CN108460323A - A kind of backsight blind area vehicle checking method of fusion vehicle mounted guidance information - Google Patents
A kind of backsight blind area vehicle checking method of fusion vehicle mounted guidance information Download PDFInfo
- Publication number
- CN108460323A CN108460323A CN201711478171.5A CN201711478171A CN108460323A CN 108460323 A CN108460323 A CN 108460323A CN 201711478171 A CN201711478171 A CN 201711478171A CN 108460323 A CN108460323 A CN 108460323A
- Authority
- CN
- China
- Prior art keywords
- detection
- target
- backsight
- model
- information
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The backsight blind area vehicle checking method of the fusion vehicle mounted guidance information of the application provides the solution merged with vehicle detecting algorithm for the vehicle mounted guidance of backsight blind area, the scene information and weather conditions that navigation provides, vehicle detecting algorithm selects different model combinations and parameter according to different environment self-adaptions, so that vehicle detecting algorithm preferably adapts to situation complicated and changeable, have higher accuracy of detection and efficiency, can preferably be applied to Automobile Electronic Industry.
Description
Technical field
This application involves a kind of blind area vehicle checking methods, specifically, belong to a kind of backsight of fusion vehicle mounted guidance information
Blind area vehicle checking method.
Background technology
With increasing rapidly for car ownership, automobile driving safe increasingly thirsts for technology by universal concern, people
The safety brought with it is convenient.Therefore, automobile ADAS systems are furtherd investigate, and are widely used in Automobile Electronic Industry, are developed into
For the core technology of automotive electronics.Since machine vision can clearly capture the information around vehicle body, to the color of object
There is preferable analytic ability with information such as textures, can effectively identify that vehicle, pedestrian, traffic police around vehicle body etc. etc. has
Huge advantage, therefore, intelligent vision module is applied on automobile, is the solution party of the great competitiveness of current driving assistance system
Case has huge market prospects.However, in complex scene and Changes in weather, the intractability of vision algorithm, shadow can be increased
Ring overall performance index.
Invention content
The present invention is at least one defect overcome described in the above-mentioned prior art, provides a kind of fusion vehicle mounted guidance information
Backsight blind area vehicle checking method.
The present invention is directed to solve above-mentioned technical problem at least to a certain extent.
The primary and foremost purpose of the present invention is to improve accuracy of detection and efficiency.
In order to solve the above technical problems, technical scheme is as follows:
A kind of backsight blind area vehicle checking method of fusion vehicle mounted guidance information, includes the following steps:
S1, detection start;
S2, it receives navigation information and passes through pretreated backsight image and carry out model initialization;
S3, carry out model selection and combine;
S4, it carries out the detection of primary layer sub-pixel using the model I combined and judges whether there is target, terminate this if no target
Secondary detection, then carries out step S5 if any target, the model I be using Pixel-level feature train come model;
S5, it carries out intermediate level edge detection using the model II combined and judges whether there is target, terminate if no target
The secondary detection, then carries out step S6 if any target, the model II be using edge feature train come model;
S6, it carries out advanced hierarchically structured detection using the model III combined and judges whether there is target, tied if no target
The beam secondary detection, then carries out step S7 if any target, the model III be using the edge feature of combination train come mould
Type;
S7, by after detection information and navigation information carry out data fusion;
S8, detection terminate.
Further, the pretreatment of backsight image includes the following steps in the step S2:
S21, the input of original backsight image data, original backsight image data packet is containing left side and right side data information;
S22, it is provided to generate the range of interest in image according to calibrating parameters and practical blind area demand, and calculates each picture number
The range-azimuth at strong point and camera;
S23, adaptive aberration correction algorithm is executed:Composite calibration parameter, image data point, the range-azimuth with camera, it is complete
At the mapping process of each point in range of interest;
S24, view transformation algorithm is executed:Pass through image transformation matrix so that the range of interest data observed are at
Best detecting state;
S25, final image to be detected is obtained, is sent into detection module and is detected.
Further, in the step S4, the purpose of primary layer sub-pixel detection is the candidate regions where filtering out vehicle
Domain, the feature of selection are exactly the luminance information inside the channels Y, specifically include following steps:
S41, the input by the backsight subvolume of interest image block in pretreated backsight image as model;
S42, pyramid level is established:The detection of target under different sizes is completed, from top to bottom, from left to right traversal can to search for
The target location of energy;
S43, for the location of pixels (x, y, width, height) in i-th layer of pyramid, make decisions ratio with model data
Compared with, and obtain corresponding score;
S44, in conjunction with the score on each position on pyramids at different levels, be normalized to 0-255, obtain probability distribution
Figure;
S45, for probability distribution graph be filtered and connected region divide, it is therefore an objective to smoothed image, remove noise, obtain it is each
Subregion;
S46, it is suitably extended out for candidate subregion, and according to score ranked candidate subregion, decision goes out candidate region
Detect priority.
Further, in the step S5, the data of intermediate level edge detection process are that the detection of primary layer sub-pixel is defeated
The candidate region frame gone out, selection are characterized as edge gradient information, carry out decision and judge to obtain the sub-block where each target.
Further, in the step S6, advanced hierarchically structured detection is the base exported in intermediate level edge detection
On plinth, each sub-block is confirmed, on the basis of ensureing target effective detection, removes false-alarm, selection is characterized as structuring
Feature, the feature of intermediate level edge detection is specifically weighted combination.
Further, in the step S3, model selection is that optimal mould is selected according to the environmental information in navigation information
Type and the parameter to match.
Further, navigation information includes road information, scene information, weather information.
Further, navigation information includes the information data for being related to highway, urban road, frontlighting, backlight, tunnel.
Further, the application provides a kind of backsight blind area vehicle using the fusion vehicle mounted guidance information as described in aforementioned
The detecting system of detection method, including navigation system, Model selection module, model composite module, detection module, Fusion Module.
Compared with prior art, the advantageous effect of technical solution of the present invention is:The present invention proposes vehicle mounted guidance and rear ablepsia
The fusion solution of area's vehicle detection selects different models and parameter for varying environment factor, and carries out in each level
Model combination is carried out, the detection of target is more advantageous to, improves recall rate, reduce false-alarm.The present invention can be carried out in test side
Successively other target detection process, primary-level offer Pixel-level another characteristic detection, the candidate region of acquisition is for middle-level
The detection of edge level another characteristic, the candidate region obtained on this basis is applied to high-level structured features detection, defeated
The information gone out is merged to obtain final as a result, this method can more effectively detect target.
Description of the drawings
Fig. 1 is detection system structure.
Fig. 2 is the backsight blind area vehicle checking method flow diagram for merging vehicle mounted guidance information.
Fig. 3 is backsight image pretreatment process schematic diagram.
Fig. 4 is primary layer sub-pixel detection process schematic diagram.
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent;It is attached in order to more preferably illustrate the present embodiment
Scheme certain components to have omission, zoom in or out, does not represent the size of actual product;To those skilled in the art,
The omitting of some known structures and their instructions in the attached drawings are understandable;Same or analogous label corresponds to same or similar
Component;The terms describing the positional relationship in the drawings are only for illustration, should not be understood as the limitation to this patent.
Specific implementation mode
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
Embodiment 1
Referring to attached drawing, the backsight blind area vehicle checking method of the fusion vehicle mounted guidance information of the application includes the following steps:
S1, detection start;
S2, it receives navigation information and passes through pretreated backsight image and carry out model initialization;
S3, carry out model selection and combine;
S4, it carries out the detection of primary layer sub-pixel using the model combined and judges whether there is target, terminate this if no target
Secondary detection, then carries out step S5 if any target, the model I be using Pixel-level feature train come model;
S5, it carries out intermediate level edge detection using the model combined and judges whether there is target, terminate this if no target
Secondary detection, then carries out step S6 if any target, the model II be using edge feature train come model;
S6, it carries out advanced hierarchically structured detection using the model combined and judges whether there is target, terminate if no target
The secondary detection, then carries out step S7 if any target, the model III be using the edge feature of combination train come model;
S7, by after detection information and navigation information carry out data fusion;
S8, detection terminate.
Embodiment 2
The present embodiment is similar to embodiment 1, since the lens type used when acquisition backsight image is fisheye camera lenses, advantage
Be observation angular field of view it is very big, the data of acquisition are more;But disadvantage is it is obvious that there are larger distortion, especially distant place
In backsight blind area, vehicle is big in image distortion, and torsional deformation is serious, influences the detection and identification of target, needs to be corrected and become
Change processing, therefore further, in the step S2 pretreatment of backsight image include the following steps:
S21, the input of original backsight image data, original backsight image data packet is containing left side and right side data information;
S22, it is provided to generate the range of interest in image according to calibrating parameters and practical blind area demand, and calculates each picture number
The range-azimuth at strong point and camera;
S23, adaptive aberration correction algorithm is executed:Composite calibration parameter, image data point, the range-azimuth with camera, it is complete
At the mapping process of each point in range of interest;
S24, view transformation algorithm is executed:Pass through image transformation matrix so that the range of interest data observed are at
Best detecting state;
S25, final image to be detected is obtained, is sent into detection module and is detected.
Embodiment 3
The present embodiment is similar to embodiment 1-2, and further, in the step S4, the purpose of primary layer sub-pixel detection is sieve
The candidate region where vehicle is selected, the feature of selection is exactly the luminance information inside the channels Y, specifically includes following steps:
S41, the input by the backsight subvolume of interest image block in pretreated backsight image as model;
S42, pyramid level is established:The detection of target under different sizes is completed, from top to bottom, from left to right traversal can to search for
The target location of energy;
S43, for the location of pixels (x, y, width, height) in i-th layer of pyramid, make decisions ratio with model data
Compared with, and obtain corresponding score;
S44, in conjunction with the score on each position on pyramids at different levels, be normalized to 0-255, obtain probability distribution
Figure;
S45, for probability distribution graph be filtered and connected region divide, it is therefore an objective to smoothed image, remove noise, obtain it is each
Subregion;
S46, it is suitably extended out for candidate subregion, and according to score ranked candidate subregion, decision goes out candidate region
Detect priority.
In step S5, the data of intermediate level edge detection process are the candidate regions of primary layer sub-pixel detection output
Frame, selection are characterized as edge gradient information, carry out decision and judge to obtain the sub-block where each target.
In step S6, advanced hierarchically structured detection is on the basis of intermediate level edge detection exports, to each height
Block is confirmed, on the basis of ensureing target effective detection, removes false-alarm, selection is characterized as the feature of structuring, specifically
The feature of intermediate level edge detection is weighted combination.
Primary layer sub-pixel detection design purpose is acquisition candidate region, and the feature of selection is exactly simple pixel value,
Fast and easy filters out nontarget area, and retains effective target;The purpose of intermediate level edge detection is special using edge gradient
Sign, searches out the position where target;The purpose of advanced hierarchically structured detection design is removal false target, and what is utilized is middle rank
The weighted array of the feature of level edge detection extraction.
Embodiment 4
The present embodiment is similar to embodiment 1-3, and further, in the step S3, model selection is according in navigation information
The parameter that environmental information selects optimal model and matches.
Navigation information includes road information, scene information, weather information.
Navigation information includes the information data for being related to highway, urban road, frontlighting, backlight, tunnel.
Embodiment 5
The application provides a kind of inspection of the backsight blind area vehicle checking method using the fusion vehicle mounted guidance information as described in aforementioned
Examining system, including navigation system, Model selection module, model composite module, detection module, Fusion Module.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
All any modification, equivalent and improvement etc., should be included in the claims in the present invention made by within the spirit and principle of invention
Protection domain within.
Claims (8)
1. a kind of backsight blind area vehicle checking method of fusion vehicle mounted guidance information, it is characterised in that:Include the following steps:
S1, detection start;
S2, it receives navigation information and passes through pretreated backsight image and carry out model initialization;
S3, carry out model selection and combine;
S4, it carries out the detection of primary layer sub-pixel using [CSA (1] I and judges whether there is target, terminate the secondary inspection if no target
Survey, then carry out step S5 if any target, the model I be using Pixel-level feature train come model;
S5, it carries out intermediate level edge detection using the model II combined and judges whether there is target, terminate if no target
The secondary detection, then carries out step S6 if any target, the model II be using edge feature train come model;
S6, it carries out advanced hierarchically structured detection using the model III combined and judges whether there is target, tied if no target
The beam secondary detection, then carries out step S7 if any target, the model III be using the edge feature of combination train come mould
Type;
S7, by after detection information and navigation information carry out data fusion;
S8, detection terminate.
2. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 1, it is characterised in that:Institute
It states in step S2, the pretreatment of backsight image includes the following steps:
S21, the input of original backsight image data, original backsight image data packet is containing left side and right side backsight image data information;
S22, it is provided to generate the range of interest in image according to calibrating parameters and practical blind area demand, and calculates each picture number
The range-azimuth at strong point and camera;
S23, adaptive aberration correction algorithm is executed:Composite calibration parameter, image data point, the range-azimuth with camera, it is complete
At the mapping process of each point in range of interest;
S24, view transformation algorithm is executed:Pass through image transformation matrix so that the range of interest data observed are at
Best detecting state;
S25, final image to be detected is obtained, is sent into detection module and is detected.
3. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 1 or 2, feature exist
In:In the step S4, the purpose of primary layer sub-pixel detection is the candidate region where filtering out vehicle, and the feature of selection is just
It is the luminance information inside the channels Y, specifically includes following steps:
S41, the input by the backsight subvolume of interest image block in pretreated backsight image as model;
S42, pyramid level is established:The detection of target under different sizes is completed, from top to bottom, from left to right traversal can to search for
The target location of energy;
S43, for the location of pixels (x, y, width, height) in i-th layer of pyramid, make decisions ratio with model data
Compared with, and obtain corresponding score;
S4 4, in conjunction with the score on each position on pyramids at different levels, score is normalized to 0-255, is obtained
Probability distribution graph;
S45, for probability distribution graph be filtered and connected region divide, it is therefore an objective to smoothed image, remove noise, obtain it is each
Subregion;
S46, it is suitably extended out for candidate subregion, and according to score ranked candidate subregion, decision goes out candidate region
Detect priority.
4. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 3, it is characterised in that:Institute
It states in step S5, the data of intermediate level edge detection process are the candidate region frame of primary layer sub-pixel detection output, selection
It is characterized as that edge gradient information, intermediate level edge detection carry out decision and judges to obtain the sub-block where each target.
5. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 4, it is characterised in that:Institute
It states in step S6, advanced hierarchically structured detection is carried out to each sub-block on the basis of intermediate level edge detection exports
Confirm, on the basis of ensureing target effective detection, removes false-alarm, selection is characterized as the feature of structuring, specifically will be intermediate
The feature of level edge detection is weighted combination.
6. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 1, it is characterised in that:Institute
It states in step S3, model selection is the parameter for selecting optimal model according to the environmental information in navigation information and matching.
7. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 6, it is characterised in that:It leads
Information of navigating includes road information, scene information, weather information.
8. the backsight blind area vehicle checking method of fusion vehicle mounted guidance information according to claim 7, it is characterised in that:It leads
Information of navigating includes being related to the information data in highway, urban road, frontlighting, backlight, tunnel.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711478171.5A CN108460323B (en) | 2017-12-29 | 2017-12-29 | Rearview blind area vehicle detection method fusing vehicle-mounted navigation information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711478171.5A CN108460323B (en) | 2017-12-29 | 2017-12-29 | Rearview blind area vehicle detection method fusing vehicle-mounted navigation information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108460323A true CN108460323A (en) | 2018-08-28 |
CN108460323B CN108460323B (en) | 2022-05-20 |
Family
ID=63221219
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711478171.5A Active CN108460323B (en) | 2017-12-29 | 2017-12-29 | Rearview blind area vehicle detection method fusing vehicle-mounted navigation information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108460323B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100312386A1 (en) * | 2009-06-04 | 2010-12-09 | Microsoft Corporation | Topological-based localization and navigation |
US20110081087A1 (en) * | 2009-10-02 | 2011-04-07 | Moore Darnell J | Fast Hysteresis Thresholding in Canny Edge Detection |
CN105512115A (en) * | 2014-09-22 | 2016-04-20 | 惠州市德赛西威汽车电子股份有限公司 | Vehicle navigation picture processing method |
CN106529530A (en) * | 2016-10-28 | 2017-03-22 | 上海大学 | Monocular vision-based ahead vehicle detection method |
-
2017
- 2017-12-29 CN CN201711478171.5A patent/CN108460323B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100312386A1 (en) * | 2009-06-04 | 2010-12-09 | Microsoft Corporation | Topological-based localization and navigation |
US20110081087A1 (en) * | 2009-10-02 | 2011-04-07 | Moore Darnell J | Fast Hysteresis Thresholding in Canny Edge Detection |
CN105512115A (en) * | 2014-09-22 | 2016-04-20 | 惠州市德赛西威汽车电子股份有限公司 | Vehicle navigation picture processing method |
CN106529530A (en) * | 2016-10-28 | 2017-03-22 | 上海大学 | Monocular vision-based ahead vehicle detection method |
Also Published As
Publication number | Publication date |
---|---|
CN108460323B (en) | 2022-05-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107392103B (en) | Method and device for detecting road lane line and electronic equipment | |
CN109460709B (en) | RTG visual barrier detection method based on RGB and D information fusion | |
CN111448478B (en) | System and method for correcting high-definition maps based on obstacle detection | |
CN111738314B (en) | Deep learning method of multi-modal image visibility detection model based on shallow fusion | |
CN107590470B (en) | Lane line detection method and device | |
CN109034047A (en) | A kind of method for detecting lane lines and device | |
CN110910378B (en) | Bimodal image visibility detection method based on depth fusion network | |
JP5631581B2 (en) | Road recognition device | |
JP2013225289A (en) | Multi-lens camera apparatus and vehicle including the same | |
CN111209780A (en) | Lane line attribute detection method and device, electronic device and readable storage medium | |
JP5180126B2 (en) | Road recognition device | |
JP2008158958A (en) | Road surface determination method and road surface determination device | |
CN106951898B (en) | Vehicle candidate area recommendation method and system and electronic equipment | |
CN112215306A (en) | Target detection method based on fusion of monocular vision and millimeter wave radar | |
CN104902261A (en) | Device and method for road surface identification in low-definition video streaming | |
JP5188429B2 (en) | Environment recognition device | |
CN111967396A (en) | Processing method, device and equipment for obstacle detection and storage medium | |
CN111144301A (en) | Road pavement defect quick early warning device based on degree of depth learning | |
CN107220632B (en) | Road surface image segmentation method based on normal characteristic | |
CN112464914A (en) | Guardrail segmentation method based on convolutional neural network | |
JP2014106739A (en) | In-vehicle image processing device | |
KR20120098292A (en) | Method for detecting traffic lane | |
JP5189556B2 (en) | Lane detection device | |
JP5091897B2 (en) | Stop line detector | |
CN112669615A (en) | Parking space detection method and system based on camera |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |