CN102840853A - Obstacle detection and alarm method for vehicle-mounted night vision system - Google Patents

Obstacle detection and alarm method for vehicle-mounted night vision system Download PDF

Info

Publication number
CN102840853A
CN102840853A CN2012102601507A CN201210260150A CN102840853A CN 102840853 A CN102840853 A CN 102840853A CN 2012102601507 A CN2012102601507 A CN 2012102601507A CN 201210260150 A CN201210260150 A CN 201210260150A CN 102840853 A CN102840853 A CN 102840853A
Authority
CN
China
Prior art keywords
interest
barrier
night vision
vision system
area
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN2012102601507A
Other languages
Chinese (zh)
Inventor
赵东旭
贾俊涛
张良
毛鑫
张同卿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Luoyang Institute of Electro Optical Equipment AVIC
Original Assignee
Luoyang Institute of Electro Optical Equipment AVIC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Luoyang Institute of Electro Optical Equipment AVIC filed Critical Luoyang Institute of Electro Optical Equipment AVIC
Priority to CN2012102601507A priority Critical patent/CN102840853A/en
Publication of CN102840853A publication Critical patent/CN102840853A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)

Abstract

Belonging to the field of computer vision technologies, the invention relates to an obstacle detection and alarm method for a vehicle-mounted night vision system. The method consists of: first reading infrared image data output by a thermal imager, and conducting pretreatment to filter out noise, then using a self-adaptive threshold segmentation method to carry out image segmentation, marking the segmentation result so as to obtain a region of interest, extracting a histogram of oriented gradient from the region of interest to compose a characteristic vector, which is input into a multi-stage classifier to perform classification and identification, and judging whether the region of interest is an obstacle. The method provided in the invention can automatically detect the obstacle in the front, has a high detection identification rate, and can be implemented by hardware in real time.

Description

A kind of detection of obstacles and alarm method that is used for the on-vehicle night vision system
Technical field
The present invention relates to a kind of detection of obstacles and alarm method that is used for the on-vehicle night vision system, belong to technical field of computer vision.
Background technology
Safe DAS (Driver Assistant System) utilizes the barrier in sensor technology detection vehicle the place ahead, and in time alerting driver is noted, reduces loss and casualties that traffic hazard brings.Along with China's automobile pollution increases sharply, develop safe DAS (Driver Assistant System) and have urgency and realistic meaning.Utilize digital image processing techniques; Reception is from the infrared image of thermal imaging system; Automatically the barriers such as motor vehicle, bicycle, pedestrian and larger animal that occur in the detected image; Provide warning information and remind the driver in time to dodge, in future transportation safety, play an important role, be with a wide range of applications.
Obstacle detection method based on vision can be divided into: method for testing motion, stereoscopic vision detecting method and based on the detection method of knowledge.Method for testing motion mainly contains optical flow method, utilizes optical flow analysis estimation kinematic parameter in the sequence frame, and calculated amount is big, and real-time is poor.The stereoscopic vision method mainly adopts binocular vision, utilizes parallax to detect barrier, need carry out the solid coupling, and Camera calibration is easy to generate drift.Based on the detection method of knowledge, utilize the priori of barrier in image, to seek the target position, this method is simple relatively, and hardware is realized easily, can satisfy the requirement of moving vehicle real-time.Obstacle detection method based on knowledge mainly contains based on the detection method of template matches with based on the detection method of cutting apart.Wherein based on the detection method of template matches, realize detecting through selecting suitable barrier template, this method need be done multiple dimensioned search in image, and calculated amount is big, can't real-time implementation.
Summary of the invention
The detection of obstacles and the alarm method that the purpose of this invention is to provide a kind of on-vehicle night vision system are need do big, the problem that can't real-time implementation of the calculated amount that multiple dimensioned search caused in image in the testing process that solves barrier in the present on-vehicle night vision system.
The present invention solves the problems of the technologies described above the obstacle detection method that a kind of on-vehicle night vision system is provided, and the step of this detection method is following:
1). read the infrared picture data of thermal imaging system output, it is carried out the pre-service filtering noise;
2). utilize the adaptive threshold dividing method to carry out image segmentation to pretreated infrared image, obtain area-of-interest;
3). extract the proper vector that area-of-interest gradient orientation histogram characteristic is formed;
4). with carrying out the differentiation result that Classification and Identification obtains area-of-interest in the proper vector input multistage classifier that extracts, determine whether it is barrier.
Said step 2) concrete steps are following:
A. the maximal value of statistical picture gray-scale value and average are calculated its self-adaptation segmentation threshold;
B. utilize segmentation threshold that view data is cut apart;
C. cut zone is carried out zone marker;
D. calculate the eigenwert of marked region.
The concrete steps of described step 3) are following:
A. adopt gradient operator that image is handled;
B. the histogram on the compute gradient direction;
C. with gradient orientation histogram composition characteristic vector.
The unresolved above-mentioned technical matters of the present invention also provides the barrier alarm method of a kind of on-vehicle night vision system, and the step of this alarm method is following:
1). read the infrared picture data of thermal imaging system output, it is carried out the pre-service filtering noise;
2). utilize the adaptive threshold dividing method to carry out image segmentation to pretreated infrared image, obtain area-of-interest;
3). extract the proper vector that area-of-interest gradient orientation histogram characteristic is formed;
4). with carrying out Classification and Identification in the proper vector input multistage classifier that extracts, judge in the area-of-interest whether be barrier;
5). to being judged as the extracted region size and the positional information of barrier, and carrying out danger classes and demarcate, provide corresponding prompt messages according to the danger classes of demarcating.
Described step 2) concrete steps are following:
A. the maximal value of statistical picture gray-scale value and average are calculated its self-adaptation segmentation threshold;
B. utilize segmentation threshold that view data is cut apart;
C. cut zone is carried out zone marker;
D. calculate the eigenwert of marked region.
The concrete steps of said step 3) are following:
A. adopt gradient operator that image is handled;
B. the histogram on the compute gradient direction;
C. with gradient orientation histogram composition characteristic vector.
Described step 5) comprises the steps:
A). extract the size and the positional information of barrier region;
B). estimate the distance between barrier and the video camera;
C). demarcate danger classes according to estimated distance.
The invention has the beneficial effects as follows: the present invention through reading thermal imaging system output earlier infrared picture data and it is carried out the pre-service filtering noise; Utilize the adaptive threshold dividing method to carry out image segmentation then; Segmentation result is carried out mark obtain area-of-interest; Area-of-interest is extracted gradient orientation histogram characteristic composition characteristic vector, proper vector is imported carried out Classification and Identification in the multistage classifier, judge whether area-of-interest is barrier.The present invention can detect the place ahead barrier automatically, and it is high to detect discrimination, but and hardware real-time realization.
Description of drawings
Fig. 1 is the process flow diagram that the present invention is used for the barrier alarm method of on-vehicle night vision system;
Fig. 2 is the obstacle detection method embodiment infrared image area-of-interest signature that the present invention is used for the on-vehicle night vision system;
Fig. 3 is the obstacle detection method embodiment infrared image detection of obstacles figure as a result that the present invention is used for the on-vehicle night vision system.
Embodiment
Be further described below in conjunction with the accompanying drawing specific embodiments of the invention
A kind of embodiment that is used for the obstacle detection method of on-vehicle night vision system of the present invention
The obstacle detection method that is used for the on-vehicle night vision system of the present invention; Its main process comprises: the infrared picture data that at first reads thermal imaging system output; Carry out the pre-service filtering noise; Utilize the adaptive threshold dividing method to carry out image segmentation then, segmentation result is carried out mark obtain area-of-interest, area-of-interest is extracted gradient orientation histogram characteristic composition characteristic vector; Proper vector imported carry out Classification and Identification in the multistage classifier, judge whether area-of-interest is barrier.Its concrete implementation step is following:
1. read the infrared picture data of thermal imaging system output;
2. image is carried out pre-service, filtering noise;
3. to the result images data of step 2, statistics maximal value and average are calculated the self-adaptation segmentation threshold, utilize the self-adaptation segmentation threshold that image is cut apart, and the computing formula of self-adaptation segmentation threshold is: Th=α f Max+ (1-α) f Mean
4. the segmentation result image of step 3 being handled that obtains carries out zone marker, obtains area-of-interest, and is as shown in Figure 2;
5. utilize gradient operator that the area-of-interest that obtains in the step 4 is handled, gradient operator is a Δ H=[1 0 1], Δ V = - 1 0 1 ;
6. utilize step 5 process result to calculate the gradient orientation histogram of area-of-interest, the composition characteristic vector, the computing formula of gradient orientation histogram is: D ( x , y ) = Arctan ( G H ( x , y ) G V ( x , y ) ) , H ( d ) = Σ i = 1 n Σ j = 1 m δ [ D ( i , j ) - d ] * W ( i , j ) ;
7. the proper vector of the area-of-interest that obtains in the step 6 is input to and manyly carries out Classification and Identification in having built somewhat, whether be barrier thereby judge this area-of-interest, as shown in Figure 3.
A kind of embodiment that is used for the barrier alarm method of on-vehicle night vision system of the present invention
The barrier alarm method that is used for the on-vehicle night vision system of the present invention, its main process is as shown in Figure 1, at first reads the infrared picture data of thermal imaging system output; Carry out the pre-service filtering noise; Utilize the adaptive threshold dividing method to carry out image segmentation then, segmentation result is carried out mark obtain area-of-interest, area-of-interest is extracted gradient orientation histogram characteristic composition characteristic vector; Proper vector imported carry out Classification and Identification in the multistage classifier; Judge whether area-of-interest is barrier, at last extracted region size and the positional information of confirming as barrier are carried out the danger classes demarcation, provide alarm prompt according to the danger classes of demarcating.Its concrete implementation step is following:
1. read the infrared picture data of thermal imaging system output;
2. image is carried out pre-service, filtering noise;
3. to the result images data of step 2, statistics maximal value and average are calculated the self-adaptation segmentation threshold, utilize the self-adaptation segmentation threshold that image is cut apart, and the computing formula of self-adaptation segmentation threshold is: Th=α f Max+ (1-α) f Mean
4. the segmentation result image of step 3 being handled that obtains carries out zone marker, obtains area-of-interest, and is as shown in Figure 2;
5. utilize gradient operator that the area-of-interest that obtains in the step 4 is handled, gradient operator is a Δ H=[1 0 1], Δ V = - 1 0 1 ;
6. utilize step 5 process result to calculate the gradient orientation histogram of area-of-interest, the composition characteristic vector, the computing formula of gradient orientation histogram is: D ( x , y ) = Arctan ( G H ( x , y ) G V ( x , y ) ) , H ( d ) = Σ i = 1 n Σ j = 1 m δ [ D ( i , j ) - d ] * W ( i , j ) ;
7. the proper vector of the area-of-interest that obtains in the step 6 is input to and carries out Classification and Identification in the multistage classifier, whether be barrier thereby judge this area-of-interest, as shown in Figure 3;
8. its size of extracted region and the positional information to confirming as barrier, and image coordinate system transformed to world coordinate system, its coordinate conversion formula is:
Y P = h ( Htg γ 0 + 2 y p tg α 0 ) H - 2 tg γ 0 tg α 0 y p X P = h 2 + Y P 2 2 x p tg β 0 W
Estimate the distance between barrier and the video camera, demarcate the danger classes of this barrier based on estimated distance;
9. the danger classes of demarcating according to barrier region in the step 8 provides alarm prompt, thereby the barrier of realizing the on-vehicle night vision system is reported to the police.

Claims (7)

1. obstacle detection method that is used for the on-vehicle night vision system, it is characterized in that: the step of this detection method is following:
1). read the infrared picture data of thermal imaging system output, it is carried out the pre-service filtering noise;
2). utilize the adaptive threshold dividing method to carry out image segmentation to pretreated infrared image, obtain area-of-interest;
3). extract the proper vector that area-of-interest gradient orientation histogram characteristic is formed;
4). with carrying out the differentiation result that Classification and Identification obtains area-of-interest in the proper vector input multistage classifier that extracts, determine whether it is barrier.
2. the obstacle detection method that is used for the on-vehicle night vision system according to claim 1 is characterized in that: concrete steps said step 2) are following:
A. the maximal value of statistical picture gray-scale value and average are calculated its self-adaptation segmentation threshold;
B. utilize segmentation threshold that view data is cut apart;
C. cut zone is carried out zone marker;
D. calculate the eigenwert of marked region.
3. the obstacle detection method that is used for the on-vehicle night vision system according to claim 1 is characterized in that: the concrete steps of described step 3) are following:
A. adopt gradient operator that image is handled;
B. the histogram on the compute gradient direction;
C. with gradient orientation histogram composition characteristic vector.
4. barrier alarm method that is used for the on-vehicle night vision system, it is characterized in that: the step of this alarm method is following:
1). read the infrared picture data of thermal imaging system output, it is carried out the pre-service filtering noise;
2). utilize the adaptive threshold dividing method to carry out image segmentation to pretreated infrared image, obtain area-of-interest;
3). extract the proper vector that area-of-interest gradient orientation histogram characteristic is formed;
4). with carrying out Classification and Identification in the proper vector input multistage classifier that extracts, judge in the area-of-interest whether be barrier;
5). to being judged as the extracted region size and the positional information of barrier, and carrying out danger classes and demarcate, provide corresponding prompt messages according to the danger classes of demarcating.
5. the barrier alarm method that is used for the on-vehicle night vision system according to claim 4, it is characterized in that: concrete steps described step 2) are following:
A. the maximal value of statistical picture gray-scale value and average are calculated its self-adaptation segmentation threshold;
B. utilize segmentation threshold that view data is cut apart;
C. cut zone is carried out zone marker;
D. calculate the eigenwert of marked region.
6. the barrier alarm method that is used for the on-vehicle night vision system according to claim 4, it is characterized in that: the concrete steps of said step 3) are following:
A. adopt gradient operator that image is handled;
B. the histogram on the compute gradient direction;
C. with gradient orientation histogram composition characteristic vector.
7. the barrier alarm method that is used for the on-vehicle night vision system according to claim 4, it is characterized in that: described step 5) comprises the steps:
A). extract the size and the positional information of barrier region;
B). estimate the distance between barrier and the video camera;
C). demarcate danger classes according to estimated distance.
CN2012102601507A 2012-07-25 2012-07-25 Obstacle detection and alarm method for vehicle-mounted night vision system Pending CN102840853A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012102601507A CN102840853A (en) 2012-07-25 2012-07-25 Obstacle detection and alarm method for vehicle-mounted night vision system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2012102601507A CN102840853A (en) 2012-07-25 2012-07-25 Obstacle detection and alarm method for vehicle-mounted night vision system

Publications (1)

Publication Number Publication Date
CN102840853A true CN102840853A (en) 2012-12-26

Family

ID=47368433

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012102601507A Pending CN102840853A (en) 2012-07-25 2012-07-25 Obstacle detection and alarm method for vehicle-mounted night vision system

Country Status (1)

Country Link
CN (1) CN102840853A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103093214A (en) * 2013-02-01 2013-05-08 浙江捷尚视觉科技有限公司 Pedestrian detection method based on on-board infrared camera
CN104751138A (en) * 2015-03-27 2015-07-01 东华大学 Vehicle mounted infrared image colorizing assistant driving system
CN111251994A (en) * 2018-11-30 2020-06-09 华创车电技术中心股份有限公司 Method and system for detecting objects around vehicle
CN111552289A (en) * 2020-04-28 2020-08-18 苏州高之仙自动化科技有限公司 Detection method, virtual radar device, electronic apparatus, and storage medium
CN117148363A (en) * 2023-09-06 2023-12-01 广州优创电子有限公司 Method and system for detecting obstacle height by ultrasonic wave

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101660908A (en) * 2009-09-11 2010-03-03 天津理工大学 Visual locating and navigating method based on single signpost
CN102447911A (en) * 2010-10-01 2012-05-09 魏载荣 Image acquisition unit, acquisition method, and associated control unit
CN102467821A (en) * 2010-11-04 2012-05-23 北京汉王智通科技有限公司 Road distance detection method based on video image and apparatus thereof

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101660908A (en) * 2009-09-11 2010-03-03 天津理工大学 Visual locating and navigating method based on single signpost
CN102447911A (en) * 2010-10-01 2012-05-09 魏载荣 Image acquisition unit, acquisition method, and associated control unit
CN102467821A (en) * 2010-11-04 2012-05-23 北京汉王智通科技有限公司 Road distance detection method based on video image and apparatus thereof

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ENZWEILER M ETC.: "monocular pedestrian detection: survey and experiments", 《IEEE》, vol. 31, no. 12, 31 December 2009 (2009-12-31), XP011266787, DOI: doi:10.1109/TPAMI.2008.260 *
溪海燕: "基于线性SVM的车辆前方行人检测方法", 《天津工业大学学报》, vol. 31, no. 1, 29 February 2012 (2012-02-29) *
贾俊涛: "一种串联组合分类的红外行人检测方法", 《应用光学》, vol. 33, no. 4, 15 July 2012 (2012-07-15) *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103093214A (en) * 2013-02-01 2013-05-08 浙江捷尚视觉科技有限公司 Pedestrian detection method based on on-board infrared camera
CN103093214B (en) * 2013-02-01 2015-08-19 浙江捷尚视觉科技股份有限公司 A kind of pedestrian detection method based on vehicle mounted infrared camera
CN104751138A (en) * 2015-03-27 2015-07-01 东华大学 Vehicle mounted infrared image colorizing assistant driving system
CN104751138B (en) * 2015-03-27 2018-02-23 东华大学 A kind of vehicle mounted infrared image colorization DAS (Driver Assistant System)
CN111251994A (en) * 2018-11-30 2020-06-09 华创车电技术中心股份有限公司 Method and system for detecting objects around vehicle
CN111251994B (en) * 2018-11-30 2021-08-24 华创车电技术中心股份有限公司 Method and system for detecting objects around vehicle
CN111552289A (en) * 2020-04-28 2020-08-18 苏州高之仙自动化科技有限公司 Detection method, virtual radar device, electronic apparatus, and storage medium
CN117148363A (en) * 2023-09-06 2023-12-01 广州优创电子有限公司 Method and system for detecting obstacle height by ultrasonic wave
CN117148363B (en) * 2023-09-06 2024-05-07 广州优创电子有限公司 Method and system for detecting obstacle height by ultrasonic wave

Similar Documents

Publication Publication Date Title
CN102765365B (en) Pedestrian detection method based on machine vision and pedestrian anti-collision warning system based on machine vision
CN105711597B (en) Front locally travels context aware systems and method
CN112700470B (en) Target detection and track extraction method based on traffic video stream
CN106611512B (en) Method, device and system for processing starting of front vehicle
EP3296923B1 (en) A method of detecting an overtaking vehicle, related processing system, overtaking vehicle detection system and vehicle
CN105488453B (en) A kind of driver based on image procossing does not fasten the safety belt detection recognition method
CN106647776B (en) Method and device for judging lane changing trend of vehicle and computer storage medium
CN104408932A (en) Drunk driving vehicle detection system based on video monitoring
CN106096525A (en) A kind of compound lane recognition system and method
CN106599832A (en) Method for detecting and recognizing various types of obstacles based on convolution neural network
CN202071799U (en) Intelligent automobile safety warning device
KR101735557B1 (en) System and Method for Collecting Traffic Information Using Real time Object Detection
CN103287372A (en) Automobile anti-collision safe protecting method based on image processing
CN102840853A (en) Obstacle detection and alarm method for vehicle-mounted night vision system
CN104200668A (en) Image-analysis-based detection method for helmet-free motorcycle driving violation event
Raviteja et al. Advanced driver assitance system (ADAS)
CN102768726A (en) Pedestrian detection method for preventing pedestrian collision
CN109389060A (en) A kind of vehicle week anti-collision warning method of view-based access control model
Tanaka et al. Vehicle Detection Based on Perspective Transformation Using Rear‐View Camera
CN108694363A (en) The method and apparatus that the pedestrian of vehicle periphery is detected
Huang Real-time multi-vehicle detection and sub-feature based tracking for traffic surveillance systems
CN206961331U (en) A kind of high threat vehicle monitoring early warning system
CN114119955A (en) Method and device for detecting potential dangerous target
Liu et al. Vehicle–bicyclist dynamic position extracted from naturalistic driving videos
CN203020184U (en) Electric vehicle lane marking line detecting device based on double charge coupled devices (CCD)

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20121226