestrian's 3-D view and pass through Characteristic Contrast unit is contrasted with the model in pedestrian dummy storehouse, and the pedestrian dummy storehouse in the memory cell is provided with feature Storehouse, posture and build of the feature database based on pedestrian dummy are respectively equipped with pose parameter θ and shape parameter β, pedestrian's mould Any of which pedestrian dummy in type storehouse is respectively provided with corresponding a pose parameter θ and shape parameter β, and pedestrian dummy posture Parameter θ and shape parameter β respectively constitute the first collision coefficient α and the second collision coefficient γ and the first collision coefficient α and second is touched Hit coefficient gamma sum and constitute a total collision coefficient α γ.
2. a kind of drive recorder pedestrian monitoring system according to claim 1, it is characterised in that also including alarm module And power module, the alarm module includes display screen and voice, and the alarm module sends alarm according to the result of contrast and carry Show;The power module is electrically connected with the acquisition module, processing module and alarm module respectively.
3. a kind of drive recorder pedestrian monitoring system according to claim 2, it is characterised in that described image processing is single Member rebuild pedestrian's 3-D view the step of be:Using multiframe sum-average arithmetic and two-sided filter by during acquisition depth image stream The noise remove of generation, is inputted the point cloud after denoising as variable model fitting algorithm;Detailed process is:Assuming that depth image Flow point cloud T and include point T (q), wherein, q ∈ [1:Q], Q is number of vertices, secondly, the build of pedestrian dummy in pedestrian dummy storehouse Parameter is β, and pose parameter is θ, and the pedestrian dummy with shape parameter β and pose parameter θ is designated as Mβ, θ, Mβ, θOn point be denoted as Mβ, θ (p), wherein p ∈ [1:P], P is the number of vertices of pedestrian dummy, based on summit T (q) in above-mentioned depth image stream and pedestrian's mould The summit M of pedestrian dummy in type storehouseβ, θ (p)Corresponding point is searched, pedestrian's 3-D view is then generated by way of Mesh Fitting, So that it is determined that the shape parameter β and pose parameter θ of the pedestrian.
4. a kind of drive recorder pedestrian monitoring system according to claim 3, it is characterised in that the shape parameter β Including being less than 90 degree when pedestrian's epigastric angle, its second collision coefficient γ is 0.1-1.5;When 90 degree of pedestrian's epigastric angle, it second is touched Coefficient gamma is hit for 1.6-2.5;When pedestrian's epigastric angle is more than 90 degree, its second collision coefficient γ is 2.6-3.5.
5. a kind of drive recorder pedestrian monitoring system according to claim 4, it is characterised in that the pose parameter θ Including being walking postures as pedestrian, its first collision coefficient α is 0.1-1.5;It is trot posture, its first collision coefficient as pedestrian α is 1.6-2.5;It is posture of running as pedestrian, its first collision coefficient α is 2.6-3.5.
6. a kind of drive recorder pedestrian monitoring system according to claim 5, it is characterised in that done what is required of social etiquette in different Under condition, described image processing unit determines the shape parameter β and pose parameter θ of the pedestrian, and generates corresponding first collision system Number α and the second collision coefficient γ, by the first collision coefficient α and the second collision coefficient γ of generation respectively with being gone in the model library The the first collision coefficient α and the second collision coefficient γ of people's model are contrasted, and are by the collisions of the first collision coefficient α and second The collision coefficient α γ of number γ sums and a certain pedestrian dummy in the model library are contrasted and sentenced according to above-mentioned comparing result Whether break has the risk of collision.
7. a kind of drive recorder pedestrian monitoring system according to claim 6, it is characterised in that in pedestrian's mould of collection In type, if the first collision coefficient α and the second collision coefficient γ is not less than 1.6 and total collision coefficient is not less than 5, for collision mould Type;If the first collision coefficient α or the second collision coefficient γ is not less than 1.6 and total collision coefficient is less than 5, for warning model;If First collision coefficient α and the second collision coefficient γ is less than 1.6, then is security model.
8. a kind of drive recorder pedestrian monitoring system according to claim 6, it is characterised in that if by with the mould Pedestrian dummy contrast is collision model in type storehouse, then the pedestrian contour is shown as dangerous color simultaneously in the display screen of the alarm module Audio alert, if by being alerted with pedestrian dummy contrast in the model library in model, the display screen of the alarm module The pedestrian contour is shown as warning color, if by being security model with pedestrian dummy contrast in the model library, not alarming aobvious Show.
CN201710230018.4A 2017-04-10 2017-04-10 A kind of drive recorder pedestrian monitoring system Pending CN107240167A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710230018.4A CN107240167A (en) 2017-04-10 2017-04-10 A kind of drive recorder pedestrian monitoring system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710230018.4A CN107240167A (en) 2017-04-10 2017-04-10 A kind of drive recorder pedestrian monitoring system

Publications (1)

Publication Number Publication Date
CN107240167A true CN107240167A (en) 2017-10-10

Family

ID=59983472

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710230018.4A Pending CN107240167A (en) 2017-04-10 2017-04-10 A kind of drive recorder pedestrian monitoring system

Country Status (1)

Country Link
CN (1) CN107240167A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107845073A (en) * 2017-10-19 2018-03-27 华中科技大学 A kind of local auto-adaptive three-dimensional point cloud denoising method based on depth map

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102096803A (en) * 2010-11-29 2011-06-15 吉林大学 Safe state recognition system for people on basis of machine vision
US20130253733A1 (en) * 2012-03-26 2013-09-26 Hon Hai Precision Industry Co., Ltd. Computing device and method for controlling unmanned aerial vehicle in flight space
CN103366565A (en) * 2013-06-21 2013-10-23 浙江理工大学 Method and system of detecting pedestrian running red light based on Kinect
CN103448670A (en) * 2012-05-31 2013-12-18 天翔系统整合股份有限公司 Vehicle monitoring system
CN105551114A (en) * 2016-01-29 2016-05-04 深圳市美好幸福生活安全系统有限公司 Automobile data recorder and method for using automobile data recorder

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102096803A (en) * 2010-11-29 2011-06-15 吉林大学 Safe state recognition system for people on basis of machine vision
US20130253733A1 (en) * 2012-03-26 2013-09-26 Hon Hai Precision Industry Co., Ltd. Computing device and method for controlling unmanned aerial vehicle in flight space
CN103448670A (en) * 2012-05-31 2013-12-18 天翔系统整合股份有限公司 Vehicle monitoring system
CN103366565A (en) * 2013-06-21 2013-10-23 浙江理工大学 Method and system of detecting pedestrian running red light based on Kinect
CN105551114A (en) * 2016-01-29 2016-05-04 深圳市美好幸福生活安全系统有限公司 Automobile data recorder and method for using automobile data recorder

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
MAX BAJRACHARYA,ETC: "Results from a Real-time Stereo-based Pedestrian Detection System on a Moving Vehicle", 《IN WORKSHOP ON PEOPLE DETECTION AND TRACKING,PROCEEDINGS OF IEEE ICRA 2009》 *
何奎莲: "《体态语——现代教师的必修课》", 31 May 2016, 西南交通大学出版 *
周瑾: "基于Kinect深度相机的三维人体重建技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
张毅,等: "《移动机器人技术基础与制作》", 31 January 2013, 哈尔滨工业大学出版社 *
范凯熹: "《信息交互设计》", 28 February 2015, 中国海洋大学出版社 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107845073A (en) * 2017-10-19 2018-03-27 华中科技大学 A kind of local auto-adaptive three-dimensional point cloud denoising method based on depth map
CN107845073B (en) * 2017-10-19 2020-02-14 华中科技大学 Local self-adaptive three-dimensional point cloud denoising method based on depth map

Similar Documents

Publication Publication Date Title
CN103858156B (en) Vehicle vicinity monitoring device
CN106991377A (en) With reference to the face identification method, face identification device and electronic installation of depth information
CN102203837B (en) System for monitoring the area around a vehicle
CN107563304A (en) Unlocking terminal equipment method and device, terminal device
CN104899869A (en) Plane and barrier detection method based on RGB-D camera and attitude sensor
CN106650708A (en) Visual detection method and system for automatic driving obstacles
CN105718863A (en) Living-person face detection method, device and system
CN101587544B (en) Based on the carried on vehicle antitracking device of computer vision
CN109643366A (en) For monitoring the method and system of the situation of vehicle driver
CN107704805B (en) Method for detecting fatigue driving, automobile data recorder and storage device
CN101877051A (en) Driver attention state monitoring method and device
JP2003189293A (en) Device for displaying state of surroundings of vehicle and image-providing system
CN104766071A (en) Rapid traffic light detection algorithm applied to pilotless automobile
CN102099842B (en) Vehicle vicinity monitoring device
CN109636924A (en) Vehicle multi-mode formula augmented reality system based on real traffic information three-dimensional modeling
CN109145750A (en) A kind of driver identity rapid authentication method and system
CN104881956A (en) Fatigue driving early warning system
CN203838770U (en) Automobile data recorder
CN108021899A (en) Vehicle intelligent front truck anti-collision early warning method based on binocular camera
CN106412420A (en) Interactive photographing implementation method
CN109147037A (en) Effect processing method, device and electronic equipment based on threedimensional model
CN107240167A (en) A kind of drive recorder pedestrian monitoring system
CN106447828A (en) Automobile data recorder with iris recognition function
Rezaee et al. Real-time intelligent alarm system of driver fatigue based on video sequences
CN105844227A (en) Driver identity authentication method for school bus safety

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20171010