CN108805910A - More mesh Train-borne recorders, object detection method, intelligent driving system and automobile - Google Patents
More mesh Train-borne recorders, object detection method, intelligent driving system and automobile Download PDFInfo
- Publication number
- CN108805910A CN108805910A CN201810557384.5A CN201810557384A CN108805910A CN 108805910 A CN108805910 A CN 108805910A CN 201810557384 A CN201810557384 A CN 201810557384A CN 108805910 A CN108805910 A CN 108805910A
- Authority
- CN
- China
- Prior art keywords
- picture
- wide angle
- camera
- target
- stereo
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/292—Multi-camera tracking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/32—Determination of transform parameters for the alignment of images, i.e. image registration using correlation-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
- G06T7/85—Stereo camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
This application discloses a kind of more mesh Train-borne recorders, object detection method, intelligent driving system and automobiles, more mesh Train-borne recorders include wide angle camera, binocular solid camera, image analysis processor and memory, wherein, wide angle camera is respectively used to the wide angle picture and stereo-picture of acquisition vehicle front target with binocular solid camera；Image analysis processor carries out target detection to stereo-picture, obtain the target range information of detection target, and the target identification information of detection target is obtained by the detection target re-projection to wide angle picture in stereo-picture according to the image registration results of wide angle picture and stereo-picture.More mesh Train-borne recorders obtain accurate detection target and target range information using the advantage of binocular solid camera, and it will be in the detection target re-projection to wide angle picture in stereo-picture, the target identification information of display detection target so that more mesh Train-borne recorders have both that the wide angle camera visual field is wide and the accurate advantage of binocular solid camera ranging.
Description
Technical field
This application involves automobile assistant driving technical field more particularly to a kind of more mesh Train-borne recorders, target detection sides
Method, intelligent driving system and automobile.
Background technology
Train-borne recorder includes camera and car-mounted terminal at present, and camera is typically mounted at vehicle front, is acquired by camera
Vehicle front road image, car-mounted terminal detects the barrier of vehicle front road by the road image that camera acquires, to driving
The personnel of sailing send out early warning.
Train-borne recorder includes the Train-borne recorder based on wide angle camera and the vehicle-mounted note based on binocular solid camera at present
Record instrument, wherein the Train-borne recorder based on wide angle camera is mostly the camera with wide-angle lens optical system using one, is had
The characteristics of big visual angle, small blind area, becomes the mainstream scheme of current Train-borne recorder, and more and more auxiliary drive function and are added
In Train-borne recorder, including pedestrian detection, vehicle identification, Lane detection, traffic lights identification etc.；And it is based on binocular solid
The Train-borne recorder of camera is to use binocular solid camera, is shot in front of running car by two cameras at different location
Road image, car-mounted terminal can calculate the range information for obtaining barrier on image by the parallax of two images, according to distance
Information accurately sends out early warning.
But the usual pattern distortion of Train-borne recorder based on wide angle camera is larger, can not be provided for automobile assistant driving
Accurate range information carries out target detection merely with two-dimensional signals such as the color of target, texture, shapes, is easily illuminated by the light, shade
The influence of equal image-forming conditions, missing inspection flase drop situation are more；And wide angle camera is suitable only for the target detection of middle short distance, Wu Fajian
Distant object is surveyed, and the auxiliary of not applicable high speed traveling drives early warning demand.And the vehicle-mounted note based on binocular solid camera
Instrument is recorded because using the smaller long focal length lens optical system of distorting, causing the visual angle that binocular solid camera covers smaller, blind area compared with
Greatly, it cannot be satisfied the primary demand of big field-of-view image acquisition and record.Therefore, how to provide a kind of with big visual angle, small blind area
And the Train-borne recorder that can precisely detect target becomes those skilled in the art's urgent problem to be solved.
Invention content
This application provides a kind of more mesh Train-borne recorders, object detection method, intelligent driving system and automobiles, to solve
A kind of the technical issues of having big visual angle, small blind area and capable of accurately detecting the Train-borne recorder of target, can not be provided at present.
In order to solve the above-mentioned technical problem, the embodiment of the present application discloses following technical solution：
In a first aspect, the embodiment of the present application discloses a kind of more mesh Train-borne recorders, including wide angle camera, binocular solid phase
Machine, image analysis processor and memory, wherein
The wide angle camera, the wide angle picture for acquiring vehicle front target；
The binocular solid camera, the stereo-picture for acquiring vehicle front target；
Described image analysis processor obtains the target of detection target for carrying out target detection to the stereo-picture
Range information；For carrying out image registration to the wide angle picture and stereo-picture, according to image registration results by the solid
In detection target re-projection to the wide angle picture in image, the target identification information of the detection target is obtained；
The memory, for storing the wide angle picture, stereo-picture, target identification information and target range information.
Optionally, described image analysis processor includes image registration device, and described image aligner is used for the wide-angle
The characteristic information of image zooming-out and the characteristic information of stereo-picture extraction carry out characteristic matching, and institute is determined according to characteristic matching result
State the coordinate transformation relation of wide angle picture and corresponding pixel points in stereo-picture, wherein the coordinate transformation relation is to institute
It states stereo-picture and carries out image registration with wide angle picture.
Optionally, described image analysis processor further includes re-projection device, and the re-projection device is used for according to the coordinate
Detection target projection in the stereo-picture is converted into the wide angle picture by transformational relation, shows the detection target
Target identification information.
Optionally, described image analysis processor further includes camera calibration device, and the camera calibration device is used for described wide
Angle camera is demarcated with stereoscopic camera, is assembled, so that the visual field overlapping region position of the wide angle camera and binocular solid camera
In the central area of the wide angle picture.
Second aspect, the embodiment of the present application disclose a kind of mesh being registrated with binocular solid camera image based on wide angle camera
Detection method is marked, including：
The wide angle picture and binocular solid camera for obtaining wide angle camera acquisition vehicle front target acquire vehicle front target
Stereo-picture；
Target detection is carried out to the stereo-picture, the target range for obtaining detection target to the binocular solid camera is believed
Breath；
Image registration is carried out to the stereo-picture and wide angle picture, according to described image registration result by the stereogram
In the detection target re-projection to the wide angle picture as in, the target identification information of the detection target is obtained.
Optionally, image registration is carried out to the stereo-picture and wide angle picture, including：
The interested of the wide angle picture is determined according to the visual field overlapping region of the wide angle camera and binocular solid camera
Region；
Extract the characteristic information of area-of-interest described in the wide angle picture and the characteristic information in the stereo-picture；
Characteristic matching is carried out to the characteristic information and the characteristic information of stereo-picture extraction of wide angle picture extraction；
Determine that the wide angle picture and the coordinate of corresponding pixel points in stereo-picture are converted according to the characteristic matching result
Relationship, wherein the coordinate transformation relation is to the stereo-picture and wide angle picture progress image registration.
Optionally, according to described image registration result by the detection target re-projection in the stereo-picture to described
In wide angle picture, the target identification information of the detection target is obtained, including：
The detection target projection in the stereo-picture is transformed into the wide angle picture according to the coordinate transformation relation
In；
The target identification information of the detection target is shown by the wide angle picture.
Optionally, before the wide angle picture and the binocular solid camera acquisition automobile that obtain wide angle camera acquisition vehicle front target
Square mesh target stereo-picture includes before：
The wide angle camera and binocular solid camera are demarcated respectively；
Calibrated wide angle camera and binocular solid camera are assembled, so that the wide angle camera and binocular solid phase
The visual field overlapping region of machine is located at the central area of the wide angle picture；
Acquire the image of vehicle front target respectively with binocular solid camera by the wide angle camera after assembly.
The third aspect, the embodiment of the present application disclose a kind of intelligent driving system, including more mesh vehicles described in first aspect
Recorder is carried, more mesh Train-borne recorders carry out target detection by the object detection method described in second aspect.
Fourth aspect, the embodiment of the present application disclose a kind of intelligent driving automobile, including the intelligence described in the third aspect is driven
Sail system.
Compared with prior art, the application has the beneficial effect that：
A kind of more mesh Train-borne recorders of the embodiment of the present application offer, object detection method, intelligent driving system and automobile, should
More mesh Train-borne recorders include wide angle camera, binocular solid camera, image analysis processor and memory, wherein wide angle camera
Wide angle picture for acquiring vehicle front target；Binocular solid camera is used to acquire the stereo-picture of vehicle front target；Figure
As analysis processor is used to carry out target detection, the target range information of acquisition detection target to stereo-picture；For to wide-angle
Image and stereo-picture carry out image registration, according to image registration results by the target information re-projection detected in stereo-picture extremely
In wide angle picture, the target range information of detection target is obtained；Memory is known for storing wide angle picture, stereo-picture, target
Other information and target range information.More mesh Train-borne recorders provided by the embodiments of the present application, with the vertical of binocular solid camera acquisition
Main Basiss of the body image as target detection carry out target detection to stereo-picture, can using the advantage of binocular solid camera
Accurate detection target and target range information are obtained, and image registration is carried out to stereo-picture and wide angle picture, according to figure
As registration result in the detection target re-projection to wide angle picture in stereo-picture, will obtain the identification information of detection target, profit
Has the advantages that the wide visual field with wide angle camera, therefore more mesh Train-borne recorders provided by the present application can have both the wide angle camera visual field
The accurate advantage of wide advantage and binocular solid camera ranging can be provided for DAS (Driver Assistant System) and more be improved accurately completely
Image information can be used for the auxiliary such as front truck anti-collision warning, lane line deviation and drive field.
It should be understood that above general description and following detailed description is only exemplary and explanatory, not
The application can be limited.
Description of the drawings
In order to illustrate more clearly of the technical solution of the application, letter will be made to attached drawing needed in the embodiment below
Singly introduce, it should be apparent that, for those of ordinary skills, without creative efforts, also
It can be obtain other attached drawings according to these attached drawings.
Fig. 1 is a kind of structure composition schematic diagram of more mesh Train-borne recorders provided by the embodiments of the present application；
Fig. 2 is the structural schematic diagram of image analysis processor in more mesh Train-borne recorders provided by the embodiments of the present application；
Fig. 3 is pinhole camera model schematic；
Fig. 4 is the schematic diagram before more mesh Train-borne recorder neutral body image stereo calibrations provided by the embodiments of the present application；
Fig. 5 is the schematic diagram after more mesh Train-borne recorder neutral body image stereo calibrations provided by the embodiments of the present application；
Fig. 6 is the stereo calibration design sketch of more mesh Train-borne recorder neutral body images provided by the embodiments of the present application；
Fig. 7 is a kind of dress of wide angle camera and binocular solid camera in more mesh Train-borne recorders provided by the embodiments of the present application
With schematic diagram；
Fig. 8 is wide angle camera and binocular solid camera weight after being assembled in more mesh Train-borne recorders provided by the embodiments of the present application
Folded area schematic；
Fig. 9 is that the feature registration of wide angle picture and stereo-picture is shown in more mesh Train-borne recorders provided by the embodiments of the present application
It is intended to；
Figure 10 is showing for more mesh Train-borne recorder neutral body images provided by the embodiments of the present application and wide angle picture re-projection
It is intended to；
Figure 11 is a kind of target being registrated with binocular solid camera image based on wide angle camera provided by the embodiments of the present application
The flow chart of detection method；
Figure 12 is the target detection provided by the embodiments of the present application being registrated with binocular solid camera image based on wide angle camera
The detail flowchart of S300 in method.
Specific implementation mode
In order to make those skilled in the art better understand the technical solutions in the application, below in conjunction with the application reality
The attached drawing in example is applied, technical solutions in the embodiments of the present application is clearly and completely described, it is clear that described implementation
Example is only some embodiments of the present application, rather than whole embodiments.Based on the embodiment in the application, the common skill in this field
The every other embodiment that art personnel are obtained without making creative work should all belong to the application protection
Range.
With vehicle intellectualized development trend, just the grinding as academia and industrial quarters such as auxiliary drivings, automatic Pilot
Study carefully hot spot, lot of domestic and foreign orthodox car manufacturer, high-tech enterprise etc. are devoted to release respective solution.In recent years,
Scheme based on millimetre-wave radar and laser radar is widely used for high-end automobile, and the main advantage of millimetre-wave radar is embodied in
With the very strong ability for resisting environmental disturbances, can penetrating fog, cigarette, dust etc., the ability with the work of round-the-clock round-the-clock can
Directly detect the distance and speed of front vehicles；The main advantage of laser radar is embodied in pole accurately range capability and superelevation point
Resolution.However, whether millimetre-wave radar or laser radar, there are some fatal defects, for example, vertical field of view angle it is relatively narrow,
Longitudinal resolution is relatively low, can not provide color and texture information etc., and these information are extremely important for many other tasks
, such as pedestrian detection, vehicle identification, Traffic Sign Recognition etc..In comparison, the method for view-based access control model can be very good more
The deficiency for mending radar sensor, is widely applied in reality, and especially Train-borne recorder etc. is just becoming the indispensability of private vehicle
Product.
Current Train-borne recorder is mostly the camera with wide-angle lens optical system using one, has big visual angle, small blind
The features such as area, becomes the mainstream scheme of current Train-borne recorder, and more and more auxiliary drive function and Train-borne recorder is added
In, including pedestrian detection, vehicle identification, Lane detection, traffic lights identification etc..But wide-angle lens usually distortion compared with
Greatly, it is difficult to accurate measurement distance is only capable of carrying out target detection using two-dimensional signals such as the color of target, texture, profile and shapes,
The image-forming conditions such as its testing result is easily illuminated by the light, shade influence, and missing inspection flase drop is more, and larger visual angle causes distant object to exist
It can not almost be observed in image, therefore wide angle camera is suitable only for the target detection of middle short distance, be not particularly suited for running at high speed
Auxiliary drive early warning demand.And the Train-borne recorder based on binocular solid camera then carries out more target using three-dimensional information
Accurately detection, but binocular solid camera avoids the error brought due to lens optical distortion etc. for accurate measurement distance,
Generally use distorts smaller long focal length lens optical system, and telephoto lens not only distorts smaller but also remote in being relatively more suitable for
The target detection etc. of distance, but this field angle for also causing stereoscopic camera to cover is smaller, blind area is larger, cannot be satisfied the big visual field
The primary demand of Image Acquisition and record.
To solve the above-mentioned problems, the embodiment of the present application provides a kind of more mesh Train-borne recorders, more mesh vehicle-mounted recordings
Image registration of the instrument based on wide angle camera Yu binocular solid camera carries out target detection, to binocular solid phase using stereoscopic vision
The stereo-picture of machine acquisition carries out target detection, and relative two dimensional image detection accuracy is more preferable, and utilizes wide angle camera and binocular
The image registration techniques of stereoscopic camera so that Train-borne recorder has both the wide advantage in the wide angle camera visual field and binocular solid camera is surveyed
Away from accurate advantage.
It is a kind of structure composition schematic diagram of more mesh Train-borne recorders provided by the embodiments of the present application referring to Fig. 1.
As shown in Figure 1, more mesh Train-borne recorders provided by the embodiments of the present application include wide angle camera 1, binocular solid camera
2, image analysis processor 3 and memory 4, wherein
Wide angle camera 1 is used to acquire the wide angle picture of vehicle front target, and wide angle picture is transmitted at image analysis
Manage device 3；Binocular solid camera 2 is used to acquire the stereo-picture of vehicle front target, and stereo-picture is transmitted to image analysis
Processor 3；Image analysis processor 3 be used for stereo-picture carry out target detection, and by stereo-picture detection target with
Wide angle picture carries out matching calibration, obtains target range information and target identification information；Memory 4 for store wide angle picture,
Stereo-picture, target identification information and target range information, convenient for users to subsequently checking.
In the embodiment of the present application, more mesh Train-borne recorders be mounted on intelligent driving automobile on, specifically, wide angle camera 1 with
Binocular solid camera 2 is mounted on the front upper place of body of a motor car, to shoot the front situation of automobile.Since wide angle camera 1 has
Wider visual angle can accommodate more scenery range in limited distance, can also carry out rough object positioning；Binocular solid
2 field range of camera is relatively narrow, but for its object within the vision, can accurately be positioned relatively, the application carries
The more mesh Train-borne recorders supplied include wide angle camera 1 and binocular solid camera 2, by the analyzing processing of image analysis processor 3,
The advantages of may make more mesh Train-borne recorders to have both the wide wide angle camera visual field and binocular solid camera precise positioning.
In the embodiment of the present application, wide angle camera 1 and binocular solid camera 2 each independently with image analysis processor 3 it
Between by being wirelessly connected or being transmitted by wired connection, to which wide angle picture and stereo-picture are transmitted to image analysis processing
Device.Connection type between wide angle camera 1 or binocular solid camera 2 and image analysis processor 3 is those skilled in the art's root
Selection can be carried out according to needs, respective advantage and disadvantage are those skilled in the art can weigh as needed.
In the embodiment of the present application, as shown in Fig. 2, image analysis processor 3 includes camera calibration device 31, stereo-picture processing
Device 32, image registration device 33 and re-projection device 34, wherein camera calibration device 31 is used for wide angle camera 1 and binocular solid camera 2
It is demarcated, obtains the internal reference of wide angle camera 1, the interior participation distortion parameter of binocular solid camera 2.
In the application example, before wide angle camera 1 and binocular solid camera 2 carry out Stereo matching and target detection, need
Wide angle camera 1 and binocular solid camera 2 are demarcated, including the plane reference of wide angle camera 1 and binocular solid camera 2
Plane reference and stereo calibration.What camera calibration mainly solved is from world coordinates XwYwZwTo camera coordinates XcYcZc, then arrive figure
As physical coordinates xyz, then to the Transformation Relation of Projection between image coordinate uv, as shown in figure 3, obtain projection matrix, i.e. camera
Internal reference (inner parameter) and outer ginseng (external parameter).The meaning of inner parameter includes that image is sat from camera coordinates system to image plane
Mark system transformation parameter and shooting image when distortion factor, in the case that external parameter is embodied in more camera lenses, each camera lens it
Between position relationship etc..
Using the geometrical relationship of pinhole imaging system, the transformational relation that can obtain image coordinate and world coordinates is as follows：
In formula, s --- twist factor；
K --- Intrinsic Matrix, by parameter ku, kv, s, u0, v determinations, these parameters and the internal structure of video camera have
It closes；And ku=f/dx, kv=f/dy；
The external parameter square of [R T] --- video camera, is determined by the position of video camera；
P --- projection matrix is 3x4 matrixes.
Pin-hole model is a kind of ideal model, also has some differences with actual camera model.Actual camera shooting
There are lens distortions for machine model.Lens distortion generally comprises radial distortion, centrifugal distortion and thin prism and distorts three kinds.Radial distortion
Distortion only is generated in radial position, and there is more than the distortion of radial position, also tangential positions for centrifugal distortion and thin prism distortion
It sets.
The mathematical model of radial distortion can be expressed as：
In above formula,--- the distance of imaging point to picture centre；
k1, k2, k3... --- coefficient of radial distortion.
The mathematical model of centrifugal distortion can be expressed as：
Wherein, p1, p2--- tangential distortion coefficient.
The mathematical model of thin prism distortion can be expressed as：
In above formula, s1, s2--- thin prism distortion factor.
Assuming that actual image physical coordinates are (xd,yd), ideal image physical coordinates are (xu,yu), then school of distorting
Transformational relation between just front and back image physical coordinates is：
Wherein, distortion parameter includes k1, k2, k3, p1, p2And s1, s2。
It in practice, can be according to actual conditions to distortion since present camera design can reach very high precision
Parameter carries out centainly ignoring simplification, such as higher-priced camera lens, and thin prism distortion very little can be ignored substantially, radial abnormal
Become the distortion factor etc. for only considering preceding two rank, to simplify the complexity calculated.
The embodiment of the present application uses plane gridiron pattern standardization, acquires the plane reference plate figure of at least two different angles
Picture, to scaling board image zooming-out angle point, demarcate respectively the left and right camera of wide angle camera and binocular solid camera inner parameter and
Distortion parameter；Then limit of utilization constraint carries out the stereo calibration of binocular solid camera, obtains the intrinsic square of binocular solid camera
Battle array, basis matrix, spin matrix and translation matrix so that after the left and right picture elimination distortion of binocular solid camera in the horizontal direction
Close alignment, two width pictures to polar curve just in the same horizontal line, any point on such width picture is with it another
Match point on piece image centainly on a same row, as shown in Figure 4, Figure 5.
The stereo-picture that the left and right camera of binocular solid camera 2 acquires after stereo calibration is as shown in fig. 6, left camera
Any point on the stereo-picture of acquisition is with its match point on the stereo-picture that right camera acquires on a same row (as schemed
Shown in middle white line).
It,, be to wide-angle when predetermined situations occur other than carrying out camera calibration when initialization in the embodiment of the present application
Camera 1 is re-scaled with binocular solid camera 2.Such as predetermined situations include at least one of following item or its group
It closes：The predetermined time is seen in the past away from last time calibration；The orientation of any of binocular solid camera, wide angle camera changes；Week
Change etc. beyond predetermined threshold occurs for the scene for enclosing things.
After camera calibration device 31 demarcates wide angle camera 1 and binocular solid camera 2, need to wide angle camera 1 with it is double
Mesh stereoscopic camera 2 is reasonably assembled, so that wide angle camera 1 and the visual field overlapping region of binocular solid camera 2 are located at wide-angle
The near central regions of camera.The field range of wide angle camera 1 is more than the field range of binocular solid camera 2, and wide angle camera 1
It, can be by wide since the optical center of wide angle camera 1 and its distortion nearby are smaller with the visual field of binocular solid camera 2 there are Chong Die
The reasonable assembly of angle camera 1 and binocular solid camera 2, to ensure the visual field overlapping region of wide angle camera 1 and binocular solid camera 2
Positioned at the near central regions of wide angle camera.
In the embodiment of the present application, wide angle camera 1 can be located at the lower section of binocular solid camera 2, as shown in fig. 7, wide angle camera 1
It is integral structure with binocular solid camera 2, wherein binocular solid camera 2 includes left mesh camera module and right mesh camera module,
Wide angle camera 1 is set between left mesh camera module and right mesh camera module, and wide angle camera 1 is located at binocular solid camera 2
Lower section.But, alternatively, wide angle camera 1 can also be split structure with binocular solid camera 2.
Wide angle camera 1 may be alternatively located at the centre position of binocular solid camera 2 or the arbitrary side of left and right sides, wide angle camera
1 may be alternatively located at the lower section of binocular solid camera 2, may be a fixed connection between wide angle camera 1 and binocular solid camera 2, also may be used
To be movable connection (such as wide angle camera is rotatable, can move up and down), i.e., the embodiment of the present application is to wide angle camera and binocular solid
The specific assembly method of camera is not required with relative position design, as long as can guarantee wide angle camera and binocular solid camera
Visual field overlapping region is located at the near central regions of wide angle camera, as shown in Figure 8.
After camera calibration device 31 demarcates wide angle camera 1 and binocular solid camera 2,32 basis of stereo image processor
The interior participation distortion parameter of binocular solid camera 2 carries out distortion to stereo-picture and handles, and removes the wide angle picture after distorting and stands
Body image is registrated by image registration device 33, and according to image registration results by the detection target re-projection in stereo-picture
Into wide angle picture, the target identification of detection target is realized.
The image registration of the embodiment of the present application uses the method for registering of feature based, as shown in figure 9, extracting wide-angle figure respectively
As the characteristic information with stereo-picture, characteristic matching is carried out to the characteristic information of extraction, in conjunction with wide angle camera 1 and binocular solid phase
The inner parameter and distortion parameter of machine 2 realize that stereo-picture and the coordinate of corresponding pixel points in wide angle picture are converted, and realize wide-angle
The image registration of image and stereo-picture.Preferably, in order to reduce invalid feature extraction and search package space, the present embodiment
Area-of-interest (region of interesting, ROI) is set to wide angle picture, only carrying out feature to ROI region carries
It takes.ROI region determined according to the visual field overlapping range of wide angle camera and binocular solid camera, is substantially regarded with stereo-picture
Wild range is identical, and positioned at the optical center near zone of wide angle picture.
After image registration device 33 carries out image registration to wide angle picture and stereo-picture, image analysis processor 3 is to solid
Image carries out Stereo matching, and the target of Stereo matching is to find matched respective pixel in the left images that two rows correct
Point calculates the parallax of these corresponding pixel points by calculating x coordinate of these corresponding pixel points in the two images of left and right, finally
Export a disparity map.The left images of binocular solid camera for going distortion and stereo calibration to cross are matched, may be used
Solid matching method based on part, the solid matching method based on global solid matching method or based on half overall situation, are adopted
Which kind of solid matching method to have no effect on the essence of the present embodiment with.
Stereo Matching Algorithm is a kind of common technology in computer vision, and target is from two dimensional image or image sequence
The middle understanding and understanding for obtaining external world utilize two dimensional image to restore the aggregate information of object in three-dimensional environment, such as
Shape, position, posture, movement etc., and can describe, identify and connect.
Stereo image processor 32 is additionally operable to carry out target detection to stereo-picture, obtains the target range letter of detection target
Breath.Object detection method of the tradition based on two dimensional image is only capable of the two-dimensional signals such as color, texture, shape using target, easily by
The environment such as illumination, shade influence, and cause testing result inaccurate, missing inspection and flase drop situation are serious.And stereo-picture is stood
Body matching after, two dimensional image is reverted into 3-D view, using the object detection method based on 3-D view to stereo-picture into
Row target detection obtains the range information of detection target, and precise positioning is carried out to detection target.
Image analysis processor 3 further includes re-projection device 34, and re-projection device 34 is according to image registration results by stereo-picture
In detection target projection to wide angle picture in, obtain detection target target identification information.Target inspection is carried out to stereo-picture
It, will be in stereo-picture according to image registration results (coordinate transformation relation of wide angle picture pixel corresponding with stereo-picture) after survey
Detection target projection transform in wide angle picture, as shown in Figure 10, to show in wide angle picture detect target target
Identification information, such as detect shape, the color information of target.
More mesh Train-borne recorders provided by the embodiments of the present application are in addition to having the Image Acquisition of general Train-borne recorder and guarantor
It deposits outside function, is also equipped with the functions such as target detection, using the field range of binocular solid camera as detection early warning range, by wide-angle
The field range of camera is as monitoring record range, particularly suitable in automobile assistant driving, being provided for DAS (Driver Assistant System)
Accurately completely image information is more improved, can be used for the auxiliary such as front truck anti-collision warning, lane line deviation and drive field.
More mesh Train-borne recorders provided by the embodiments of the present application include wide angle camera, binocular solid camera, at image analysis
Manage device and memory, wherein wide angle camera is used to acquire the wide angle picture of vehicle front target；Binocular solid camera is for acquiring
The stereo-picture of vehicle front target；Image analysis processor is used to carry out target detection to stereo-picture, obtains detection target
Target range information, image registration is carried out to wide angle picture and stereo-picture, will be in stereo-picture according to image registration results
Detection target re-projection to wide angle picture in, obtain detection target target identification information, specifically, image analysis processor
Including image registration device, feature of the image registration device for characteristic information and stereo-picture extraction for being extracted to wide angle picture
Information carries out characteristic matching, and the coordinate of the wide angle picture and corresponding pixel points in stereo-picture is determined according to characteristic matching result
Transformational relation, wherein coordinate transformation relation is to stereo-picture and wide angle picture progress image registration；Memory is for storing
Wide angle picture, stereo-picture, target identification information and target range information.More mesh Train-borne recorders provided by the present application are based on wide
Angle camera and binocular solid camera are going the stereo-picture of distortion enterprising using stereo-picture as the Main Basiss of target detection
Row target detection obtains accurate detection target and target range information using the advantage of binocular solid camera, and to solid
Image and wide angle picture carry out image registration, according to image registration results by the detection target projection in stereo-picture to wide-angle figure
As in, the identification information of detection target is obtained, has the advantages that the wide visual field using wide angle camera.More mesh Train-borne recorders use
Stereoscopic vision carries out target detection, and relative two dimensional image detection accuracy is more preferable, and utilizes wide angle camera and binocular solid camera
Image registration techniques so that the wide advantage in the Train-borne recorder spacing wide angle camera visual field and binocular solid camera ranging are accurate
Advantage.
Based on more mesh Train-borne recorders provided by the embodiments of the present application, the embodiment of the present application also provides one kind being based on wide-angle
The object detection method that camera is registrated with binocular solid camera image.
As shown in figure 11, the target provided by the embodiments of the present application being registrated with binocular solid camera image based on wide angle camera
Detection method includes：
S100：The wide angle picture and binocular solid camera for obtaining wide angle camera acquisition vehicle front target acquire vehicle front
The stereo-picture of target.
In the embodiment of the present application, wide angle camera is mounted on binocular solid camera on body of a motor car, before shooting automobile
Square road conditions need to carry out some initialization operations, including to wide-angle before wide angle camera and the work of binocular solid camera
Camera and binocular solid camera are demarcated, and the internal reference of wide angle camera, the internal reference and distortion parameter of binocular solid camera are obtained, double
The internal reference of mesh stereoscopic camera includes the internal reference of the internal reference and right mesh camera of left mesh camera, and internal reference goes to distortion processing institute for image
It needs；Distortion parameter includes the relevant outer ginseng of position relationship between left mesh camera and right mesh camera.
After wide angle camera and binocular solid camera calibration, it is also necessary to calibrated wide angle camera and binocular solid camera
It is assembled, since the field range of wide angle camera is more than the field range of binocular solid camera, the field range of the two exists
Overlapping region, and the optical center of wide angle camera and its distortion nearby are smaller, it is therefore desirable to wide angle camera and binocular solid camera into
Row rationally assembly, with ensure the visual field overlapping region of wide angle camera and binocular solid camera be located at wide angle camera central area it is attached
Closely.
S200：Target detection is carried out to stereo-picture, obtains the target range information for detecting target to binocular solid camera.
The embodiment of the present application carries out target detection using the object detection method based on 3-D view to stereo-picture, obtains
Target is detected to the target range information of binocular solid camera, precise positioning is carried out to the detection target in stereo-picture.It can adopt
The target range information that target is detected in stereo-picture is obtained with stereo matching method, specifically, to the acquisition of binocular solid camera
Stereo-picture is detected into row distance, and the distance for obtaining each pixel may include：With Stereo Matching Algorithm to the image of acquisition into
Row matching, obtains respective pixel range difference in the picture, by the internal reference and distortion parameter of binocular solid camera calculate to
The distance of each pixel in the image of a few acquisition.
Object detection method based on 3-D view takes full advantage of the information such as the two-dimensional color of target, texture and three-dimensional
Range information etc. can effectively overcome the influence to testing result such as illumination, shade.In the embodiment of the present application, it is based on 3-D view
Object detection method can be the object detection method based on disparity map, can also be the target detection side based on three-dimensional point cloud
Method, common methods include the target detection based on UV figures, the target detection based on gridding method, the target detection side based on RGBD figures
Method, the target detection etc. based on convolutional neural networks.
Technology customary in the art, this non-the application can be used to want in relation to Stereo Matching Algorithm and object detection method
Point, details are not described herein again.
S300：Image registration is carried out to stereo-picture and wide angle picture, it will be in stereo-picture according to image registration results
It detects in target re-projection to wide angle picture, obtains the target identification information for stating detection target.
Using the object detection method based on 3-D view to stereo-picture carry out target detection after, using wide angle camera with
The image registration results of binocular solid camera transform to the detection target projection in stereo-picture in wide angle picture, thus
Display target identification information in wide angle picture.Image re-projection refers to being projected by the reference picture to arbitrary viewpoint to produce
Raw new image is since image re-projection can change the direction of visual lines of generated image, therefore it is in multi-view image
The fields such as real-time rendering, View morph, image mosaic applied.
It obtains wide angle camera and the method for the image registration results of binocular solid camera is as shown in figure 12.
S301：The region of interest of wide angle picture is determined according to the visual field overlapping region of wide angle camera and binocular solid camera
Domain.
Wide angle camera and binocular solid camera are carried out after rationally assembling, the visual field weight of wide angle camera and binocular solid camera
Folded region is located at the near central regions of wide angle camera, and the distortion of the central area is smaller, thus according to wide angle camera with pair
The visual field overlapping region of mesh stereoscopic camera determines the area-of-interest (ROI region) of wide angle picture, only analyzes the spy in ROI region
Reference ceases.
S302：Extract the characteristic information of area-of-interest and the characteristic information in stereo-picture in wide angle picture.
Since the central area distortion of wide angle picture is smaller, and the distortion in other regions is larger, therefore only to wide angle picture
ROI region carry out feature information extraction, extract point, line, edge in a left side (or right) image of wide angle picture and stereo-picture
Equal characteristic informations.The algorithm for extracting feature includes but not limited to harris algorithms, susan algorithms, canny algorithms, SIFT, SURF
Deng, can go distortion image on carry out feature extraction, feature extraction can also be carried out on original image.
S303：Characteristic matching is carried out to the characteristic information and the characteristic information of stereo-picture extraction of wide angle picture extraction.
Characteristic matching, estimation are carried out to the characteristic information extracted in the characteristic information and stereo-picture that are extracted in wide angle picture
Spin matrix, translation matrix, eigenmatrix and basis matrix between wide angle camera and binocular solid camera etc..It is missed to reduce
Difference promotes registration accuracy, generally also needs to carry out registration optimization, what optimization method can for least square method, linearly or nonlinearly
LM (Levenberg-Marquardt) method etc..
S304：Determine that wide angle picture and the coordinate of corresponding pixel points in stereo-picture convert pass according to characteristic matching result
System.
After carrying out characteristic matching to the characteristic information of extraction, in conjunction with the inner parameter of wide angle camera and binocular solid camera and
Distortion parameter realizes that stereo-picture and the coordinate of corresponding pixel points in wide angle picture are converted, obtains wide angle picture and stereo-picture
Coordinate transformation relation, the coordinate transformation relation to the stereo-picture and wide angle picture carrying out image registration.
S305：The detection target projection in stereo-picture is transformed into wide angle picture according to coordinate transformation relation.
S306：The target identification information of detection target is shown by wide angle picture.
Using the object detection method based on 3-D view to stereo-picture carry out target detection after, according to wide angle picture with
Detection target projection in stereo-picture is transformed into wide angle picture by the coordinate transformation relation of stereo-picture, and detection target is existed
It is shown in wide angle picture, to obtain the target identification information of detection target, such as the shape of target, color information.
Using the image registration parameter of wide angle camera and binocular solid camera, by the detection target weight of stereo-picture coordinate system
It projects in wide angle picture coordinate system.The application can only calculate re-projection coordinate of the target frame vertex on wide angle picture,
Projection frame of the target frame on wide angle picture can be calculated.
According to more mesh Train-borne recorders provided by the embodiments of the present application, the embodiment of the present application also provides a kind of intelligent drivings
System comprising above-mentioned more mesh Train-borne recorders, more mesh Train-borne recorders are based on wide angle camera and binocular solid using above-mentioned
The object detection method of camera image registration carries out target detection.
Brain of the intelligent driving system as intelligent driving automobile is believed using the target identification that more mesh Train-borne recorders provide
Breath and target range information to make intelligent driving decision, and send out control instruction to execution unit is driven, such as turn to, slow down,
The instructions such as lane-change, brake.
According to intelligent driving system provided by the embodiments of the present application, the embodiment of the present application also provides a kind of intelligent driving vapour
Vehicle, including above-mentioned intelligent driving system.More mesh Train-borne recorders provided by the present application are particularly suitable in intelligent driving system
For the identification of target, the target in a wide range of can either be identified, and can identify the target range within the scope of emphasis, it can
Preferable input data is provided for intelligent driving system, ensures the safe driving of intelligent driving automobile.
Since embodiment of above is that reference combination illustrates on other modes, have between different embodiments
There is an identical part, just to refer each other for identical, similar part between each embodiment in this specification.Herein no longer in detail
It illustrates.
It should be noted that in the present specification, such as the terms "include", "comprise" or its any other variant are intended to
Cover non-exclusive inclusion, so that circuit structure, article or equipment including a series of elements include not only those
Element, but also include other elements that are not explicitly listed, or further include for this circuit structure, article or equipment
Intrinsic element.In the absence of more restrictions, there is the element that sentence "including a ..." limits, it is not excluded that
There is also other identical elements in circuit structure, article or equipment including the element.
Those skilled in the art will readily occur to its of the application after considering specification and putting into practice the disclosure invented here
His embodiment.This application is intended to cover the present invention any variations, uses, or adaptations, these modifications, purposes or
Person's adaptive change follows the general principle of the application and includes the undocumented common knowledge in the art of the application
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the application are wanted by right
The content asked is pointed out.
Above-described the application embodiment does not constitute the restriction to the application protection domain.
Claims (10)
1. a kind of more mesh Train-borne recorders, which is characterized in that including wide angle camera, binocular solid camera, image analysis processor
And memory, wherein
The wide angle camera, the wide angle picture for acquiring vehicle front target；
The binocular solid camera, the stereo-picture for acquiring vehicle front target；
Described image analysis processor obtains the target range of detection target for carrying out target detection to the stereo-picture
Information；For carrying out image registration to the wide angle picture and stereo-picture, according to image registration results by the stereo-picture
In detection target re-projection to the wide angle picture in, obtain it is described detection target target identification information；
The memory, for storing the wide angle picture, stereo-picture, target identification information and target range information.
2. more mesh Train-borne recorders according to claim 1, which is characterized in that described image analysis processor includes image
Aligner, the characteristic information of the characteristic information that described image aligner is used to extract the wide angle picture and stereo-picture extraction
Characteristic matching is carried out, determines that the wide angle picture and the coordinate of corresponding pixel points in stereo-picture are converted according to characteristic matching result
Relationship, wherein the coordinate transformation relation is to the stereo-picture and wide angle picture progress image registration.
3. more mesh Train-borne recorders according to claim 2, which is characterized in that described image analysis processor further includes weight
Projector, the re-projection device are used to be converted the detection target projection in the stereo-picture according to the coordinate transformation relation
To the target identification information in the wide angle picture, showing the detection target.
4. more mesh Train-borne recorders according to claim 3, which is characterized in that described image analysis processor further includes phase
Machine calibration device, the camera calibration device is for demarcating the wide angle camera with stereoscopic camera, being assembled, so that the wide-angle
The visual field overlapping region of camera and binocular solid camera is located at the central area of the wide angle picture.
5. a kind of object detection method being registrated with binocular solid camera image based on wide angle camera, which is characterized in that the side
Method includes：
Obtain the wide angle picture of wide angle camera acquisition vehicle front target and standing for binocular solid camera acquisition vehicle front target
Body image；
Target detection is carried out to the stereo-picture, obtains the target range information for detecting target to the binocular solid camera；
Image registration is carried out to the stereo-picture and wide angle picture, it will be in the stereo-picture according to described image registration result
The detection target re-projection to the wide angle picture in, obtain it is described detection target target identification information.
6. object detection method according to claim 5, which is characterized in that carried out to the stereo-picture and wide angle picture
Image registration, including：
The area-of-interest of the wide angle picture is determined according to the visual field overlapping region of the wide angle camera and binocular solid camera；
Extract the characteristic information of area-of-interest described in the wide angle picture and the characteristic information in the stereo-picture；
Characteristic matching is carried out to the characteristic information and the characteristic information of stereo-picture extraction of wide angle picture extraction；
The coordinate transformation relation of the wide angle picture and corresponding pixel points in stereo-picture is determined according to the characteristic matching result,
Wherein, the coordinate transformation relation is to the stereo-picture and wide angle picture progress image registration.
7. object detection method according to claim 6, which is characterized in that will be described vertical according to described image registration result
In the detection target re-projection to the wide angle picture in body image, the target identification information of the detection target is obtained,
Including：
The detection target projection in the stereo-picture is transformed in the wide angle picture according to the coordinate transformation relation；
The target identification information of the detection target is shown by the wide angle picture.
8. object detection method according to claim 5, which is characterized in that obtain wide angle camera and acquire vehicle front target
Wide angle picture and binocular solid camera acquisition vehicle front target stereo-picture, include before：
The wide angle camera and binocular solid camera are demarcated respectively；
Calibrated wide angle camera and binocular solid camera are assembled, so that the wide angle camera and binocular solid camera
Visual field overlapping region is located at the central area of the wide angle picture；
Acquire the image of vehicle front target respectively with binocular solid camera by the wide angle camera after assembly.
9. a kind of intelligent driving system, which is characterized in that including the more mesh Train-borne recorders of claim 1-4 any one of them,
More mesh Train-borne recorders carry out target detection by claim 5-8 any one of them object detection methods.
10. a kind of intelligent driving automobile, which is characterized in that including the intelligent driving system described in claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810557384.5A CN108805910A (en) | 2018-06-01 | 2018-06-01 | More mesh Train-borne recorders, object detection method, intelligent driving system and automobile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810557384.5A CN108805910A (en) | 2018-06-01 | 2018-06-01 | More mesh Train-borne recorders, object detection method, intelligent driving system and automobile |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108805910A true CN108805910A (en) | 2018-11-13 |
Family
ID=64090189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810557384.5A Pending CN108805910A (en) | 2018-06-01 | 2018-06-01 | More mesh Train-borne recorders, object detection method, intelligent driving system and automobile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108805910A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949205A (en) * | 2019-01-30 | 2019-06-28 | 广东工业大学 | A kind of automatic Pilot image perception System and method for for simulating human eye |
| CN110096993A (en) * | 2019-04-28 | 2019-08-06 | 深兰科技（上海）有限公司 | The object detection apparatus and method of binocular stereo vision |
| CN110543813A (en) * | 2019-07-22 | 2019-12-06 | 深思考人工智能机器人科技（北京）有限公司 | Face image and gaze counting method and system based on scene |
| CN111089548A (en) * | 2019-12-16 | 2020-05-01 | 天津大学 | Medium-and-long-distance three-dimensional imaging system with high-resolution texture |
| CN111559314A (en) * | 2020-04-27 | 2020-08-21 | 长沙立中汽车设计开发股份有限公司 | Depth and image information fused 3D enhanced panoramic looking-around system and implementation method |
| CN111914715A (en) * | 2020-07-24 | 2020-11-10 | 廊坊和易生活网络科技股份有限公司 | Intelligent vehicle target real-time detection and positioning method based on bionic vision |
| CN112233185A (en) * | 2020-09-24 | 2021-01-15 | 浙江大华技术股份有限公司 | Camera calibration method, image registration method, camera device and storage device |
| CN112415010A (en) * | 2020-09-30 | 2021-02-26 | 成都中信华瑞科技有限公司 | Imaging detection method and system |
| CN112465921A (en) * | 2020-12-14 | 2021-03-09 | 南京理工大学 | Device and method for measuring front visual field of vehicle driver |
| CN112785649A (en) * | 2019-11-11 | 2021-05-11 | 北京京邦达贸易有限公司 | Laser radar and camera calibration method and device, electronic equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103048995A (en) * | 2011-10-13 | 2013-04-17 | 中国科学院合肥物质科学研究院 | Wide-angle binocular vision identifying and positioning device for service robot |
| CN103455144A (en) * | 2013-08-22 | 2013-12-18 | 深圳先进技术研究院 | Vehicle-mounted man-machine interaction system and method |
| CN103513295A (en) * | 2013-09-25 | 2014-01-15 | 青海中控太阳能发电有限公司 | Weather monitoring system and method based on multi-camera real-time shoot and image processing |
| CN107122770A (en) * | 2017-06-13 | 2017-09-01 | 驭势（上海）汽车科技有限公司 | Many mesh camera systems, intelligent driving system, automobile, method and storage medium |
-
2018
- 2018-06-01 CN CN201810557384.5A patent/CN108805910A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103048995A (en) * | 2011-10-13 | 2013-04-17 | 中国科学院合肥物质科学研究院 | Wide-angle binocular vision identifying and positioning device for service robot |
| CN103455144A (en) * | 2013-08-22 | 2013-12-18 | 深圳先进技术研究院 | Vehicle-mounted man-machine interaction system and method |
| CN103513295A (en) * | 2013-09-25 | 2014-01-15 | 青海中控太阳能发电有限公司 | Weather monitoring system and method based on multi-camera real-time shoot and image processing |
| CN107122770A (en) * | 2017-06-13 | 2017-09-01 | 驭势（上海）汽车科技有限公司 | Many mesh camera systems, intelligent driving system, automobile, method and storage medium |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949205A (en) * | 2019-01-30 | 2019-06-28 | 广东工业大学 | A kind of automatic Pilot image perception System and method for for simulating human eye |
| CN110096993A (en) * | 2019-04-28 | 2019-08-06 | 深兰科技（上海）有限公司 | The object detection apparatus and method of binocular stereo vision |
| CN110543813A (en) * | 2019-07-22 | 2019-12-06 | 深思考人工智能机器人科技（北京）有限公司 | Face image and gaze counting method and system based on scene |
| CN110543813B (en) * | 2019-07-22 | 2022-03-15 | 深思考人工智能机器人科技（北京）有限公司 | Face image and gaze counting method and system based on scene |
| CN112785649A (en) * | 2019-11-11 | 2021-05-11 | 北京京邦达贸易有限公司 | Laser radar and camera calibration method and device, electronic equipment and medium |
| CN111089548A (en) * | 2019-12-16 | 2020-05-01 | 天津大学 | Medium-and-long-distance three-dimensional imaging system with high-resolution texture |
| CN111559314B (en) * | 2020-04-27 | 2021-08-24 | 长沙立中汽车设计开发股份有限公司 | Depth and image information fused 3D enhanced panoramic looking-around system and implementation method |
| CN111559314A (en) * | 2020-04-27 | 2020-08-21 | 长沙立中汽车设计开发股份有限公司 | Depth and image information fused 3D enhanced panoramic looking-around system and implementation method |
| CN111914715A (en) * | 2020-07-24 | 2020-11-10 | 廊坊和易生活网络科技股份有限公司 | Intelligent vehicle target real-time detection and positioning method based on bionic vision |
| CN111914715B (en) * | 2020-07-24 | 2021-07-16 | 廊坊和易生活网络科技股份有限公司 | Intelligent vehicle target real-time detection and positioning method based on bionic vision |
| CN112233185A (en) * | 2020-09-24 | 2021-01-15 | 浙江大华技术股份有限公司 | Camera calibration method, image registration method, camera device and storage device |
| CN112233185B (en) * | 2020-09-24 | 2024-06-11 | 浙江大华技术股份有限公司 | Camera calibration method, image registration method, image pickup device and storage device |
| CN112415010A (en) * | 2020-09-30 | 2021-02-26 | 成都中信华瑞科技有限公司 | Imaging detection method and system |
| CN112415010B (en) * | 2020-09-30 | 2024-06-04 | 成都中信华瑞科技有限公司 | Imaging detection method and system |
| CN112465921A (en) * | 2020-12-14 | 2021-03-09 | 南京理工大学 | Device and method for measuring front visual field of vehicle driver |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108805910A (en) | More mesh Train-borne recorders, object detection method, intelligent driving system and automobile | |
| CN102313536B (en) | Method for barrier perception based on airborne binocular vision | |
| US10909395B2 (en) | Object detection apparatus | |
| JP5455124B2 (en) | Camera posture parameter estimation device | |
| CN113196007B (en) | Camera system applied to vehicle | |
| US20180122086A1 (en) | Image processing apparatus, image processing method, and program | |
| CN107122770B (en) | Multi-camera system, intelligent driving system, automobile, method and storage medium | |
| CN102792333A (en) | Image processing device, method, program and storage medium | |
| CN110009682A (en) | A kind of object recognition and detection method based on monocular vision | |
| CN113888639B (en) | Visual odometer positioning method and system based on event camera and depth camera | |
| US9373175B2 (en) | Apparatus for estimating of vehicle movement using stereo matching | |
| CN110288659A (en) | A kind of Depth Imaging and information acquisition method based on binocular vision | |
| CN106991378A (en) | Facial orientation detection method, detection means and electronic installation based on depth | |
| CN110675436A (en) | Laser radar and stereoscopic vision registration method based on 3D feature points | |
| CN108362205A (en) | Space ranging method based on fringe projection | |
| CN113643345A (en) | Multi-view road intelligent identification method based on double-light fusion | |
| CN104949657A (en) | Object detection device, object detection method, and computer readable storage medium comprising objection detection program | |
| KR20170014916A (en) | Method for calculating an object's coordinates in an image using single camera and gps | |
| CN110750153A (en) | Dynamic virtualization device of unmanned vehicle | |
| Kinzig et al. | Real-time seamless image stitching in autonomous driving | |
| CN112330747A (en) | Multi-sensor combined detection and display method based on unmanned aerial vehicle platform | |
| KR102003387B1 (en) | Method for detecting and locating traffic participants using bird's-eye view image, computer-readerble recording medium storing traffic participants detecting and locating program | |
| CN111986248B (en) | Multi-vision sensing method and device and automatic driving automobile | |
| CN112639864B (en) | Method and apparatus for ranging | |
| RU2552123C2 (en) | Method of selecting objects on remote background |
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 |
Application publication date: 20181113 |
|
| RJ01 | Rejection of invention patent application after publication |