CN107909024A - Vehicle tracking system, method and vehicle based on image recognition and infrared obstacle avoidance - Google Patents
Vehicle tracking system, method and vehicle based on image recognition and infrared obstacle avoidance Download PDFInfo
- Publication number
- CN107909024A CN107909024A CN201711116535.5A CN201711116535A CN107909024A CN 107909024 A CN107909024 A CN 107909024A CN 201711116535 A CN201711116535 A CN 201711116535A CN 107909024 A CN107909024 A CN 107909024A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- tracking
- avoidance
- image
- classifier
- 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
- 238000000034 method Methods 0.000 title claims description 51
- 238000012545 processing Methods 0.000 claims abstract description 45
- 238000011161 development Methods 0.000 claims abstract description 27
- 238000012544 monitoring process Methods 0.000 claims abstract description 16
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 41
- 238000012549 training Methods 0.000 claims description 28
- 238000005457 optimization Methods 0.000 claims description 10
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical compound OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 7
- 238000000205 computational method Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000006116 polymerization reaction Methods 0.000 claims description 6
- 238000009434 installation Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 19
- 238000005286 illumination Methods 0.000 abstract description 2
- 230000006872 improvement Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000005611 electricity Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007717 exclusion Effects 0.000 description 2
- 238000003780 insertion Methods 0.000 description 2
- 230000037431 insertion Effects 0.000 description 2
- 241001522296 Erithacus rubecula Species 0.000 description 1
- 241000209140 Triticum Species 0.000 description 1
- 235000021307 Triticum Nutrition 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
- Closed-Circuit Television Systems (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of vehicle tracking system based on image recognition and infrared obstacle avoidance, including：Host computer watchdog subsystem, tracking avoidance subsystem and vehicle control subsystems；The host computer watchdog subsystem is used for the pace and direction of advance for monitoring vehicle；The tracking avoidance subsystem is used for tracking and the avoidance of vehicle；The vehicle control subsystems are used for the pace and direction of advance for controlling vehicle；The host computer watchdog subsystem includes Wifi receiving modules, display module and monitoring module；The tracking avoidance subsystem includes：Camera module, infrared sensor module, Wifi sending modules, embedded image processing development board and tracking avoidance processing module；Solve easily be subject to caused by simple algorithm in the prior art illumination, block, object deformation etc. influences；Though also solves the poor technical problem of relative complex algorithm real-time.
Description
Technical field
The present invention relates to field in intelligent robotics, more particularly to a kind of vehicle tracking based on image recognition and infrared obstacle avoidance
System, method and vehicle.
Background technology
Object tracking has very big answer in recognition of face, independent navigation, vehicle identification, visual early warning in real time for a long time
Use prospect.Although the simple algorithm real-time used at present is preferable, easily be subject to illumination, block, object deformation etc. is influenced；
And although the relative complex algorithm used is not easy higher by external influence, precision, but real-time is poor.In addition, both
Algorithm be not implemented for front end camera information synchronous transfer to host computer monitor process.
The content of the invention
It is an object of the invention in view of the above-mentioned problems, provide a kind of vehicle based on image recognition and infrared obstacle avoidance with
Track system, method and vehicle.
The present invention can be realized using following system：A kind of vehicle tracking system based on image recognition and infrared obstacle avoidance
System, including：Host computer watchdog subsystem, tracking avoidance subsystem and vehicle control subsystems；
The host computer watchdog subsystem is used for the pace and direction of advance for monitoring vehicle；The tracking avoidance subsystem
Tracking and avoidance of the system for vehicle；The vehicle control subsystems are used for the pace and direction of advance for controlling vehicle；
The host computer watchdog subsystem includes Wifi receiving modules, display module and monitoring module；
The tracking avoidance subsystem includes：It is camera module, infrared sensor module, Wifi sending modules, embedded
Image procossing development board and tracking avoidance processing module；
The tracking avoidance processing module includes：(SuSE) Linux OS kernel submodule, ROS submodules and TLD are calculated
Submodule；
The vehicle control subsystems include：Controller and driving motor；
The realtime graphic that the Wifi sending modules are used to send the vehicle to be tracked that the camera module obtains is supreme
Position machine watchdog subsystem；The Wifi receiving modules are used for the real-time figure for receiving the vehicle to be tracked that the camera module obtains
Picture；The embedded image processing development board is used for the real-time image information of the vehicle to be tracked obtained to the camera module
It is acquired；The tracking avoidance processing module is used to handle the data through embedded image processing development board collection；
The TLD calculating sub modules include；It is excellent that object module improves unit, arest neighbors image sheet improvement unit, grader
Change unit and GPU accelerator modules；The object module improves unit and is used for Further aim model；The arest neighbors image sheet changes
It is used to improve arest neighbors image sheet into unit；The classifier optimization unit is used for Optimum Classification device；The GPU accelerator modules are used
In the calculating speed for accelerating TLD calculating sub modules；
The specific method of the Further aim model is：Preserve the realtime graphic for the vehicle to be tracked that camera module obtains
As off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out TLD calculating sub modules
Initialization；The parameter includes：Pass through all patch of Ensemble classifier, the image of the node of ten trees of Ensemble classifier
The positional value of coordinate points, by all positive negative samples of nearest neighbor classifier, also has each grader of cascade classifier
Threshold value；
The controller is used for real-time testing result and the processing module processing of tracking avoidance according to infrared sensor module
The output control driving motor of data, the data refer to the data of embedded image processing development board collection.
Further, the specific method of the improvement arest neighbors image sheet is：According to time and confidence level computational methods, from
Take-off time and positive sample image sheet and negative sample image sheet of the confidence level in values, institute in off-line training sample
The sum of quantity of positive sample image sheet and negative sample image sheet is stated as 20.
Further, the specific method of the Optimum Classification device is：By apart from clustering method, exclude beyond preset value away from
From vehicle frames images, the frames images of the vehicle in preset value distance range are clustered, calculate new target frame,
And reappraised with arest neighbors method；And judge when the result by nearest neighbor classifier as 0 patch when, again
Ensemble classifier is polymerize, while arest neighbors classification thresholds, the patch after polymerization is sent to nearest neighbor classifier weight again
New training.
Further, the GPU accelerator modules use GPU parallel acceleration methods, specially will be serial using CUDA platforms
Operation is changed to the concurrent operation of multithreading.
The present invention realizes with the following method：A kind of wireless vehicle tracking based on image recognition and infrared obstacle avoidance,
S201, the realtime graphic for obtaining by camera vehicle to be tracked；
S202, handle development board by embedded image the real-time image information for obtaining vehicle to be tracked be acquired,
S203, will collect real-time image information and be handled using TLD trackings, before calculating vehicle to be tracked
Into speed and direction of advance；
S204, according to the output control of the real-time testing result of infrared sensor and tracking avoidance processing data drive electricity
Machine, the data refer to the data of embedded image processing development board collection；And the real-time of vehicle will be tracked by WIFI module
Image information is uploaded to host computer monitoring；
Wherein, the TLD trackings include：Further aim model, improve arest neighbors image sheet and Optimum Classification device；
The specific method of the Further aim model is：Preserve the realtime graphic for the vehicle to be tracked that camera module obtains
As off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out TLD calculating sub modules
Initialization；The parameter includes：Pass through all patch of Ensemble classifier, the image of the node of ten trees of Ensemble classifier
The positional value of coordinate points, by all positive negative samples of nearest neighbor classifier, also has each grader of cascade classifier
Threshold value；
Further, the specific method of the improvement arest neighbors image sheet is：According to time and confidence level computational methods, from
Take-off time and positive sample image sheet and negative sample image sheet of the confidence level in values in off-line training sample.
Further, the sum of quantity of the positive sample image sheet and negative sample image sheet is 20.
Further, the specific method of the Optimum Classification device is：By apart from clustering method, exclude beyond preset value away from
From vehicle frames images, the frames images of the vehicle in preset value distance range are clustered, calculate new target frame,
And reappraised with arest neighbors method；And judge when the result by nearest neighbor classifier as 0 patch when, again
Ensemble classifier is polymerize, while arest neighbors classification thresholds, the patch after polymerization is sent to nearest neighbor classifier weight again
New training.
Further, the GPU accelerator modules use GPU parallel acceleration methods, specially will be serial using CUDA platforms
Operation is changed to the concurrent operation of multithreading.
The present invention also provides a kind of vehicle based on image recognition and infrared obstacle avoidance, which installs just like claim 1
To 5 any vehicle tracking systems based on image recognition and infrared obstacle avoidance.
To sum up, a kind of vehicle tracking system based on image recognition and infrared obstacle avoidance, including：Host computer watchdog subsystem,
Track avoidance subsystem and vehicle control subsystems；The host computer watchdog subsystem be used to monitoring the pace of vehicle with
Direction of advance；The tracking avoidance subsystem is used for tracking and the avoidance of vehicle；The vehicle control subsystems are used to control car
Pace and direction of advance；The host computer watchdog subsystem includes Wifi receiving modules, display module and monitoring
Module；The tracking avoidance subsystem includes：Camera module, infrared sensor module, Wifi sending modules, embedded image
Handle development board and tracking avoidance processing module；The tracking avoidance processing module includes：(SuSE) Linux OS kernel submodule
Block, ROS submodules and TLD calculating sub modules；The vehicle control subsystems include：Controller and driving motor；It is described
Realtime graphic to the host computer that Wifi sending modules are used to send the vehicle to be tracked that the camera module obtains monitors subsystem
System；The Wifi receiving modules are used for the realtime graphic for receiving the vehicle to be tracked that the camera module obtains；The insertion
The real-time image information for the vehicle to be tracked that formula image procossing development board is used to obtain the camera module is acquired；Institute
Tracking avoidance processing module is stated to be used to handle the data through embedded image processing development board collection；The TLD calculates son
Module includes；Object module improves unit, arest neighbors image sheet improves unit, classifier optimization unit and GPU accelerator modules；
The object module improves unit and is used for Further aim model；The arest neighbors image sheet improves unit and is used to improve arest neighbors figure
Photo；The classifier optimization unit is used for Optimum Classification device；The GPU accelerator modules are used to accelerate TLD calculating sub modules
Calculating speed；The specific method of the Further aim model is：Preserve the real-time figure for the vehicle to be tracked that camera module obtains
As being used as off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out TLD calculating sub modules
Initialization and only arest neighbors image sheet and variance yields are preserved；The controller is used for according to infrared sensor module
The output control driving motor of real-time testing result and tracking avoidance processing module processing data, the data refer to embedded figure
As the data of processing development board collection.
Have the beneficial effect that：
1. the present invention can carry out real-time tracing by camera to vehicle to be tracked, and by infrared sensor to periphery
Environment carries out avoidance detection；
2. using improved object module algorithm, the realtime graphic conduct for the vehicle to be tracked that camera module obtains is preserved
Off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out the initialization of TLD computational algorithms,
Realize offline inspection and on-line study ability；Reduce the occupied space of model so that the time of storage and loaded targets model
Greatly shorten；
Brief description of the drawings
In order to illustrate more clearly of technical scheme, letter will be made to attached drawing needed in the embodiment below
Singly introduce, it should be apparent that, drawings in the following description are only some embodiments described in the present invention, for this area
For those of ordinary skill, without creative efforts, other attached drawings can also be obtained according to these attached drawings.
Fig. 1 is the vehicle tracking system example structure figure provided by the invention based on image recognition and infrared obstacle avoidance；
Fig. 2 is the wireless vehicle tracking embodiment flow chart provided by the invention based on image recognition and infrared obstacle avoidance；
Fig. 3 is to track avoidance in the vehicle tracking system provided by the invention based on image recognition and infrared obstacle avoidance to handle mould
Block structural diagram.
Embodiment
The present invention gives a kind of vehicle tracking system and embodiment of the method based on image recognition and infrared obstacle avoidance, in order to
Those skilled in the art is more fully understood the technical solution in the embodiment of the present invention, and make the above-mentioned purpose of the present invention, spy
Advantage of seeking peace can be more obvious understandable, and technical solution in the present invention is described in further detail below in conjunction with the accompanying drawings：
Present invention firstly provides the vehicle tracking system based on image recognition and infrared obstacle avoidance, as shown in Figure 1, host computer
Watchdog subsystem 10, tracking avoidance subsystem 20 and vehicle control subsystems 30；
The host computer watchdog subsystem 10 is used for the pace and direction of advance for monitoring vehicle；Tracking avoidance
System 20 is used for tracking and the avoidance of vehicle；The vehicle control subsystems 30 are used for pace and the advance side for controlling vehicle
To；
The host computer watchdog subsystem 10 includes Wifi receiving modules 101, display module 102 and monitoring module 103；
The tracking avoidance subsystem 20 includes：Camera module 201, infrared sensor module 202, Wifi sending modules
203rd, embedded image processing development board 204 and tracking avoidance processing module 205；
The tracking avoidance processing module 205, including：(SuSE) Linux OS kernel submodule 2051, ROS submodules
2052 and TLD calculating sub modules 2053；
The vehicle control subsystems 30 include：Controller 301 and driving motor 302；
Wherein, host computer watchdog subsystem be one based under linux system P2P server clients receive model,
It receives the real time data of the camera collection of WIFI module transmission and display.Essence on development board is handled by embedded image
Clever process Client clients can read the picture that TLD computational methods methods are stored under specified path in real time, pass through Linux networks
Picture is sent to host computer and is monitored in real time by programmed method.
Wherein, controller employs STM32 controllers, and controller uses driving motor driving wheat according to processing data result
Ke Namu omni-directional wheels are moved, and reach the effect of tracking, and at the same time, infrared sensor measures real-time surrounding enviroment, will examine
The data of survey pass STM32 controllers back and carry out avoidance judgement.
Wherein, tracking obstacle avoidance algorithm operation principle is：By improve TLD algorithms and ROS systems realize target it is automatic with
Track, the Automatic Target Tracking algorithm steps include：Three target detection, target study, target following modules, by mutual
Supervision and study, reach long-time target following effect.
The Wifi sending modules 203 are used for the real-time figure for sending the vehicle to be tracked that the camera module 201 obtains
As to host computer watchdog subsystem 10；The Wifi receiving modules 101 be used to receiving the camera module 201 obtains treat with
The realtime graphic of track vehicle；What embedded image processing development board 204 was used to obtain the camera module 201 treat with
The real-time image information of track vehicle is acquired；The tracking avoidance processing module 205 is used to handle through the embedded image
Handle the data of development board collection；
The TLD calculating sub modules 2053 include；Object module improves unit 20531, arest neighbors image sheet improves unit
20532nd, classifier optimization unit 20533 and GPU accelerator modules 20534；The object module improves unit 20531 and is used to change
Into object module；The arest neighbors image sheet improves unit 20532 and is used to improve arest neighbors image sheet；The classifier optimization list
Member 20533 is used for Optimum Classification device；The GPU accelerator modules 20534 are used for the calculating speed for accelerating TLD calculating sub modules；
The specific method of the Further aim model is：Preserve the realtime graphic for the vehicle to be tracked that camera module obtains
As off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out TLD calculating sub modules
Initialization；The parameter includes：Pass through all patch of Ensemble classifier, the image of the node of ten trees of Ensemble classifier
The positional value of coordinate points, by all positive negative samples of nearest neighbor classifier, also has each grader of cascade classifier
Threshold value；
Wherein, the parameter of the first two field picture includes all data of first frame by variance grader；Including but not limited to
Pass through the positive negative sample of Ensemble classifier, each threshold value, variance by the positive negative sample of nearest neighbor classifier, cascade classifier
The threshold value of grader, the prior probability of Ensemble classifier, the threshold value of nearest neighbor classifier, on-time model 20 are opened, Ensemble classifier
10 are set, picture position of each node in each tree etc.；The process of initialization is to being saved in above-mentioned vector data
Inside one file, the form of matrix is saved as, the data in file are directly read when starting shooting for second, so as to carry out initial
Change；Used nearest neighbor classifier, is defined the quantity of on-time model, and preceding 10 initialization are motionless, and latter 10 are adopted
It is cyclically updated with Round robin algorithms.
The controller is used for real-time testing result and the processing module processing of tracking avoidance according to infrared sensor module
The output control driving motor of data, the data refer to the data of embedded image processing development board collection.
Preferably, the specific method of the improvement arest neighbors image sheet is：According to time and confidence level computational methods, from from
Take-off time and positive sample image sheet and negative sample image sheet of the confidence level in values in line training sample.
Above-mentioned improvement can reduce arest neighbors calculate and matched object module in image sheet so that reduce calculation amount with
And the running memory in accounting for
Preferably, the sum of quantity of the positive sample image sheet and negative sample image sheet is 20.
Preferably, the specific method of the Optimum Classification device is：By exceeding preset value distance apart from clustering method, exclusion
Vehicle frames images, the frames images of the vehicle in preset value distance range are clustered, calculate new target frame, and
Reappraised with arest neighbors method；And judge when the result by nearest neighbor classifier as 0 patch when, it is again right
Ensemble classifier is polymerize, while arest neighbors classification thresholds, and the patch after polymerization is sent to nearest neighbor classifier again again
Training..
Preferably, the GPU accelerator modules use GPU parallel acceleration methods, specially will serially be transported using CUDA platforms
Row is changed to the concurrent operation of multithreading, handles multiple scanning windows at the same time using multithreading, handling result is come back for assessing.
Present invention also offers the wireless vehicle tracking embodiment based on image recognition and infrared obstacle avoidance, as shown in Fig. 2, bag
Include：
The present invention realizes with the following method：A kind of wireless vehicle tracking based on image recognition and infrared obstacle avoidance,
S201, the realtime graphic for obtaining by camera vehicle to be tracked；
S202, handle development board by embedded image the real-time image information for obtaining vehicle to be tracked be acquired；
S203, will collect real-time image information and be handled using TLD trackings, before calculating vehicle to be tracked
Into speed and direction of advance；
S204, according to the output control of the real-time testing result of infrared sensor and tracking avoidance processing data drive electricity
Machine, the data refer to the data of embedded image processing development board collection；And the real-time of vehicle will be tracked by WIFI module
Image information is uploaded to host computer monitoring；
Wherein, the TLD trackings include：Further aim model, improve arest neighbors image sheet and Optimum Classification device；
The specific method of the Further aim model is：Preserve the realtime graphic for the vehicle to be tracked that camera module obtains
As off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out TLD calculating sub modules
Initialization；The parameter includes：Pass through all patch of Ensemble classifier, the image of the node of ten trees of Ensemble classifier
The positional value of coordinate points, by all positive negative samples of nearest neighbor classifier, also has each grader of cascade classifier
Threshold value；
Preferably, the specific method of the improvement arest neighbors image sheet is：According to time and confidence level computational methods, from from
Take-off time and positive sample image sheet and negative sample image sheet of the confidence level in values, described in line training sample
The sum of quantity of positive sample image sheet and negative sample image sheet is 20.
The above method reduces the image sheet in arest neighbors calculating and matched object module, in reducing calculation amount and accounting for
Running memory.
Preferably, the specific method of the Optimum Classification device is：By exceeding preset value distance apart from clustering method, exclusion
Vehicle frames images, the frames images of the vehicle in preset value distance range are clustered, calculate new target frame, and
Reappraised with arest neighbors method；And judge when the result by nearest neighbor classifier as 0 patch when, it is again right
Ensemble classifier is polymerize, while arest neighbors classification thresholds, and the patch after polymerization is sent to nearest neighbor classifier again again
Training.
Preferably, the GPU accelerator modules use GPU parallel acceleration methods, specially will serially be transported using CUDA platforms
Row is changed to the concurrent operation of multithreading.
The present invention also provides a kind of vehicle based on image recognition and infrared obstacle avoidance, which installs just like claim 1
To 5 any vehicle tracking systems based on image recognition and infrared obstacle avoidance.
In conclusion the present invention writes algorithm using time and confidence level, taken out from model most possess it is representational just
Image sheet and negative image piece, and the quantity for limiting positive and negative image sheet calculates and matched target at 20 so as to reduce arest neighbors
Image sheet in model, the running memory in reducing calculation amount and accounting for；The present invention is also by apart from clustering algorithm, will be far from mesh
Target frames images exclude, and are clustered in the frames images of target proximity, and synthesis goes out new target frame, and is commented again with arest neighbors
Estimate so that the classifying quality of grader is more accurate；In addition the present invention carries out algorithm using CUDA multithreads computings technology
Optimization, by serial computing method is rewritten into parallel computation on CPU originally, improves the speed of service of algorithm, when shortening operation
Between.
The present invention provides a kind of vehicle tracking system based on image recognition and infrared obstacle avoidance, including：Host computer monitoring
System, tracking avoidance subsystem and vehicle control subsystems；The host computer watchdog subsystem is used for the advance for monitoring vehicle
Speed and direction of advance；The tracking avoidance subsystem is used for tracking and the avoidance of vehicle；The vehicle control subsystems are used for
Control the pace and direction of advance of vehicle；The host computer watchdog subsystem include Wifi receiving modules, display module with
And monitoring module；The tracking avoidance subsystem includes：Camera module, infrared sensor module, Wifi sending modules, insertion
Formula image procossing development board and tracking avoidance processing module；The tracking avoidance processing module includes：In (SuSE) Linux OS
Nucleon module, ROS submodules and TLD calculating sub modules；The vehicle control subsystems include：Controller and driving electricity
Machine；Realtime graphic to the host computer that the Wifi sending modules are used to send the vehicle to be tracked that the camera module obtains is supervised
Depending on subsystem；The Wifi receiving modules are used for the realtime graphic for receiving the vehicle to be tracked that the camera module obtains；Institute
The real-time image information for stating the vehicle to be tracked that embedded image processing development board is used to obtain the camera module carries out
Collection；The tracking avoidance processing module is used to handle the data through embedded image processing development board collection；The TLD
Calculating sub module includes；Object module improves unit, arest neighbors image sheet improves unit, classifier optimization unit and GPU and adds
Fast unit；The object module improves unit and is used for Further aim model；The arest neighbors image sheet improves unit and is used to improve
Arest neighbors image sheet；The classifier optimization unit is used for Optimum Classification device；The GPU accelerator modules are used to accelerate TLD calculating
The calculating speed of submodule.The present invention by varying former algorithm can only manual setting initialized target, then track process every time,
Invention saves the information of the first frame image data as initialization；The cascade sort brought because GPU accelerates is improved at the same time
The time-consuming problem of device, the present invention are that other side's difference grader is accelerated, and former algorithm is run on cpu, it utilizes NVIDIA
TK1 planks carry video card embedded plank, with GPU parallel computations, accelerate the calculating of variance grader.It is used most
Nearest Neighbor Classifier, is defined the quantity of on-time model, solves former algorithm and unrestrictedly increases, embedded platform is not
The problem of supporting the quantity by cascade classifier, and aggregating algorithm has been used, plurality of pictures normalizing, ensure that tracking just
True property.
Above example is to illustrative and not limiting technical scheme.Appointing for spirit and scope of the invention is not departed from
What modification or local replacement, should all cover among scope of the presently claimed invention.
Claims (9)
- A kind of 1. vehicle tracking system based on image recognition and infrared obstacle avoidance, it is characterised in that including：Host computer monitors subsystem System, tracking avoidance subsystem and vehicle control subsystems；The host computer watchdog subsystem is used for the pace and direction of advance for monitoring vehicle；The tracking avoidance subsystem is used Tracking and avoidance in vehicle；The vehicle control subsystems are used for the pace and direction of advance for controlling vehicle；The host computer watchdog subsystem includes Wifi receiving modules, display module and monitoring module；The tracking avoidance subsystem includes：Camera module, infrared sensor module, Wifi sending modules, embedded image Handle development board and tracking avoidance processing module；The tracking avoidance processing module includes：(SuSE) Linux OS kernel submodule Block, ROS submodules and TLD calculating sub modules；The vehicle control subsystems include：Controller and driving motor；The Wifi sending modules are used to sending the realtime graphic of the vehicle to be tracked that the camera module obtains to host computer Watchdog subsystem；The Wifi receiving modules are used for the realtime graphic for receiving the vehicle to be tracked that the camera module obtains； Embedded image processing development board be used for the real-time image information of the vehicle to be tracked obtained to the camera module into Row collection；The tracking avoidance processing module is used to handle the data through embedded image processing development board collection；The TLD calculating sub modules include；Object module improves unit, arest neighbors image sheet improves unit, classifier optimization list Member and GPU accelerator modules；The object module improves unit and is used for Further aim model；The arest neighbors image sheet improves single Member is used to improve arest neighbors image sheet；The classifier optimization unit is used for Optimum Classification device；The GPU accelerator modules are used to add The calculating speed of fast TLD calculating sub modules；The specific method of the Further aim model is：Preserve the realtime graphic conduct for the vehicle to be tracked that camera module obtains Off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out the initial of TLD calculating sub modules Change；The parameter includes：Pass through all patch of Ensemble classifier, the image coordinate of the node of ten trees of Ensemble classifier The positional value of point, by all positive negative samples of nearest neighbor classifier, also has the threshold of each grader of cascade classifier Value；The controller is used for real-time testing result and tracking avoidance processing module processing data according to infrared sensor module Output control driving motor, the data refer to embedded image processing development board collection data.
- 2. the vehicle tracking system based on image recognition and infrared obstacle avoidance as claimed in claim 1, it is characterised in that described to change Specific method into arest neighbors image sheet is：According to time and confidence level computational methods, the take-off time from off-line training sample With positive sample image sheet and negative sample image sheet of the confidence level in values, the positive sample image sheet and negative sample The sum of quantity of image sheet is 20.
- 3. the vehicle tracking system based on image recognition and infrared obstacle avoidance as claimed in claim 1, it is characterised in thatThe specific method of the Optimum Classification device is：By apart from clustering method, excluding the figure for exceeding the vehicle of preset value distance Frame, the frames images of the vehicle in preset value distance range are clustered, and calculate new target frame, and with arest neighbors side Method is reappraised；And judge when the result by nearest neighbor classifier as 0 patch when, again to Ensemble classifier It is polymerize, while arest neighbors classification thresholds, the patch after polymerization is sent to nearest neighbor classifier re -training again.
- 4. the vehicle tracking system according to claim 1 based on image recognition and infrared obstacle avoidance, it is characterised in that described GPU accelerator modules use GPU parallel acceleration methods, and serial operation is specially changed to the parallel fortune of multithreading using CUDA platforms Calculate.
- A kind of 5. wireless vehicle tracking based on image recognition and infrared obstacle avoidance, it is characterised in that including：S201, the realtime graphic for obtaining by camera vehicle to be tracked；S202, handle development board by embedded image the real-time image information for obtaining vehicle to be tracked be acquired,S203, will collect real-time image information and be handled using TLD trackings, calculate the advance speed of vehicle to be tracked Degree and direction of advance；S204, according to the output control of the real-time testing result of infrared sensor and tracking avoidance processing data drive motor, institute State the data that data refer to embedded image processing development board collection；And the realtime graphic for tracking vehicle is believed by WIFI module Breath is uploaded to host computer monitoring；Wherein, the TLD trackings include：Further aim model, improve arest neighbors image sheet and Optimum Classification device；The specific method of the Further aim model is：Preserve the realtime graphic conduct for the vehicle to be tracked that camera module obtains Off-line training sample, and the parameter for reading the first two field picture that off-line training is completed automatically carries out the initial of TLD calculating sub modules Change；The parameter includes：Pass through all patch of Ensemble classifier, the image coordinate of the node of ten trees of Ensemble classifier The positional value of point, by all positive negative samples of nearest neighbor classifier, also has the threshold of each grader of cascade classifier Value.
- 6. the wireless vehicle tracking based on image recognition and infrared obstacle avoidance as claimed in claim 5, it is characterised in that described to change Specific method into arest neighbors image sheet is：According to time and confidence level computational methods, the take-off time from off-line training sample With positive sample image sheet and negative sample image sheet of the confidence level in values, the positive sample image sheet and negative sample The sum of quantity of image sheet is 20.
- 7. the wireless vehicle tracking based on image recognition and infrared obstacle avoidance as claimed in claim 5, it is characterised in thatThe specific method of the Optimum Classification device is：By apart from clustering method, excluding the figure for exceeding the vehicle of preset value distance Frame, the frames images of the vehicle in preset value distance range are clustered, and calculate new target frame, and with arest neighbors side Method is reappraised；And judge when the result by nearest neighbor classifier as 0 patch when, again to Ensemble classifier It is polymerize, while arest neighbors classification thresholds, the patch after polymerization is sent to nearest neighbor classifier re -training again.
- 8. the wireless vehicle tracking according to claim 5 based on image recognition and infrared obstacle avoidance, it is characterised in that described GPU accelerator modules use GPU parallel acceleration methods, and serial operation is specially changed to the parallel fortune of multithreading using CUDA platforms Calculate.
- 9. a kind of vehicle based on image recognition and infrared obstacle avoidance, it is characterised in that installation is just like any institute of claim 1 to 5 The vehicle tracking system based on image recognition and infrared obstacle avoidance stated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711116535.5A CN107909024B (en) | 2017-11-13 | 2017-11-13 | Vehicle tracking system and method based on image recognition and infrared obstacle avoidance and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711116535.5A CN107909024B (en) | 2017-11-13 | 2017-11-13 | Vehicle tracking system and method based on image recognition and infrared obstacle avoidance and vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107909024A true CN107909024A (en) | 2018-04-13 |
| CN107909024B CN107909024B (en) | 2021-11-05 |
Family
ID=61843896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711116535.5A Expired - Fee Related CN107909024B (en) | 2017-11-13 | 2017-11-13 | Vehicle tracking system and method based on image recognition and infrared obstacle avoidance and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107909024B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110825123A (en) * | 2019-10-21 | 2020-02-21 | 哈尔滨理工大学 | Control system and method for automatic following loading vehicle based on motion algorithm |
| CN112347953A (en) * | 2020-11-11 | 2021-02-09 | 上海伯镭智能科技有限公司 | Recognition device for road condition irregular obstacles of unmanned vehicle |
| CN112508865A (en) * | 2020-11-23 | 2021-03-16 | 深圳供电局有限公司 | Unmanned aerial vehicle inspection obstacle avoidance method and device, computer equipment and storage medium |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140022378A1 (en) * | 2004-12-23 | 2014-01-23 | Magna Electronics Inc. | Driver assistance system for vehicle |
| CN103838244A (en) * | 2014-03-20 | 2014-06-04 | 湖南大学 | Portable target tracking method and system based on four-axis air vehicle |
| CN105631798A (en) * | 2016-03-04 | 2016-06-01 | 北京理工大学 | Low-power consumption portable real-time image target detecting and tracking system and method thereof |
| CN105844664A (en) * | 2016-03-21 | 2016-08-10 | 辽宁师范大学 | Monitoring video vehicle detection tracking method based on improved TLD |
| CN106303461A (en) * | 2016-09-08 | 2017-01-04 | 福建师范大学 | Movable-type intelligent safety device based on video analysis |
| WO2017044550A1 (en) * | 2015-09-11 | 2017-03-16 | Intel Corporation | A real-time multiple vehicle detection and tracking |
| US20170116488A1 (en) * | 2015-10-23 | 2017-04-27 | MAGNETI MARELLI S.p.A. | Method for identifying an incoming vehicle and corresponding system |
| CN106846362A (en) * | 2016-12-26 | 2017-06-13 | 歌尔科技有限公司 | A kind of target detection tracking method and device |
| CN106886748A (en) * | 2016-12-28 | 2017-06-23 | 中国航天电子技术研究院 | A kind of mutative scale method for tracking target suitable for unmanned plane based on TLD |
| CN107105159A (en) * | 2017-04-13 | 2017-08-29 | 山东万腾电子科技有限公司 | The real-time detecting and tracking system and method for embedded moving target based on SoC |
| WO2017185503A1 (en) * | 2016-04-29 | 2017-11-02 | 高鹏 | Target tracking method and apparatus |
-
2017
- 2017-11-13 CN CN201711116535.5A patent/CN107909024B/en not_active Expired - Fee Related
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140022378A1 (en) * | 2004-12-23 | 2014-01-23 | Magna Electronics Inc. | Driver assistance system for vehicle |
| CN103838244A (en) * | 2014-03-20 | 2014-06-04 | 湖南大学 | Portable target tracking method and system based on four-axis air vehicle |
| WO2017044550A1 (en) * | 2015-09-11 | 2017-03-16 | Intel Corporation | A real-time multiple vehicle detection and tracking |
| US20170116488A1 (en) * | 2015-10-23 | 2017-04-27 | MAGNETI MARELLI S.p.A. | Method for identifying an incoming vehicle and corresponding system |
| CN105631798A (en) * | 2016-03-04 | 2016-06-01 | 北京理工大学 | Low-power consumption portable real-time image target detecting and tracking system and method thereof |
| CN105844664A (en) * | 2016-03-21 | 2016-08-10 | 辽宁师范大学 | Monitoring video vehicle detection tracking method based on improved TLD |
| WO2017185503A1 (en) * | 2016-04-29 | 2017-11-02 | 高鹏 | Target tracking method and apparatus |
| CN106303461A (en) * | 2016-09-08 | 2017-01-04 | 福建师范大学 | Movable-type intelligent safety device based on video analysis |
| CN106846362A (en) * | 2016-12-26 | 2017-06-13 | 歌尔科技有限公司 | A kind of target detection tracking method and device |
| CN106886748A (en) * | 2016-12-28 | 2017-06-23 | 中国航天电子技术研究院 | A kind of mutative scale method for tracking target suitable for unmanned plane based on TLD |
| CN107105159A (en) * | 2017-04-13 | 2017-08-29 | 山东万腾电子科技有限公司 | The real-time detecting and tracking system and method for embedded moving target based on SoC |
Non-Patent Citations (7)
| Title |
|---|
| CLAUDIO CARAFFI 等: "A System for Real-time Detection and Tracking of Vehicles from a Single Car-mounted Camera", 《IEEE CONFERENCE ON INTELLIGENT TRANSPORTATION SYSTEMS 》 * |
| V. CADENAT 等: "An hybrid control for avoiding obstacles during a vision-based tracking task", 《EUROPEAN CONTROL CONFERENCE》 * |
| ZDENEK KALAL 等: "Tracking-Learning-Detection", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》 * |
| 刘阔 等: "低分辨率条件下基于TLD的鲁棒车辆跟踪算法", 《计算机应用与软件》 * |
| 强彦 等: "基于红外避障的智能小车的设计", 《微电子学与计算机》 * |
| 曲海成 等: "检测区域动态调整的TLD目标跟踪算法", 《计算机应用》 * |
| 雷鹏飞 等: "红外传感器在智能车避障系统的应用", 《电脑与信息技术》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110825123A (en) * | 2019-10-21 | 2020-02-21 | 哈尔滨理工大学 | Control system and method for automatic following loading vehicle based on motion algorithm |
| CN112347953A (en) * | 2020-11-11 | 2021-02-09 | 上海伯镭智能科技有限公司 | Recognition device for road condition irregular obstacles of unmanned vehicle |
| CN112347953B (en) * | 2020-11-11 | 2021-09-28 | 上海伯镭智能科技有限公司 | Recognition device for road condition irregular obstacles of unmanned vehicle |
| CN112508865A (en) * | 2020-11-23 | 2021-03-16 | 深圳供电局有限公司 | Unmanned aerial vehicle inspection obstacle avoidance method and device, computer equipment and storage medium |
| CN112508865B (en) * | 2020-11-23 | 2024-02-02 | 深圳供电局有限公司 | Unmanned aerial vehicle inspection obstacle avoidance method, unmanned aerial vehicle inspection obstacle avoidance device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107909024B (en) | 2021-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11893785B2 (en) | Object annotation method and apparatus, movement control method and apparatus, device, and storage medium | |
| Leira et al. | Automatic detection, classification and tracking of objects in the ocean surface from UAVs using a thermal camera | |
| CN109685141B (en) | Robot article sorting visual detection method based on deep neural network | |
| CN111612841B (en) | Target positioning method and device, mobile robot and readable storage medium | |
| CN110533687B (en) | Multi-target three-dimensional track tracking method and device | |
| CN107909024A (en) | Vehicle tracking system, method and vehicle based on image recognition and infrared obstacle avoidance | |
| Ahmad et al. | Overhead view person detection using YOLO | |
| US11227374B2 (en) | Method and apparatus for realtime object detection in unmanned aerial vehicle image | |
| CN111913177A (en) | Method and device for detecting target object and storage medium | |
| Lou et al. | Measuring loblolly pine crowns with drone imagery through deep learning | |
| Coşkun et al. | Implementation of tracking of a moving object based on camshift approach with a UAV | |
| CN110992378A (en) | Dynamic update visual tracking aerial photography method and system based on rotor flying robot | |
| Khudov et al. | Devising a method for segmenting complex structured images acquired from space observation systems based on the particle swarm algorithm | |
| Ajakwe et al. | Tractable minacious drones aerial recognition and safe-channel neutralization scheme for mission critical operations | |
| Yazdanpanah et al. | Sky segmentation by fusing clustering with neural networks | |
| Le et al. | Human detection and tracking for autonomous human-following quadcopter | |
| CN117474950A (en) | Cross-modal target tracking method based on visual semantics | |
| CN111814653A (en) | Method, device, equipment and storage medium for detecting abnormal behaviors in video | |
| Wang et al. | Online drone-based moving target detection system in dense-obstructer environment | |
| WO2022214821A2 (en) | Monocular depth estimation | |
| Makirin et al. | Onboard visual drone detection for drone chasing and collision avoidance | |
| CN114648568A (en) | Object size recognition method and device, computer equipment and storage medium | |
| CN112508970A (en) | Point cloud data segmentation method and device | |
| Zhou et al. | A vision-based autonomous detection scheme for obstacles on the runway | |
| CN104182990A (en) | A method for acquiring a sequence image motion target area in real-time |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20211105 |