CN107798688A - Motion estimate method, method for early warning and automobile anti-rear end collision prior-warning device - Google Patents
Motion estimate method, method for early warning and automobile anti-rear end collision prior-warning device Download PDFInfo
- Publication number
- CN107798688A CN107798688A CN201711046921.1A CN201711046921A CN107798688A CN 107798688 A CN107798688 A CN 107798688A CN 201711046921 A CN201711046921 A CN 201711046921A CN 107798688 A CN107798688 A CN 107798688A
- Authority
- CN
- China
- Prior art keywords
- pixel
- mrow
- block
- pixels
- value
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/248—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/215—Motion-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of motion estimate method, method for early warning and automobile anti-rear end collision prior-warning device, wherein recognition methods includes step：According to the heterochromia degree where each pixel in the two field picture of acquisition between block of pixels and surrounding pixel block, the distance between and with surrounding pixel block, calculate the color of each pixel and the otherness measure value of distance, multiple block of pixels of threshold value are estimated as reference less than otherness using otherness measure value, calculate the multiple with reference to otherness measure value of each pixel, obtain the notable feature value of each pixel, the pixel of characteristic threshold value is more than according to notable feature value and its notable feature value obtains color and apart from notable hum pattern, motion feature image is determined using continuous multiple frames color and apart from notable hum pattern, and determine the moving target in current frame image.The heterochromia of above-mentioned motion estimate method, combining target and background, the nearlyer likelihood probability of block of pixels distance be bigger and Target Motion Character, and target can be accurately and efficiently identified from two field picture.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of motion estimate method, method for early warning with
And automobile anti-rear end collision prior-warning device.
Background technology
Target identification technology based on image feature information extraction all has a wide range of applications in every field, target identification
Technology can use image analysis method, and extraction identification is carried out to the target in image.Such as in safety monitoring warning aspect, especially
It is the fields such as the anti-early warning of knocking into the back of automobile, the information such as the vehicle to automobile front and back, pedestrian can be used to carry out video monitoring simultaneously
The motion state of target prediction target is extracted, early warning is sent when target may knock into the back, this requires to enter to video
During row target identification with motion estimate method there is the real-time and accuracy of height.
Existing motion estimate method, extraction identification is carried out to the target in image according to clarification of objective, but
Extraction identification often to target is not accurate enough, causes the accuracy deficiency in application, such as because target is known during early warning analysis
It is not inaccurate and cause to prejudge mistake, the warning information of mistake is obtained, reduces the accuracy and reliability of safe early warning.
The content of the invention
Based on this, it is necessary to for conventional target identification technology accuracy it is low the problem of, there is provided a kind of motion estimate
Method, method for early warning and automobile anti-rear end collision prior-warning device.
A kind of motion estimate method, comprises the following steps：
According to the heterochromia degree where each pixel in the current frame image of acquisition between block of pixels and surrounding pixel block, with
And the distance between block of pixels and surrounding pixel block where the pixel, calculate the color of each pixel and the otherness of distance is estimated
Value；
Otherness is less than using the otherness measure value and estimates block of pixels where the pixel of threshold value as reference pixels block, is divided
The reference otherness measure value of block of pixels where not calculating each pixel, each pixel is obtained with reference to otherness measure value according to described
Notable feature value；
The pixel that notable feature value in current frame image is more than to notable feature threshold value is set to notable feature pixel, according to institute
State notable feature pixel and its notable feature value obtains the color of the current frame image and apart from notable hum pattern；
Determine that motion is special using the color of the continuous multiple frames image comprising the current frame image and apart from notable hum pattern
Image is levied, and the moving target in the current frame image is identified according to the motion feature image.
A kind of motion estimate device is also provided, including：
First computing module, for block of pixels and surrounding pixel block where each pixel in the current frame image according to acquisition it
Between heterochromia degree, and the pixel where the distance between block of pixels and surrounding pixel block, calculate each pixel color and
The otherness measure value of distance；
Second computing module, for being less than the pixel where otherness estimates the pixel of threshold value with the otherness measure value
Block is reference pixels block, the reference otherness measure value of block of pixels where calculating each pixel respectively, according to described with reference to otherness
Measure value obtains the notable feature value of each pixel；
Hum pattern acquisition module, the pixel for notable feature value in current frame image to be more than to notable feature threshold value are set to
Notable feature pixel, the color of the current frame image is obtained and apart from aobvious according to the notable feature pixel and its notable feature value
Write hum pattern；
Target Acquisition module, for the color using the continuous multiple frames image comprising the current frame image and apart from notable
Hum pattern determines motion feature image, and identifies the moving target in the current frame image according to the motion feature image.
Above-mentioned motion estimate method and apparatus, combine the color of target and the difference of background, block of pixels distance are got over
Many features such as nearly likelihood probability is bigger and target is moved, color, block of pixels distance and fortune are carried out to video frame images
Dynamic combination features computing extraction, rejects the background of video frame images, obtains target, the motion estimate method is to target
Identification precise and high efficiency, target can be accurately extracted from video frame images.
A kind of moving target method for early warning, comprises the following steps：
Obtain the two field picture shot to the environmental information of direction initialization around carrier；
All targets in current frame image are identified, according to pair of each pixel and actual range in default two field picture
Relation acquisition and the closest target of carrier are answered, and obtains the distance between the nearest target and carrier information；Wherein, adopt
With the target in the motion estimate method identification two field picture described in as above any one；
It is and described according to the time interval of the setting of adjacent two field pictures and the carrier present speed and acceleration that obtain
The distance between target and carrier prejudge the relative position information of next setting time section target and carrier；
According to the next setting time section target and carrier of default speed-acceleration-Safety distance model and anticipation
Relative position information carries out early warning.
Above-mentioned moving target method for early warning, with above-mentioned motion estimate method identify shooting two field picture in carrier
Closest target, early warning is carried out by analyzing the position of next setting time section target of anticipation.Due to carrying out target
Combine that the color of target and the difference of background, the nearlyer likelihood probability of block of pixels distance be bigger during identification and target motion etc. is more
The feature of aspect, video frame images are carried out with color, block of pixels distance and the combination features computing of motion and is extracted, rejects video
The background of two field picture, accurately efficiently it can extract target from the video frame images of shooting.So as to current
The motion state of carrier surrounding objects object makes more accurately anticipation and early warning, lifted to the accuracy of moving target early warning and
Reliability.
A kind of automobile anti-rear end collision prior-warning device, described device are configured to perform as described above when performing and preventing knocking into the back early warning
Moving target method for early warning.
Above-mentioned automobile anti-rear end collision prior-warning device, shoots to the environmental information of motor vehicle environment direction initialization, accurate and high
Effect ground extracts target from the video frame images of shooting, the motion state of current motor vehicle environment target object is made precisely and
When prejudge, carry out early warning when that may knock into the back, the accuracy and reliability of lifting automobile anti-rear end collision early warning.
Brief description of the drawings
Fig. 1 is motion estimate method flow diagram；
Fig. 2 is motion estimate apparatus structure schematic diagram；
Fig. 3 is moving target method for early warning flow chart；
Fig. 4 is the automobile anti-rear end collision prior-warning device structural representation of one embodiment.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Referring to Fig. 1, Fig. 1 is motion estimate method flow diagram, and the motion estimate method comprises the following steps：
S101, according to the heterochromia where each pixel in the current frame image of acquisition between block of pixels and surrounding pixel block
The distance between block of pixels and surrounding pixel block where degree, and the pixel, calculate the color of each pixel and the otherness of distance
Measure value.
Often existing heterochromia between moving target pixel and background pixel, such as automobile and road are in color and bright
Feature on degree etc. has differences, therefore can extract specific block of pixels according to the heterochromia of target and background.
Meanwhile the positional distance between block of pixels and block also possesses notable feature.Background block often exists in the video frame
Many similar characteristics, when it with those is considered as that striking contrast occurs when gathering together in notable block, if some picture
Plain block possesses color notable feature, then and the sub-block closer to the block of pixels is that the probability of the notable block of color is bigger, otherwise more
The remote block of pixels is that the probability of notable block is lower.
Therefore, can be with the heterochromia degree where each pixel of summary between block of pixels and surrounding pixel block, and be somebody's turn to do
Two characteristic factors of the distance between block of pixels and surrounding pixel block where pixel, define a kind of otherness measure value of pixel
Function is calculated, the function can be proportionate with the heterochromia degree between block of pixels and surrounding pixel block, with picture where the pixel
The distance between plain block and surrounding pixel block is negatively correlated, so as to reflect the color between the pixel and other block of pixels with
The diversity factor of distance, if the otherness measure value of certain pixel is bigger, the color and distance feature of the pixel are more notable.
In one embodiment, block of pixels where each pixel and surrounding pixel block in the current frame image according to acquisition
Between heterochromia degree, and the distance between block of pixels and surrounding pixel block where the pixel calculate the color of each pixel
Include with the step of otherness measure value of distance：
Current frame image is split according to default pixel block size, obtains each pixel in current frame image and is somebody's turn to do
Block of pixels corresponding to pixel；
If computing is carried out to all pixels point, it will ample resources is expended, can be by two field picture in order to lift operation efficiency
The block of pixels of default size is divided into, computing is carried out in units of block of pixels, the numerical value of each pixel can be by calculating picture
Block of pixels where vegetarian refreshments determines.Specifically, the segmentation size of block of pixels can be set according to being actually needed, such as can be
4 × 4,8 × 8 or 16 × 16 block of pixels, the result of the smaller calculating of block of pixels is more accurate, expends demand to calculation resources accordingly
It is bigger.
In view of single block of pixels, it is assumed that i represents certain pixel, and pi represents the block of pixels where pixel i.Such as fruit block pi
Various features be different from other blocks, then pixel i can be considered as to possess notable feature.Color notable feature value can by pi and
Tried to achieve after pj vectorizations in Lab hue coordinates space using Euclidean distance, wherein Lab hue coordinates space can be described as again
Lab color model, the color model are International Commission on Illumination (Commission InternationaledeL '
Eclairage, abbreviation CIE) tissue determine one include the visible institute's the colorful one color mode of human eye in theory.
Block of pixels calculates the color notable feature value of each pixel according to corresponding to each pixel in Lab hue coordinates space,
Wherein, the calculation formula of the color notable feature value of pixel is as follows：
Lc(pi, pj)=| | pi-pj | |2
In formula, i is pixel, and the block of pixels of setting ranges of the pi where pixel i, j is any pixel in two field picture,
The block of pixels of setting ranges of the pj where j, Lc(pi, pj) is pixel i color notable feature value；
Obtain each pixel apart from notable feature value, wherein, the calculation formula apart from notable feature value of pixel is as follows：
Lp(pi, pj)=| | pi-pj | |2
In formula, Lp(pi, pj) is pixel i apart from notable feature value；
Obtain according to the color notable feature value of each pixel and apart from notable feature value the color of each pixel and the difference of distance
Different in nature measure value, wherein, the calculation formula of the color of pixel and the otherness measure value of distance is as follows：
In formula, c is constant, can value be otherness measure value that 3, d (pi, pj) is pixel i.
When pixel i and other blocks in image pixel difference are larger, i.e. d (pi, pj) is sufficiently large, then pixel i is considered as
Possesses the pixel of notable feature.
S102, otherness is less than using the otherness measure value and estimates block of pixels where the pixel of threshold value as reference pixels
Block, the reference otherness measure value of block of pixels where calculating each pixel respectively, obtained respectively with reference to otherness measure value according to described
The notable feature value of pixel.
The otherness measure value of block of pixels is bigger, then the block of pixels is that the probability of target pixel block is higher, conversely, block of pixels
Otherness measure value it is smaller, then the block of pixels is that the probability of blocks of background pixels is higher；Therefore, the color of foregoing operation can be chosen
The otherness that color and range difference opposite sex measure value is less than setting estimates block of pixels where the pixel of threshold value as background i.e. reference
Block of pixels, each block of pixels and possible blocks of background pixels are subjected to contrast computing, the reference otherness for obtaining each pixel is surveyed
Angle value, higher with reference to otherness measure value, then the difference of the pixel and background is bigger, and the pixel is the target for having notable feature
The probability of pixel is higher.
In one embodiment, it is described with the otherness measure value be less than otherness estimate it is more where the pixel of threshold value
Individual block of pixels is reference pixels block, multiple reference otherness measure values of block of pixels where calculating each pixel respectively, according to described
The step of obtaining the notable feature value of each pixel with reference to otherness measure value includes：
Obtain otherness measure value and estimate multiple block of pixels corresponding to the pixel of threshold value not have less than the otherness set
The block of pixels of standby notable feature；Wherein, the pixel set of blocks for not possessing notable feature is represented byN is not possess
The total number of the block of pixels of notable feature；
Using the block of pixels for not possessing notable feature as reference, multiple reference otherness measure values of certain pixel are drawn,
The calculation formula of any one reference otherness measure value is as follows：
In formula, qk is k-th of block of pixels for not possessing notable feature, and d (pi, qk) is that pixel i is poor relative to qk reference
Different in nature measure value；
According to the notable feature value that certain pixel color and distance are calculated with reference to otherness measure value, the notable feature
Value is drawn by following formula：
In formula, SiFor pixel i notable feature value, N is the total quantity for the block of pixels for not possessing notable feature
The purpose of above formula exponent arithmetic is so that SiValue be normalized to [0,1] scope.If the S of certain pixeliValue is bigger,
Illustrate that the pixel more has notable feature relative to other pixels, the pixel is that the probability of object pixel is higher, thus can be carried
Take out the block of pixels for existing and there is notable feature on color and positional distance.
S103, the pixel that notable feature value in current frame image is more than to notable feature threshold value are set to notable feature pixel,
The color of the current frame image is obtained and apart from notable hum pattern according to the notable feature pixel and its notable feature value.
The notable feature value of pixel is bigger, and the pixel is that the probability of object pixel is higher, therefore can set one significantly
Characteristic threshold value filters out notable feature value and is characterized pixel higher than the pixel of the threshold value, wherein, the threshold value can be by offline
Training obtains an Objective extraction accuracy rate highest threshold value.Draw point of the notable feature value of character pixel in the two field picture
Butut, be the two field picture color and apart from notable hum pattern.
S104, determine to transport using the color of the continuous multiple frames image comprising the current frame image and apart from notable hum pattern
Dynamic characteristic image, and the moving target in the current frame image is identified according to the motion feature image.
Video sequence figure does not only exist color and the notable feature of positional distance, and there is also the notable feature of motion.With inspection
Survey for the purpose of the moving target of notable feature, reject static pixel, can be to the color of continuous multiple frames image and apart from notable
Hum pattern carries out the calculus of differences of consecutive frame, then will transport counted result and carry out logical AND, and current frame image pair can be calculated
The motion feature image answered.Dynamic background is rejected by the segmentation threshold of setting again, you can obtain the target in current frame image.
In one embodiment, it is described using the color of the continuous multiple frames image comprising current frame image and apart from notable letter
Breath figure determines motion feature image corresponding to current frame image, and is determined according to the motion feature image in current frame image
The step of moving target, includes：
Calculate color and the first difference result figure apart from notable hum pattern of current frame image and previous frame image；
Calculate the color of current frame image and latter two field picture and the second difference result figure apart from notable hum pattern；
Logical AND is carried out to the first difference result figure and the second difference result figure and obtains motion feature image；
Motion feature image split according to the segmentation threshold of setting to obtain bianry image, wherein the bianry image
In the value of each pixel drawn by following formula：
In above formula, Dm.i,tFor the first difference result figure, Dm.i,t+1For the second difference result figure, T is the segmentation threshold of setting,
BSm.i,tFor the value of each pixel in bianry image；
Work as BSm.i,tValue when being 1, represent prospect, i.e. target works as BSm.i,tValue be 0 when, be expressed as background pixel.T generations
Table segmentation threshold, can value be 20.It is possible thereby to reject background element, target object is obtained.
The pixel that bianry image intermediate value is 0 is rejected as background pixel, using the pixel that bianry image intermediate value is 1 as mesh
Mark pixel to retain, obtain the target in current frame image.
In one embodiment, it is described using the color of the continuous multiple frames image comprising current frame image and apart from notable letter
Before the step of breath figure determines motion feature image corresponding to current frame image, in addition to step：To including current frame image
The color of continuous multiple frames image and apart from notable hum pattern carry out medium filtering noise reduction process.
The video frame images that carrier is shot in the process of moving, may be due to the equipment precision deficiency or equipment of shooting
Some noises be present in reason, the videos of shooting such as shake.Medium filtering can be effectively retained local edge during noise reduction,
Target pixel region edge feature can be retained during noise reduction is carried out to video frame images, therefore calculate video sequence fortune
Medium filtering noise reduction process first can be carried out to color and apart from notable hum pattern before dynamic notable feature, reject the noise in figure
Influence to recognition result, the accuracy of lifter motion target identification.
Above-mentioned motion estimate method, combines the color of target and the difference of background, block of pixels distance are more near similar
Many features such as probability is bigger and target is moved, color, block of pixels distance and motion are carried out more to video frame images
Characteristic synthetic computing is extracted, and is rejected the background of video frame images, is obtained target, the identification of the motion estimate method to target
Precise and high efficiency, target can be accurately extracted from video frame images.
Referring to Fig. 2, Fig. 2 is motion estimate apparatus structure schematic diagram, and the motion estimate device includes：
First computing module, for block of pixels and surrounding pixel block where each pixel in the current frame image according to acquisition it
Between heterochromia degree, and the pixel where the distance between block of pixels and surrounding pixel block, calculate each pixel color and
The otherness measure value of distance；
Second computing module, for being less than the pixel where otherness estimates the pixel of threshold value with the otherness measure value
Block is reference pixels block, the reference otherness measure value of block of pixels where calculating each pixel respectively, according to described with reference to otherness
Measure value obtains the notable feature value of each pixel；
Hum pattern acquisition module, the pixel of the notable feature value for notable feature value in current frame image to be more than to setting
Be set to notable feature pixel, according to the notable feature pixel and its notable feature value obtain the current frame image color and away from
From notable hum pattern；
Target Acquisition module, for the color using the continuous multiple frames image comprising the current frame image and apart from notable
Hum pattern determines motion feature image, and identifies the moving target in the current frame image according to the motion feature image.
Above-mentioned motion estimate device, combines the color of target and the difference of background, block of pixels distance are more near similar
Many features such as probability is bigger and target is moved, color, block of pixels distance and motion are carried out more to video frame images
Characteristic synthetic computing is extracted, and is rejected the background of video frame images, is obtained target, the identification of the motion estimate method to target
Precise and high efficiency, target can be accurately extracted from video frame images.
The motion estimate device of the present invention and the motion estimate method of the present invention correspond, in above-mentioned motion
The technical characteristic and its advantage that the embodiment of target identification method illustrates are applied to the implementation of motion estimate device
In example, hereby give notice that.
The present invention also provides a kind of moving target method for early warning, and shown in Figure 3, Fig. 3 is moving target method for early warning stream
Cheng Tu, the moving target method for early warning comprise the following steps：
S301, obtain the two field picture shot to the environmental information of direction initialization around carrier.
In above-mentioned steps, the carrier can be to possess any carrier for performing the moving target method for early warning function,
Such as can be the vehicles such as automobile；The direction of setting can be any direction around carrier, such as automobile,
Can shoot vehicle front and/or the video information at rear, obtain the video frame images of shooting.
S302, identify all targets in current frame image, according to each pixel in default two field picture with it is actual away from
From corresponding relation obtain with the closest target of carrier, and obtain the distance between the nearest target and carrier information；
Wherein, using the target in the motion estimate method identification two field picture described in as above any one.
In the step, any one of foregoing motion estimate method can be used to the motion in current video two field picture
Target is identified, and can be carried out calculating afterwards according to the corresponding relation of each pixel and actual range in default two field picture and divided
Analysis, draws the distance between each target and carrier, selects the target closest wherein with carrier as tracking target.
Wherein, the corresponding relation of each pixel and actual range can be obtained by off-line training in the two field picture,
In one embodiment, before the step of all targets in the identification current frame image, in addition to step：By instructing offline
Practice the corresponding relation for obtaining each pixel and actual range in two field picture.Pass through the corresponding relation, you can obtain by computing
The actual range of each target and carrier in video frame images.
In addition, in order to by estimating that the distance between target latter period and carrier judge target whether in safety
Within distance, it is also necessary to establish speed-acceleration-Safety distance model of carrier, in one embodiment, the identification is current
Before the step of all targets in two field picture, in addition to step：Establish speed-acceleration-Safety distance model of carrier.
S303, according to the time interval of the setting of adjacent two field pictures and obtain carrier present speed and acceleration, with
And the distance between the target and carrier prejudge the relative position information of next setting time section target and carrier.
In the two field pictures obtained according to the time interval of adjacent two field pictures and abovementioned steps between target and carrier
Distance, and the current speed of carrier and acceleration obtained, can be calculated the relative displacement of target object and this car in two frames,
And then the average speed of target object can be estimated, estimate relative position of the target next with carrier such as this automobile by calculating
Confidence ceases.
S304, according to default speed-acceleration-Safety distance model and next setting time section target of anticipation with carrying
The relative position information of body carries out early warning.
, can be according to default speed-acceleration-Safety distance model, and next setting of anticipation in above-mentioned steps
The relative position of period target and carrier, judges whether subsequent time period target is in safe distance, and according to target
Motion state sends early warning, for example, can when anticipation may exceed safe distance to the distance between target and carrier automobile,
Send early warning.In one embodiment, it is described to be set according to default speed-acceleration-Safety distance model and the next of anticipation
The step of relative position information of section target of fixing time and carrier progress early warning, includes：
According to the next setting time section target and carrier of default speed-acceleration-Safety distance model and anticipation
Relative position information judges whether potential safety hazard；
When potential safety hazard be present, potential safety hazard early warning is sent and/or according to default by indicator lamp or buzzer
Braking mode to motion carrier carry out retarding braking.
Wherein, the potential safety hazard can be target object possible collision between carrier such as bicycle, pedestrian
Hidden danger, then can be the rear-end collision hidden danger between vehicle for automobile.
Above-mentioned moving target method for early warning, with above-mentioned motion estimate method identify shooting two field picture in carrier
Closest target, early warning is carried out by analyzing the position of next setting time section target of anticipation.Due to carrying out target
Combine that the color of target and the difference of background, the nearlyer likelihood probability of block of pixels distance be bigger during identification and target motion etc. is more
The feature of aspect, video frame images are carried out with color, block of pixels distance and the combination features computing of motion and is extracted, rejects video
The background of two field picture, accurately efficiently it can extract target from the video frame images of shooting.So as to current
The motion state of carrier surrounding objects object makes more accurately anticipation and early warning, lifted to the accuracy of moving target early warning and
Reliability.
In addition, also providing a kind of automobile anti-rear end collision prior-warning device, described device is configured to hold when performing and preventing knocking into the back early warning
The capable as above moving target method for early warning described in any one.
Above-mentioned automobile anti-rear end collision prior-warning device, shoots to the environmental information of motor vehicle environment direction initialization, accurate and high
Effect ground extracts target from the video frame images of shooting, the motion state of current motor vehicle environment target object is made precisely and
When prejudge, carry out early warning when that may knock into the back, the accuracy and reliability of lifting automobile anti-rear end collision early warning.
In one embodiment, the automobile anti-rear end collision prior-warning device can include automobile parameter reading device, video is adopted
Collection processing unit, anti-knock into the back early warning device, automobile control process device and processor, shown in Figure 4, automobile ginseng
Number reading device, video acquisition processing unit, anti-early warning device, the automobile control process device of knocking into the back connect with processor respectively
Connect, processor is configured to perform the as above moving target method for early warning described in any one.
Wherein, automobile parameter reading device can be used for obtaining vehicle condition parameter, including speed, acceleration etc..Due to vapour
The controller of the parameter such as the existing measuring speed in in-car portion and acceleration, therefore need to carry can be with vehicle communication bus phase for the present apparatus
The communication protocol and interface of adaptation go to obtain the parameter of automobile.
Wherein, video acquisition processing unit can be used for obtaining vehicle front and/or the video information at rear, and to collection
To video information handled, for example with any one it is foregoing moving target analysis method extraction video frame images in mesh
Mark.The video information of video acquisition processing unit collection can come from the drive recorder that automobile carries or in addition
The camera of installation, the sensor of carrying can be reduced from drive recorder collection video information, reduces cost.
Wherein, the anti-early warning device that knocks into the back can include early warning indicator lamp and/or buzzer.After video information process
When being produced if alarm command, indicator lamp will send lamp bright signal and/or buzzer and ring is sent into alarm signal, and prompting is driven
The person of sailing pays attention to front vehicles.
Wherein, automobile control process device is used to carry out automobile the operation such as retarding braking.When alarm signal produces, drive
The person of sailing can be set directly is controlled processing by control device to automobile, can also set and automobile is carried out by driver personnel
Control process.
Below with a specific operation principle that above-mentioned automobile anti-rear end collision prior-warning device is illustrated using example, adopted with automobile
Exemplified by the video information for collecting front, the automobile anti-rear end collision prior-warning device can include such as when the anti-early warning analysis that knocks into the back is handled
Lower step：
Video acquisition processing unit is shot to the video information of vehicle front, is identified from the video frame images of acquisition
Moving target, if identifying multiple moving targets, according to the pixel and actual range obtained beforehand through off-line training
Corresponding relation, the actual range between each moving target and automobile is calculated, target wherein nearest apart from automobile is made
To track target, analyze whether target nearest apart from automobile in adjacent two frames picture is same object, if then continuing to this
Target is tracked., then, will be with using target nearest in a time later two field picture as tracking target if not same object
The distance between track target and automobile information are sent to processor in real time.
Processor is according to the distance between the tracking target of reception and automobile information, with reference between default two field pictures
Time interval, and the speed and acceleration of this car obtained in real time from automobile parameter reading device, can obtain object in two frames
The relative displacement of body and this car, and then the average speed of target object can be estimated, by calculate estimate the target next with this
The relative position information of automobile, judge whether current state continues according to default speed-acceleration-Safety distance model
In the presence of the safety problem such as knock into the back, if it is determined that there is a possibility that to knock into the back, then warning information is sent to the anti-early warning that knocks into the back
Device and automobile control process device, the anti-early warning device that knocks into the back send flashing light or buzzing early warning driver operation, or
Person's automobile control process device is slowed down or braking processing, the risk that knocks into the back is eliminated, so as to complete to vehicle front moving target
Anti- early warning of knocking into the back.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction executing device, device or equipment (such as computer based device including the device of processor or other can be held from instruction
Luggage is put, the device of device or equipment instruction fetch and execute instruction) use, or combine these instruction executing devices, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium " can any can be included, store, communicate, propagate or pass
Defeated program is for instruction executing device, device or equipment or the dress used with reference to these instruction executing devices, device or equipment
Put.
The more specifically example (non-exhaustive list) of computer-readable medium includes following：Connected up with one or more
Electrical connection section (electronic installation), portable computer diskette box (magnetic device), random access memory (RAM), read-only storage
(ROM), erasable edit read-only storage (EPROM or flash memory), fiber device, and portable optic disk is read-only deposits
Reservoir (CDROM).In addition, computer-readable medium, which can even is that, to print the paper of described program thereon or other are suitable
Medium, because can then enter edlin, interpretation or if necessary with it for example by carrying out optical scanner to paper or other media
His suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.Above-mentioned
In embodiment, software that multiple steps or method can be performed in memory and by suitable instruction executing device with storage
Or firmware is realized.If, and in another embodiment, can be with well known in the art for example, realized with hardware
Any one of row technology or their combination are realized：With the logic gates for realizing logic function to data-signal
Discrete logic, have suitable combinational logic gate circuit application specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or the spy for combining the embodiment or example description
Point is contained at least one embodiment or example of the present invention.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.Moreover, specific features, structure, material or the feature of description can be any
One or more embodiments or example in combine in an appropriate manner.
Each technical characteristic of embodiment described above can be combined arbitrarily, to make description succinct, not to above-mentioned reality
Apply all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, the scope that this specification is recorded all is considered to be.
Embodiment described above only expresses the several embodiments of the present invention, and its description is more specific and detailed, but simultaneously
Can not therefore it be construed as limiting the scope of the patent.It should be pointed out that come for one of ordinary skill in the art
Say, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (10)
- A kind of 1. motion estimate method, it is characterised in that comprise the following steps：According to the heterochromia degree where each pixel in the current frame image of acquisition between block of pixels and surrounding pixel block, and should The distance between block of pixels and surrounding pixel block where pixel, calculate the color of each pixel and the otherness measure value of distance；Otherness is less than using the otherness measure value and estimates block of pixels where the pixel of threshold value as reference pixels block, is counted respectively The reference otherness measure value of block of pixels where calculating each pixel, the notable of each pixel is obtained with reference to otherness measure value according to described Characteristic value；The pixel that notable feature value in current frame image is more than to notable feature threshold value is set to notable feature pixel, according to described aobvious Write character pixel and its notable feature value obtains the color of the current frame image and apart from notable hum pattern；Motion feature figure is determined using the color of the continuous multiple frames image comprising the current frame image and apart from notable hum pattern Picture, and the moving target in the current frame image is identified according to the motion feature image.
- 2. motion estimate method according to claim 1, it is characterised in that described utilize includes current frame image The color of continuous multiple frames image and apart from notable hum pattern determine motion feature image corresponding to current frame image the step of before, Also include step：Color to the continuous multiple frames image comprising current frame image and carry out medium filtering noise reduction process apart from notable hum pattern.
- 3. motion estimate method according to claim 1, it is characterised in that the current frame image according to acquisition In heterochromia degree where each pixel between block of pixels and surrounding pixel block, and block of pixels and surrounding pixel where the pixel The distance between block, calculate each pixel color and distance otherness measure value the step of include：Current frame image is split according to default pixel block size, obtains each pixel and pixel in current frame image Corresponding block of pixels；Block of pixels calculates the color notable feature value of each pixel according to corresponding to each pixel in Lab hue coordinates space, wherein, The calculation formula of the color notable feature value of pixel is as follows：Lc(pi, pj)=| | pi-pj | |2In formula, i is pixel, and the block of pixels of setting ranges of the pi where pixel i, j is any pixel in two field picture, and pj is The block of pixels of setting range where j, Lc(pi, pj) is pixel i color notable feature value；Obtain each pixel apart from notable feature value, wherein, the calculation formula apart from notable feature value of pixel is as follows：Lp(pi, pj)=| | pi-pj | |2In formula, Lp(pi, pj) is pixel i apart from notable feature value；Obtain according to the color notable feature value of each pixel and apart from notable feature value the color of each pixel and the otherness of distance Measure value, wherein, the calculation formula of the color of pixel and the otherness measure value of distance is as follows：<mrow> <mi>d</mi> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>p</mi> <mi>j</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <msub> <mi>L</mi> <mi>c</mi> </msub> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>p</mi> <mi>j</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mn>1</mn> <mo>+</mo> <mi>c</mi> <mo>&times;</mo> <msub> <mi>L</mi> <mi>p</mi> </msub> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>p</mi> <mi>j</mi> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>In formula, c is constant, and d (pi, pj) is pixel i otherness measure value.
- 4. motion estimate method according to claim 3, it is characterised in that described small with the otherness measure value Multiple block of pixels where otherness estimates the pixel of threshold value are reference pixels block, block of pixels where calculating each pixel respectively It is multiple with reference to otherness measure values, wrapped according to the step of notable feature value with reference to each pixel of otherness measure value acquisition Include：Otherness measure value is obtained to be less than the otherness set to estimate multiple block of pixels corresponding to the pixel of threshold value aobvious not possess Write the block of pixels of feature；Using the block of pixels for not possessing notable feature as reference, the multiple with reference to otherness measure value of certain pixel are drawn, it is described Any one is as follows with reference to the calculation formula of otherness measure value：<mrow> <mi>d</mi> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>q</mi> <mi>k</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <msub> <mi>L</mi> <mi>c</mi> </msub> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>q</mi> <mi>k</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mn>1</mn> <mo>+</mo> <mi>c</mi> <mo>&times;</mo> <msub> <mi>L</mi> <mi>p</mi> </msub> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>q</mi> <mi>k</mi> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>In formula, qk is k-th of block of pixels for not possessing notable feature, and d (pi, qk) is reference othernesses of the pixel i relative to qk Measure value；According to the notable feature value that certain pixel color and distance are calculated with reference to otherness measure value, the notable feature value by Following formula is drawn：<mrow> <msub> <mi>S</mi> <mi>i</mi> </msub> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mi>exp</mi> <mo>{</mo> <mo>-</mo> <mfrac> <mn>1</mn> <mi>N</mi> </mfrac> <munderover> <mo>&Sigma;</mo> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <mi>d</mi> <mrow> <mo>(</mo> <mi>p</mi> <mi>i</mi> <mo>,</mo> <mi>q</mi> <mi>k</mi> <mo>)</mo> </mrow> <mo>}</mo> </mrow>In formula, SiFor pixel i notable feature value, N is the total quantity for the block of pixels for not possessing notable feature.
- 5. motion estimate method according to claim 4, it is characterised in that described utilize includes current frame image The color of continuous multiple frames image and motion feature image corresponding to current frame image is determined apart from notable hum pattern, and according to described Motion feature image determines that the step of moving target in current frame image includes：Calculate color and the first difference result figure apart from notable hum pattern of current frame image and previous frame image；Calculate the color of current frame image and latter two field picture and the second difference result figure apart from notable hum pattern；Logical AND is carried out to the first difference result figure and the second difference result figure and obtains motion feature image；Motion feature image split according to the segmentation threshold of setting to obtain bianry image, wherein each in the bianry image The value of pixel is drawn by following formula：<mrow> <msub> <mi>BS</mi> <mrow> <mi>m</mi> <mo>.</mo> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>1</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>i</mi> <mi>f</mi> <mi> </mi> <msub> <mi>D</mi> <mrow> <mi>m</mi> <mo>.</mo> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> <mo>></mo> <mi>T</mi> <mo>&cap;</mo> <msub> <mi>D</mi> <mrow> <mi>m</mi> <mo>.</mo> <mi>i</mi> <mo>,</mo> <mi>t</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mo>></mo> <mi>T</mi> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <mi>O</mi> <mi>t</mi> <mi>h</mi> <mi>e</mi> <mi>r</mi> <mi>s</mi> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow>In above formula, Dm.i,tFor the first difference result figure, Dm.i,t+1For the second difference result figure, T is the segmentation threshold of setting, BSm.i,tFor the value of each pixel in bianry image；The pixel that bianry image intermediate value is 0 is rejected as background pixel, using the pixel that bianry image intermediate value is 1 as target picture Element retains, and obtains the target in current frame image.
- A kind of 6. motion estimate device, it is characterised in that including：First computing module, for where each pixel in the current frame image according to acquisition between block of pixels and surrounding pixel block The distance between block of pixels and surrounding pixel block where heterochromia degree, and the pixel, calculate the color and distance of each pixel Otherness measure value；Second computing module, for using the otherness measure value be less than otherness estimate block of pixels where the pixel of threshold value as Reference pixels block, the reference otherness measure value of block of pixels, estimates according to described with reference to otherness where calculating each pixel respectively Value obtains the notable feature value of each pixel；Hum pattern acquisition module, for the color using the continuous multiple frames image comprising the current frame image and apart from notable letter Breath figure determines motion feature image, and identifies the moving target in the current frame image according to the motion feature image；Target Acquisition module, for utilizing the color of the continuous multiple frames image comprising current frame image and true apart from notable hum pattern Motion feature image corresponding to settled prior image frame, and determine according to the motion feature image motion mesh in current frame image Mark.
- 7. a kind of moving target method for early warning, it is characterised in that the moving target method for early warning comprises the following steps：Obtain the two field picture shot to the environmental information of direction initialization around carrier；All targets in current frame image are identified, are closed according to each pixel in default two field picture is corresponding with actual range System obtains and the closest target of carrier, and obtains the distance between the nearest target and carrier information；Wherein, using such as The target in motion estimate method identification two field picture described in any one of claim 1 to 5；According to the time interval of the setting of adjacent two field pictures and the carrier present speed and acceleration that obtain, and the target The distance between carrier prejudges the relative position information of next setting time section target and carrier；According to the relative of the next setting time section target and carrier of default speed-acceleration-Safety distance model and anticipation Positional information carries out early warning.
- 8. moving target method for early warning according to claim 7, it is characterised in that the institute in the identification current frame image Before having the step of target, in addition to step：Pair of each pixel and actual range in two field picture is obtained by off-line training It should be related to, establish speed-acceleration-Safety distance model of carrier.
- 9. moving target method for early warning according to claim 7, it is characterised in that described according to default speed-acceleration The step of relative position information of the next setting time section target and carrier of degree-Safety distance model and anticipation carries out early warning bag Include：According to the relative of the next setting time section target and carrier of default speed-acceleration-Safety distance model and anticipation Positional information judges whether potential safety hazard；When potential safety hazard be present, potential safety hazard early warning is sent and/or according to default system by indicator lamp or buzzer Dynamic model formula carries out retarding braking to motion carrier.
- 10. a kind of automobile anti-rear end collision prior-warning device, it is characterised in that described device is configured to perform when performing and preventing knocking into the back early warning Moving target method for early warning as described in any one of claim 7 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711046921.1A CN107798688B (en) | 2017-10-31 | 2017-10-31 | Moving target identification method, early warning method and automobile rear-end collision prevention early warning device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711046921.1A CN107798688B (en) | 2017-10-31 | 2017-10-31 | Moving target identification method, early warning method and automobile rear-end collision prevention early warning device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107798688A true CN107798688A (en) | 2018-03-13 |
| CN107798688B CN107798688B (en) | 2020-07-28 |
Family
ID=61548530
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711046921.1A Active CN107798688B (en) | 2017-10-31 | 2017-10-31 | Moving target identification method, early warning method and automobile rear-end collision prevention early warning device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107798688B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109747644A (en) * | 2019-03-01 | 2019-05-14 | 广州杰赛科技股份有限公司 | Vehicle tracking anti-collision early warning method, device, controller, system and vehicle |
| CN110111364A (en) * | 2019-04-30 | 2019-08-09 | 腾讯科技（深圳）有限公司 | Method for testing motion, device, electronic equipment and storage medium |
| CN111914627A (en) * | 2020-06-18 | 2020-11-10 | 广州杰赛科技股份有限公司 | Vehicle identification and tracking method and device |
| CN112351191A (en) * | 2020-09-14 | 2021-02-09 | 中标慧安信息技术股份有限公司 | Mobile detection processing method and system |
| CN112577475A (en) * | 2021-01-14 | 2021-03-30 | 天津希格玛微电子技术有限公司 | Video ranging method capable of effectively reducing power consumption |
| WO2023191713A3 (en) * | 2022-03-29 | 2023-11-30 | 脸萌有限公司 | Object recognition method and apparatus, electronic device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120294476A1 (en) * | 2011-05-16 | 2012-11-22 | Microsoft Corporation | Salient Object Detection by Composition |
| US20120328161A1 (en) * | 2011-06-22 | 2012-12-27 | Palenychka Roman | Method and multi-scale attention system for spatiotemporal change determination and object detection |
| CN103106671A (en) * | 2013-01-25 | 2013-05-15 | 西北工业大学 | Method for detecting interested region of image based on visual attention mechanism |
| CN103824299A (en) * | 2014-03-11 | 2014-05-28 | 武汉大学 | Target tracking method based on significance |
| CN104933738A (en) * | 2015-06-16 | 2015-09-23 | 中国人民解放军国防科学技术大学 | Visual saliency map generation method based on local structure detection and contrast |
| CN107292917A (en) * | 2017-06-15 | 2017-10-24 | 武汉理工大学 | A kind of Dragging recognition methods based on Computer Vision |
-
2017
- 2017-10-31 CN CN201711046921.1A patent/CN107798688B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120294476A1 (en) * | 2011-05-16 | 2012-11-22 | Microsoft Corporation | Salient Object Detection by Composition |
| US20120328161A1 (en) * | 2011-06-22 | 2012-12-27 | Palenychka Roman | Method and multi-scale attention system for spatiotemporal change determination and object detection |
| CN103106671A (en) * | 2013-01-25 | 2013-05-15 | 西北工业大学 | Method for detecting interested region of image based on visual attention mechanism |
| CN103824299A (en) * | 2014-03-11 | 2014-05-28 | 武汉大学 | Target tracking method based on significance |
| CN104933738A (en) * | 2015-06-16 | 2015-09-23 | 中国人民解放军国防科学技术大学 | Visual saliency map generation method based on local structure detection and contrast |
| CN107292917A (en) * | 2017-06-15 | 2017-10-24 | 武汉理工大学 | A kind of Dragging recognition methods based on Computer Vision |
Non-Patent Citations (2)
| Title |
|---|
| LIU NINGBO等: "Target Detection Within Sea Clutter Based on Combined Fractal Characteristics", 《IEEE》 * |
| 成培瑞等: "基于多尺度区域对比的显著目标识别", 《中国光学》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109747644A (en) * | 2019-03-01 | 2019-05-14 | 广州杰赛科技股份有限公司 | Vehicle tracking anti-collision early warning method, device, controller, system and vehicle |
| CN110111364A (en) * | 2019-04-30 | 2019-08-09 | 腾讯科技（深圳）有限公司 | Method for testing motion, device, electronic equipment and storage medium |
| CN110111364B (en) * | 2019-04-30 | 2022-12-27 | 腾讯科技（深圳）有限公司 | Motion detection method and device, electronic equipment and storage medium |
| CN111914627A (en) * | 2020-06-18 | 2020-11-10 | 广州杰赛科技股份有限公司 | Vehicle identification and tracking method and device |
| CN112351191A (en) * | 2020-09-14 | 2021-02-09 | 中标慧安信息技术股份有限公司 | Mobile detection processing method and system |
| CN112351191B (en) * | 2020-09-14 | 2021-11-23 | 中标慧安信息技术股份有限公司 | Mobile detection processing method and system |
| CN112577475A (en) * | 2021-01-14 | 2021-03-30 | 天津希格玛微电子技术有限公司 | Video ranging method capable of effectively reducing power consumption |
| WO2023191713A3 (en) * | 2022-03-29 | 2023-11-30 | 脸萌有限公司 | Object recognition method and apparatus, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107798688B (en) | 2020-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107798688A (en) | Motion estimate method, method for early warning and automobile anti-rear end collision prior-warning device | |
| JP5297078B2 (en) | Method for detecting moving object in blind spot of vehicle, and blind spot detection device | |
| CN110659543B (en) | Gesture recognition-based vehicle control method and system and vehicle | |
| CN106611512B (en) | Method, device and system for processing starting of front vehicle | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN109635656A (en) | Vehicle attribute recognition methods, device, equipment and medium neural network based | |
| CN106485233A (en) | Drivable region detection method, device and electronic equipment | |
| CN106541968B (en) | The recognition methods of the subway carriage real-time prompt system of view-based access control model analysis | |
| CN106575473A (en) | Method and axle-counting device for contact-free axle counting of a vehicle and axle-counting system for road traffic | |
| CN105427342B (en) | A kind of underwater Small object sonar image target detection tracking method and system | |
| CN101739686A (en) | Moving object tracking method and system thereof | |
| WO2010006361A1 (en) | Detection of vehicles in images of a night time scene | |
| CN105654516B (en) | Satellite image based on target conspicuousness is to ground weak moving target detection method | |
| CN105426894B (en) | Railway plug pin image detecting method and device | |
| CN107274678B (en) | A kind of night vehicle flowrate and model recognizing method based on Kinect | |
| CN107016348A (en) | With reference to the method for detecting human face of depth information, detection means and electronic installation | |
| CN111967396A (en) | Processing method, device and equipment for obstacle detection and storage medium | |
| CN112613568B (en) | Target identification method and device based on visible light and infrared multispectral image sequence | |
| CN106803262A (en) | The method that car speed is independently resolved using binocular vision | |
| CN116978009A (en) | Dynamic object filtering method based on 4D millimeter wave radar | |
| CN107506753B (en) | Multi-vehicle tracking method for dynamic video monitoring | |
| CN110287957B (en) | Low-slow small target positioning method and positioning device | |
| EP4089649A1 (en) | Neuromorphic cameras for aircraft | |
| CN109740558B (en) | Moving target detection method based on improved optical flow method | |
| WO2023071874A1 (en) | Roadside assistance working node determining method and apparatus, electronic device, and storage medium |
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 |