CN106777350A - It is a kind of based on bayonet socket data scheming to search drawing method and device - Google Patents
It is a kind of based on bayonet socket data scheming to search drawing method and device Download PDFInfo
- Publication number
- CN106777350A CN106777350A CN201710033558.3A CN201710033558A CN106777350A CN 106777350 A CN106777350 A CN 106777350A CN 201710033558 A CN201710033558 A CN 201710033558A CN 106777350 A CN106777350 A CN 106777350A
- Authority
- CN
- China
- Prior art keywords
- image
- characteristic point
- target vehicle
- word
- vehicle image
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Library & Information Science (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the present invention provide it is a kind of based on bayonet socket data scheming to search drawing method and device, including:The vehicle body region in target vehicle image is obtained, the characteristic information of target vehicle image is then extracted from vehicle body region, characteristic information includes the characteristic point of target vehicle image and the yardstick of characteristic point, principal direction and relative position.The yardstick of characteristic point of the characteristic point and sample image of the sample image for then being stored in the characteristic information query characteristics database according to target vehicle image, principal direction and relative position, it is determined that the image similar to target vehicle image.In the embodiment of the present invention, multiple characteristic points of target vehicle image and the relevant information of characteristic point according to extracting find out similar image from bayonet socket database, therefore the intrinsic information such as color, vehicle, brand in vehicle is when duplicating or lacking, still similar image can be found out from bayonet socket database, improve the precision that similar vehicle image is searched from bayonet socket database.
Description
Technical field
The present embodiments relate to intelligent transportation field, more particularly to it is a kind of based on bayonet socket data with scheme to search drawing method and
Device.
Background technology
With the fast-developing of city and progress, urban safety also increasingly attracts attention.Carrying out video investigation, deck
, it is necessary to the picture according to target vehicle finds out the picture of same vehicle from bayonet socket database when car is searched, suspicion car is searched for.Pass
The bayonet vehicle searching system of system can only carry out partial condition according to license board information, vehicle color, vehicle, brand age money information
Inquiry or combination condition query.And existing vehicle fleet size is more, vehicle color, vehicle, brand age money information are duplicated
Probability is big, and target vehicle cannot be quickly and accurately searched when license board information is lacked, and causes according to target vehicle image from card
The precision that similar vehicle image is searched in mouth database is low.
The content of the invention
The embodiment of the present invention provide it is a kind of based on bayonet socket data to scheme to search drawing method and device, for solving traditional bayonet socket
Vehicle retrieval system searches the low problem of the similar vehicle precision of images according to vehicle target vehicle image from bayonet socket database.
The embodiment of the invention provides it is a kind of based on bayonet socket data to scheme to search drawing method, including:
Obtain the vehicle body region in target vehicle image;
The characteristic information of the target vehicle image, the spy are extracted from the vehicle body region in the target vehicle image
Reference breath include the yardstick of the characteristic point of the characteristic point of the target vehicle image and the target vehicle image, principal direction and
Relative position;
Yardstick, the principal direction of the characteristic point of characteristic point and the target vehicle image according to the target vehicle image
With the chi of the characteristic point of the characteristic point and the sample image of the sample image stored in relative position query characteristics database
Degree, principal direction and relative position, it is determined that the image similar to the target vehicle image, the property data base is according to bayonet socket
What the image in database determined.
Alternatively, before the vehicle body region obtained in target vehicle image, also include:
In the image of the setting quantity in the bayonet socket database being obtained as training image and determining the training image
Vehicle body region;
The characteristic point of the training image is extracted from the vehicle body region in the training image;
Characteristic point to the training image is clustered, and every category feature point is defined as in visual dictionary
Word, one word numbering of each word correspondence in the visual dictionary;
The word and word numbering are preserved into the visual dictionary.
Alternatively, the image in the bayonet socket database determines property data base, including:
The all images in the bayonet socket database are obtained as sample image and the vehicle body area of the sample image is determined
Domain;
The characteristic information of the sample image, the characteristic information bag are extracted from the vehicle body region in the sample image
Include yardstick, principal direction and the relative position of the characteristic point of the sample image and the characteristic point of the sample image;
The word that the characteristic point of the sample image is mapped in the visual dictionary and the spy for determining the sample image
Levy word numbering a little;
By the yardstick of the word of the characteristic point of sample image numbering and the characteristic point of the sample image, principal direction and
Relative position is preserved into the property data base.
Alternatively, the characteristic point of the characteristic point according to the target vehicle image and the target vehicle image
Yardstick, principal direction and relative position determine the image similar to the target vehicle image from property data base, including:
Word that the characteristic point of the target vehicle image is mapped in the visual dictionary simultaneously determines the target vehicle
The word numbering of the characteristic point of image;
The word numbering of the characteristic point according to the target vehicle image is filtered out and the mesh from the property data base
Mark the word numbering identical characteristic point of the characteristic point of vehicle image;
Word numbering identical characteristic point in the property data base with the characteristic point of the target vehicle image is corresponding
Image image as a comparison;
The vision word numbering of vision word numbering and the contrast images according to the target vehicle image determines institute
The matching characteristic point of target vehicle image and the contrast images is stated, the vision word numbering is according to word numbering, characteristic point
Yardstick, principal direction and relative position determine;
Characteristic point, the characteristic point of the contrast images and the target vehicle image according to the target vehicle image
With the similarity that the matching characteristic of contrast images point determines the target vehicle image and the contrast images;
The contrast images that the similarity meets given threshold are defined as the image similar to the target vehicle image.
Alternatively, it is described that vision word numbering symbol is determined according to word numbering, the yardstick of characteristic point, principal direction and relative position
Close following formula (1):
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein, M is vision word numbering, WidIt is the word numbering in visual dictionary, Tw is the sum of word in visual dictionary,
PidIt is the numbering of relative position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidFor
The numbering of yardstick.
The characteristic point according to the target vehicle image, the characteristic point of the contrast images and the target vehicle
The matching characteristic point of image and the contrast images determines that the target vehicle image and the similarity of the contrast images meet
Following formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and right
Than the number of the matching characteristic point of image, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the spy of contrast images
Levy number a little.
Correspondingly, the embodiment of the present invention additionally provide it is a kind of based on bayonet socket data to scheme to search map device, including:
Acquisition module, for obtaining the vehicle body region in target vehicle image;
Characteristic extracting module, for extracting the target vehicle image from the vehicle body region in the target vehicle image
Characteristic information, the characteristic point of the characteristic information including the target vehicle image and the feature of the target vehicle image
Yardstick, principal direction and the relative position put;
Processing module, for the characteristic point according to the target vehicle image and the characteristic point of the target vehicle image
Yardstick, principal direction and relative position query characteristics database in the characteristic point of sample image that stores and the sample image
The yardstick of characteristic point, principal direction and relative position, it is determined that the image similar to the target vehicle image, the characteristic
Storehouse is that the image in bayonet socket database determines.
Alternatively, the acquisition module is additionally operable to:
In the image of the setting quantity in the bayonet socket database being obtained as training image and determining the training image
Vehicle body region;
The characteristic point of the training image is extracted from the vehicle body region in the training image;
Characteristic point to the training image is clustered, and every category feature point is defined as in visual dictionary
Word, one word numbering of each word correspondence in the visual dictionary;
The word and word numbering are preserved into the visual dictionary.
Alternatively, the processing module specifically for:
The all images in the bayonet socket database are obtained as sample image and the vehicle body area of the sample image is determined
Domain;
The characteristic information of the sample image, the characteristic information bag are extracted from the vehicle body region in the sample image
Include yardstick, principal direction and the relative position of the characteristic point of the sample image and the characteristic point of the sample image;
The word that the characteristic point of the sample image is mapped in the visual dictionary and the spy for determining the sample image
Levy word numbering a little;
By the yardstick of the word of the characteristic point of sample image numbering and the characteristic point of the sample image, principal direction and
Relative position is preserved into the property data base.
Alternatively, the processing module specifically for:
Word that the characteristic point of the target vehicle image is mapped in the visual dictionary simultaneously determines the target vehicle
The word numbering of the characteristic point of image;
The word numbering of the characteristic point according to the target vehicle image is filtered out and the mesh from the property data base
Mark the word numbering identical characteristic point of the characteristic point of vehicle image;
Word numbering identical characteristic point in the property data base with the characteristic point of the target vehicle image is corresponding
Image image as a comparison;
The vision word numbering of vision word numbering and the contrast images according to the target vehicle image determines institute
The matching characteristic point of target vehicle image and the contrast images is stated, the vision word numbering is according to word numbering, characteristic point
Yardstick, principal direction and relative position determine;
Characteristic point, the characteristic point of the contrast images and the target vehicle image according to the target vehicle image
With the similarity that the matching characteristic of contrast images point determines the target vehicle image and the contrast images;
The contrast images that the similarity meets given threshold are defined as the image similar to the target vehicle image.
Alternatively, the processing module specifically for:
It is described to determine that vision word numbering meets following according to word numbering, the yardstick of characteristic point, principal direction and relative position
Formula (1):
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein, M is vision word numbering, WidIt is the word numbering in visual dictionary, Tw is the sum of word in visual dictionary,
PidIt is the numbering of relative position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidFor
The numbering of yardstick.
The characteristic point according to the target vehicle image, the characteristic point of the contrast images and the target vehicle
The matching characteristic point of image and the contrast images determines that the target vehicle image and the similarity of the contrast images meet
Following formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and right
Than the number of the matching characteristic point of image, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the spy of contrast images
Levy number a little.
The embodiment of the present invention shows, the vehicle body region in target vehicle image is obtained, then from the target vehicle image
In vehicle body region in extract the characteristic information of the target vehicle image, the characteristic information includes the target vehicle image
Characteristic point and the yardstick of characteristic point of the target vehicle image, principal direction and relative position.Then according to the target
The yardstick of the characteristic point of the characteristic point of vehicle image and the target vehicle image, principal direction and relative position query characteristics number
According to the yardstick of the characteristic point of the characteristic point and the sample image of the sample image stored in storehouse, principal direction and relative position,
It is determined that the image similar to the target vehicle image, the property data base is that the image in bayonet socket database determines
's.In the embodiment of the present invention, by extracting multiple characteristic points of target vehicle image and the relevant information of characteristic point, and according to
The characteristic point and characteristic point relevant information of extraction find out similar image from bayonet socket database, rather than by vehicle in itself
The intrinsic informations such as license board information, vehicle color, vehicle, brand, therefore there is weight in the license board information, color, vehicle, brand in vehicle
When multiple or license board information is lacked, still similar image can be found out from bayonet socket database, improve and searched from bayonet socket database
The precision of similar vehicle image.
Brief description of the drawings
Technical scheme in order to illustrate more clearly the embodiments of the present invention, below will be to that will make needed for embodiment description
Accompanying drawing is briefly introduced, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for this
For the those of ordinary skill in field, without having to pay creative labor, it can also be obtained according to these accompanying drawings
His accompanying drawing.
Fig. 1 is a kind of schematic flow sheet to scheme to search drawing method based on bayonet socket data provided in an embodiment of the present invention;
Fig. 2 is a kind of schematic diagram of the principal direction division of image provided in an embodiment of the present invention;
Fig. 3 is a kind of schematic diagram of the relative position division of image provided in an embodiment of the present invention;
Fig. 4 is another schematic flow sheet to scheme to search drawing method based on bayonet socket data provided in an embodiment of the present invention;
Fig. 5 is a kind of structural representation to scheme to search map device based on bayonet socket data provided in an embodiment of the present invention.
Specific embodiment
In order that the purpose of the present invention, technical scheme and beneficial effect become more apparent, below in conjunction with accompanying drawing and implementation
Example, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used to explain this hair
It is bright, it is not intended to limit the present invention.
Bayonet socket is road traffic public security bayonet monitoring system in the embodiment of the present invention, and the monitoring system uses advanced light
The technologies such as electricity, computer, image procossing, pattern-recognition, remote data access, car lane, non-motor vehicle to monitoring section
Road carries out round-the-clock monitor in real time and records dependent image data, view data include vehicle by time, place, traveling side
To, the information such as the number-plate number, car plate color, body color, and the information that will be got is transferred to bayonet socket by computer network
Data storage is carried out in the database of System Control Center, inquired about, compared peer processes.
Retrieved from bayonet socket database by being input into the picture containing target vehicle with scheming to search figure in the embodiment of the present invention
Go out a kind of technology of the picture comprising same vehicle.The technology relate to computer vision, image procossing, pattern-recognition, data
The subjects such as library management, information retrieval.
Based on foregoing description, Fig. 1 illustrate it is provided in an embodiment of the present invention it is a kind of based on bayonet socket data to scheme
The flow of drawing method is searched, the flow can be performed by based on bayonet socket data to scheme to search map device.
As shown in figure 1, the specific steps of the flow include:
Step S101, obtains the vehicle body region in target vehicle image.
Step S102, extracts the characteristic information of target vehicle image from the vehicle body region in target vehicle image.
Step S103, yardstick, the main side of the characteristic point of characteristic point and target vehicle image according to target vehicle image
To with relative position query characteristics database in store sample image characteristic point and sample image characteristic point yardstick,
Principal direction and relative position, it is determined that the image similar to target vehicle image.
Specifically, in step S101, before obtaining the vehicle body region in target vehicle image, off-line training vision is first needed
Dictionary, trains the detailed process of visual dictionary as follows:
Obtain bayonet socket database in setting quantity image is as training image and determines the vehicle body area in training image
Domain.Then the characteristic point of training image is extracted from the vehicle body region in training image.Then the feature to training image is clicked through
Row cluster, and every category feature point is defined as a word in visual dictionary, one word numbering of each word correspondence in visual dictionary,
Finally word and word numbering are preserved into visual dictionary.In specific implementation, the image in bayonet socket database refers to each bayonet socket reality
When the image containing vehicle that gathers and upload.Before feature information extraction is carried out to training image, first pass through and manually take off
The mode vehicle body region that determines in training image.The characteristic point of the training image for wherein extracting is the vehicle body region of training image
Local feature characteristic point, local feature can be scale invariant feature conversion (Scale-invariant feature
Transform, abbreviation SIFT) feature, or dense scale invariant feature conversion (Dense Scale-invariant
Feature transform, abbreviation DCSift) feature, one of SIFT feature is usually 128 dimensions, and DCSift features are usual
It is 216 dimensions.In specific implementation, needed to specify the size of visual dictionary, the size of visual dictionary first before visual dictionary is trained
The number of the word for being included in visual dictionary, concrete numerical value can determine according to actual conditions, such as can take 800 or 1000.
The size of visual dictionary is appointed as 1000 in the setting embodiment of the present invention, is then calculated using K mean cluster algorithm or other clusters
Method is clustered the characteristic point of all training images for extracting, and cluster sum is 1000, is regarded per category feature point correspondence after cluster
Feel a word in dictionary, one word numbering of each word correspondence, is designated as W in visual dictionaryid, the sum brief note of word in visual dictionary
It is Tw.Finally 1000 words and corresponding word numbering are preserved into visual dictionary.
In step S101, it is necessary to using accurate vehicle body detection algorithm to target vehicle after acquisition target vehicle image
Positioned, determined the vehicle body region in target vehicle image.Can be using based on integrating channel feature and level in specific implementation
The vehicle body detection algorithm of connection lifting (Cascade boosting).The vehicle body detection algorithm is extracted four spies of passage of image
Levy, four passages include:Color Channel feature and gradient magnitude feature.The vehicle body detection algorithm in terms of model training, model
Using Cascade boosting structures, powerful classification feature, cascade point are realized by cascading multiple iterative algorithm graders
Each layer of class device is an iterative algorithm grader, and an iterative algorithm grader is made up of multiple Weak Classifiers.By many
The feature that individual iterative algorithm grader selects most discrimination is stored in model, realizes optimal classification performance.Vehicle body detection is calculated
Method, using multi-scale sliding window mouthful scanning strategy, then carries out detection zone information fusion when vehicle body region is positioned.
In step s 102, characteristic information includes the characteristic point of target vehicle image and the characteristic point of target vehicle image
Yardstick, principal direction and relative position.It should be noted that in the embodiment of the present invention extract target vehicle image feature with
The feature of the training image extracted during training visual dictionary is identical.The feature of the training image extracted when such as training visual dictionary
It is SIFT feature, then the feature of the target vehicle image of extraction is also SIFT feature.Except extracting mesh in the embodiment of the present invention
Outside the characteristic point in the vehicle body region in mark vehicle image, yardstick, principal direction and the relative position of characteristic point are also extracted.Specifically
Two yardsticks of the characteristic point being extracted in target vehicle image, total scale parameter letter is calculated as Ts, the corresponding volume of current signature point
Number be Sid, the characteristic point numbering of large scale is designated as 1, and the characteristic point numbering of smaller scale is designated as 0.Spy in target vehicle image
The principal direction levied a little is always divided into 12 directions, and the total letter of principal direction is calculated as Td, the division of each principal direction as shown in Fig. 2
12 directions in Fig. 2 are numbered using numeral 0~11, and the corresponding numbering of principal direction of current signature point is Did, Did's
Value is the integer between 0~11.The relative position of the characteristic point in target vehicle image is always divided into 9 positions, with respect to position
The total letter put is calculated as Tp, the division of each relative position using numeral 0~8 to 9 relative positions in Fig. 3 as shown in figure 3, entered
Line number, the corresponding numbering of principal direction of current signature point is Pid, PidValue be integer between 0~8.
In step s 103, property data base is the image determination in bayonet socket database.Determine property data base
Process it is specific as follows:
The image in bayonet socket database is obtained as sample image and the vehicle body region of sample image is determined.Then from sample
The characteristic information of sample image is extracted in vehicle body region in image, characteristic information includes the characteristic point and sample of sample image
The yardstick of the characteristic point of image, principal direction and relative position.Then the characteristic point of sample image is mapped in visual dictionary
Word and determine sample image characteristic point word numbering, by the word of the characteristic point of sample image numbering and the feature of sample image
Yardstick, principal direction and the relative position put are preserved into property data base.
In specific implementation, due to each bayonet socket Real-time Collection and the image containing vehicle is uploaded to bayonet socket database, institute
It is being continuously increased with the image in bayonet socket database.Each image is used as sample in obtaining bayonet socket database in present invention implementation
Image.In order to realize the yardstick of the characteristic point of characteristic point and target vehicle image according to target vehicle image, principal direction and
Relative position determines the image similar to target vehicle image from property data base, the sample extracted when property data base is set up
The feature of this image need to be identical with the feature of the target vehicle image for extracting, such as set up the sample graph extracted during property data base
The feature of picture is SIFT feature, then the feature of the target vehicle image of extraction also must be SIFT feature.Likewise, characteristic point
Yardstick, principal direction and relative position be also required to adopt in a like fashion when extracting.Such as sample image and target vehicle figure
The partition of the scale of the characteristic point of picture, principal direction are divided and relative position is divided and needed unanimously.The characteristic point of sample image is mapped
Word in visual dictionary, vehicle body area maps in so every sample image, afterwards can be according to regarding into the set of multiple words
Feel that the word numbering in dictionary determines the characteristic point corresponding word numbering of every sample image, by the characteristic point of all sample images with
And the corresponding word numbering composition characteristic database of characteristic point, finally by the yardstick of the characteristic point of all sample images, principal direction and
Relative position is preserved into property data base.
Alternatively, according to target vehicle image characteristic point and yardstick, the principal direction of the characteristic point of target vehicle image
With yardstick, the master of the characteristic point of the characteristic point and sample image of the sample image stored in relative position query characteristics database
Direction and relative position, it is determined that the image similar to target vehicle image, detailed process is:
The word that the characteristic point of target vehicle image is mapped in visual dictionary and the characteristic point for determining target vehicle image
Word numbering, the characteristic point according to target vehicle image word numbering filtered out from property data base and target vehicle image
The word numbering identical characteristic point of characteristic point.Then by the word numbering phase in property data base with the characteristic point of target vehicle image
With the corresponding image of characteristic point image as a comparison.Vision word numbering and contrast images according to target vehicle image are regarded
Feel word numbering determines the matching characteristic point of target vehicle image and contrast images, and vision word numbering is according to word numbering, spy
Levy what yardstick a little, principal direction and relative position determined.Then characteristic point according to target vehicle image, the feature of contrast images
The matching characteristic point of point and target vehicle image and contrast images determines the similarity of target vehicle image and contrast images.Most
The contrast images that similarity meets given threshold are defined as the image similar to target vehicle image afterwards.
In specific implementation, after the characteristic point of extraction target vehicle image, the characteristic point of the target vehicle image that will be extracted
Be mapped to the word in visual dictionary, the vehicle body area maps in such target vehicle image into multiple words set, basis afterwards
Visual dictionary determines the corresponding word numbering of the characteristic point of target vehicle image.Then inquired from property data base and target carriage
The corresponding word numbering identical characteristic point of characteristic point of image, same characteristic features point is defined as by word numbering identical characteristic point,
The sample image for containing same characteristic features point with target vehicle image is finally defined as contrast images.
Specifically, the process that contrast images compare with target vehicle image is:First according to the vision of target vehicle image
Word is numbered and the vision word of contrast images numbers the matching characteristic point for determining target vehicle image and contrast images, wherein regarding
Feel that word numbering is determined according to word, the yardstick of characteristic point, principal direction and relative position, specifically meet following formula (1):
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein M is numbered for vision word, WidIt is the numbering of word in visual dictionary, Tw is the sum of word in visual dictionary, Pid
It is the numbering of relative position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidIt is chi
The numbering of degree.
The characteristic point and target vehicle image and comparison diagram of characteristic point, contrast images then according to target vehicle image
The matching characteristic point of picture determines the similarity of target vehicle image and contrast images, specifically meets following formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and right
Than the number of the matching characteristic point of image, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the spy of contrast images
Levy number a little.
The contrast images that similarity meets given threshold are finally defined as the image similar to target vehicle image, are set
Threshold value sets as the case may be.
In the embodiment of the present invention, when feature extraction is carried out to target vehicle image and sample image, except extracting feature
Yardstick, principal direction and the relative position of characteristic point are also extracted outside point, feature extraction algorithm is optimized.Simultaneously by target carriage
When image and contrast images are contrasted, in addition to comparative feature point, yardstick, principal direction and relative position also to characteristic point
Compare, search the precision of similar vehicle image from bayonet socket database so as to improve.In the other embodiment of the present invention
After training visual dictionary and the word numbering that the characteristic point of target vehicle image and sample image is mapped in visual dictionary, profit
Numbered with the word in visual dictionary and filter out contrast images corresponding with target vehicle image from property data base, therefore from card
When searching for the image similar to target vehicle image in mouth database, the contrast images and target vehicle image that will need to only filter out
It is compared, without all images in bayonet socket database and target vehicle image are compared, so as to accelerate extensive
The speed of image retrieval.
In order to preferably explain the embodiment of the present invention, describe the embodiment of the present invention below by specific implement scene and provide
A kind of flow to scheme to search drawing method based on bayonet socket data.
As shown in figure 4, the method is comprised the following steps:
Step S401, off-line training visual dictionary.
Step S402, every image is used as sample image and the vehicle body area of certain sample image in obtaining bayonet socket image library
Domain.
Step S403, extracts yardstick, principal direction and the relative position of characteristic point and characteristic point in sample image vehicle body region
Put.
Step S404, corresponding word is numbered during sample image characteristic point is mapped into visual dictionary.
Step S405, by the yardstick of the corresponding word numbering of sample image characteristic point and characteristic point, principal direction and relative position
Put composition characteristic database.
Step S406, obtains target vehicle image and positions the vehicle body region in target vehicle image.
Step S407, extracts the characteristic point in the vehicle body region of target vehicle image and the yardstick of characteristic point, principal direction and phase
To position.
Step S408, corresponding word is numbered during the characteristic point of target vehicle image is mapped into visual dictionary.
Step S409, filters out the word numbering identical feature with the characteristic point of target vehicle image from property data base
Point.
Step S410, will as a comparison scheme according to word numbering with sample image of the target vehicle image comprising same characteristic features point
Picture.
Step S411, target vehicle image and contrast images is compared and determines target vehicle image with contrast images
Similarity.
Step S412, the contrast images similar to target vehicle image are determined according to similarity.
From the above, it is seen that the embodiment of the present invention provide it is a kind of based on bayonet socket data scheming to search drawing method and dress
Put, including:The vehicle body region in target vehicle image is obtained, is then extracted from the vehicle body region in the target vehicle image
The characteristic information of the target vehicle image, the characteristic information includes the characteristic point and the mesh of the target vehicle image
Mark yardstick, principal direction and the relative position of the characteristic point of vehicle image.Then the characteristic point according to the target vehicle image with
And the sample graph stored in the yardstick of the characteristic point of the target vehicle image, principal direction and relative position query characteristics database
The yardstick of the characteristic point of the characteristic point of picture and the sample image, principal direction and relative position, it is determined that with the target vehicle
The similar image of image, the property data base is that the image in bayonet socket database determines.In the embodiment of the present invention, lead to
The relevant information of the multiple characteristic points and characteristic point of extracting target vehicle image is crossed, and according to the characteristic point and characteristic point extracted
Relevant information finds out similar image from bayonet socket database, rather than license board information, vehicle color, car by vehicle in itself
The intrinsic informations such as type, brand, thus the license board information, color, vehicle, brand in vehicle duplicate or license board information lack when,
Still similar image can be found out from bayonet socket database, improve the precision that similar vehicle image is searched from bayonet socket database.
Based on same idea, Fig. 5 it is exemplary show it is provided in an embodiment of the present invention it is a kind of based on bayonet socket data with
Figure searches the structure of map device, and the device can perform the flow to scheme to search figure based on bayonet socket data.
As shown in figure 5, the device includes:
Acquisition module 501, for obtaining the vehicle body region in target vehicle image;
Characteristic extracting module 502, for extracting the target vehicle from the vehicle body region in the target vehicle image
The characteristic information of image, characteristic point and the target vehicle image of the characteristic information including the target vehicle image
The yardstick of characteristic point, principal direction and relative position;
Processing module 503, for the characteristic point according to the target vehicle image and the spy of the target vehicle image
Levy the characteristic point and the sample of the sample image stored in yardstick a little, principal direction and relative position query characteristics database
The yardstick of the characteristic point of image, principal direction and relative position, it is determined that the image similar to the target vehicle image, the feature
Database is that the image in bayonet socket database determines.
Alternatively, the acquisition module 501 is additionally operable to:
In the image of the setting quantity in the bayonet socket database being obtained as training image and determining the training image
Vehicle body region;
The characteristic point of the training image is extracted from the vehicle body region in the training image;
Characteristic point to the training image is clustered, and every category feature point is defined as in visual dictionary
Word, one word numbering of each word correspondence in the visual dictionary;
The word and word numbering are preserved into the visual dictionary.
Alternatively, the processing module 503 specifically for:
The all images in the bayonet socket database are obtained as sample image and the vehicle body area of the sample image is determined
Domain;
The characteristic information of the sample image, the characteristic information bag are extracted from the vehicle body region in the sample image
Include yardstick, principal direction and the relative position of the characteristic point of the sample image and the characteristic point of the sample image;
The word that the characteristic point of the sample image is mapped in the visual dictionary and the spy for determining the sample image
Levy word numbering a little;
By the yardstick of the word of the characteristic point of sample image numbering and the characteristic point of the sample image, principal direction and
Relative position is preserved into the property data base.
Alternatively, the processing module 503 specifically for:
Word that the characteristic point of the target vehicle image is mapped in the visual dictionary simultaneously determines the target vehicle
The word numbering of the characteristic point of image;
The word numbering of the characteristic point according to the target vehicle image is filtered out and the mesh from the property data base
Mark the word numbering identical characteristic point of the characteristic point of vehicle image;
Word numbering identical characteristic point in the property data base with the characteristic point of the target vehicle image is corresponding
Image image as a comparison;
The vision word numbering of vision word numbering and the contrast images according to the target vehicle image determines institute
The matching characteristic point of target vehicle image and the contrast images is stated, the vision word numbering is according to word numbering, characteristic point
Yardstick, principal direction and relative position determine;
Characteristic point, the characteristic point of the contrast images and the target vehicle image according to the target vehicle image
With the similarity that the matching characteristic of contrast images point determines the target vehicle image and the contrast images;
The contrast images that the similarity meets given threshold are defined as the image similar to the target vehicle image.
Alternatively, the processing module 503 specifically for:
It is described to determine that vision word numbering meets following according to word numbering, the yardstick of characteristic point, principal direction and relative position
Formula (1):
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein, M is vision word numbering, WidIt is the word numbering in visual dictionary, Tw is the sum of word in visual dictionary,
PidIt is the numbering of relative position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidFor
The numbering of yardstick.
The characteristic point according to the target vehicle image, the characteristic point of the contrast images and the target vehicle
The matching characteristic point of image and the contrast images determines that the target vehicle image and the similarity of the contrast images meet
Following formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and right
Than the number of the matching characteristic point of image, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the spy of contrast images
Levy number a little.
It should be understood by those skilled in the art that, embodiments of the invention can be provided as method or computer program product.
Therefore, the present invention can be using the embodiment in terms of complete hardware embodiment, complete software embodiment or combination software and hardware
Form.And, the present invention can be used to be can use in one or more computers for wherein including computer usable program code and deposited
The shape of the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The present invention is the flow with reference to method according to embodiments of the present invention, equipment (system) and computer program product
Figure and/or block diagram are described.It should be understood that every first-class during flow chart and/or block diagram can be realized by computer program instructions
The combination of flow and/or square frame in journey and/or square frame and flow chart and/or block diagram.These computer programs can be provided
The processor of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that produced for reality by the instruction of computer or the computing device of other programmable data processing devices
The device of the function of being specified in present one flow of flow chart or multiple one square frame of flow and/or block diagram or multiple square frames.
These computer program instructions may be alternatively stored in can guide computer or other programmable data processing devices with spy
In determining the computer-readable memory that mode works so that instruction of the storage in the computer-readable memory is produced and include finger
Make the manufacture of device, the command device realize in one flow of flow chart or multiple one square frame of flow and/or block diagram or
The function of being specified in multiple square frames.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that in meter
Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented treatment, so as in computer or
The instruction performed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in individual square frame or multiple square frames.
, but those skilled in the art once know basic creation although preferred embodiments of the present invention have been described
Property concept, then can make other change and modification to these embodiments.So, appended claims are intended to be construed to include excellent
Select embodiment and fall into having altered and changing for the scope of the invention.
Obviously, those skilled in the art can carry out various changes and modification without deviating from essence of the invention to the present invention
God and scope.So, if these modifications of the invention and modification belong to the scope of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to comprising these changes and modification.
Claims (10)
1. it is a kind of based on bayonet socket data scheming to search drawing method, it is characterised in that including:
Obtain the vehicle body region in target vehicle image;
The characteristic information of the target vehicle image, the feature letter are extracted from the vehicle body region in the target vehicle image
Breath includes the yardstick of the characteristic point of the characteristic point of the target vehicle image and the target vehicle image, principal direction and relatively
Position;
The yardstick of the characteristic point of characteristic point and the target vehicle image according to the target vehicle image, principal direction and phase
To yardstick, the master of the characteristic point of the characteristic point and the sample image of the sample image of storage in the query characteristics database of position
Direction and relative position, it is determined that the image similar to the target vehicle image, the property data base is according to bayonet socket data
What the image in storehouse determined.
2. the method for claim 1, it is characterised in that before the vehicle body region in the acquisition target vehicle image,
Also include:
The image of the setting quantity in the bayonet socket database is obtained as training image and the car in the training image is determined
Body region;
The characteristic point of the training image is extracted from the vehicle body region in the training image;
Characteristic point to the training image is clustered, and every category feature point is defined as a word in visual dictionary, institute
State one word numbering of each word correspondence in visual dictionary;
The word and word numbering are preserved into the visual dictionary.
3. method as claimed in claim 2, it is characterised in that the image in the bayonet socket database determines feature
Database, including:
The all images in the bayonet socket database are obtained as sample image and the vehicle body region of the sample image is determined;
The characteristic information of the sample image is extracted from the vehicle body region in the sample image, the characteristic information includes institute
State yardstick, principal direction and the relative position of the characteristic point of sample image and the characteristic point of the sample image;
The word that the characteristic point of the sample image is mapped in the visual dictionary and the characteristic point for determining the sample image
Word numbering;
By the yardstick of the word of the characteristic point of sample image numbering and the characteristic point of the sample image, principal direction and relative
Position is preserved into the property data base.
4. method as claimed in claim 3, it is characterised in that the characteristic point and institute according to the target vehicle image
State the sample image of storage in yardstick, principal direction and the relative position query characteristics database of the characteristic point of target vehicle image
The yardstick of the characteristic point of characteristic point and the sample image, principal direction and relative position, it is determined that with the target vehicle image
Similar image, including:
Word that the characteristic point of the target vehicle image is mapped in the visual dictionary simultaneously determines the target vehicle image
Characteristic point word numbering;
The word numbering of the characteristic point according to the target vehicle image is filtered out and the target carriage from the property data base
The word numbering identical characteristic point of the characteristic point of image;
By figure corresponding with the word numbering identical characteristic point of the characteristic point of the target vehicle image in the property data base
As image as a comparison;
The vision word numbering of vision word numbering and the contrast images according to the target vehicle image determines the mesh
The matching characteristic point of mark vehicle image and the contrast images, the vision word numbering is the chi according to word numbering, characteristic point
What degree, principal direction and relative position determined;
Characteristic point, the characteristic point of the contrast images and the target vehicle image and institute according to the target vehicle image
The matching characteristic point for stating contrast images determines the similarity of the target vehicle image and the contrast images;
The contrast images that the similarity meets given threshold are defined as the image similar to the target vehicle image.
5. method as claimed in claim 4, it is characterised in that described according to word numbering, the yardstick of characteristic point, principal direction and phase
Determine that vision word numbering meets following formula (1) to position:
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein, M is vision word numbering, WidIt is the word numbering in visual dictionary, Tw is the sum of word in visual dictionary, PidIt is phase
To the numbering of position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidIt is yardstick
Numbering;
The characteristic point according to the target vehicle image, the characteristic point of the contrast images and the target vehicle image
Determine that the target vehicle image and the similarity of the contrast images meet following with the matching characteristic of contrast images point
Formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and comparison diagram
The number of the matching characteristic point of picture, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the characteristic point of contrast images
Number.
6. it is a kind of based on bayonet socket data scheming to search map device, it is characterised in that including:
Acquisition module, for obtaining the vehicle body region in target vehicle image;
Characteristic extracting module, the spy for extracting the target vehicle image from the vehicle body region in the target vehicle image
Reference ceases, the characteristic point of characteristic point and the target vehicle image of the characteristic information including the target vehicle image
Yardstick, principal direction and relative position;
Processing module, for the characteristic point according to the target vehicle image and the chi of the characteristic point of the target vehicle image
The characteristic point of the sample image stored in degree, principal direction and relative position query characteristics database and the spy of the sample image
Yardstick a little, principal direction and relative position are levied, it is determined that the image similar to the target vehicle image, the property data base is
What the image in bayonet socket database determined.
7. device as claimed in claim 6, it is characterised in that the acquisition module is additionally operable to:
The image of the setting quantity in the bayonet socket database is obtained as training image and the car in the training image is determined
Body region;
The characteristic point of the training image is extracted from the vehicle body region in the training image;
Characteristic point to the training image is clustered, and every category feature point is defined as a word in visual dictionary, institute
State one word numbering of each word correspondence in visual dictionary;
The word and word numbering are preserved into the visual dictionary.
8. device as claimed in claim 7, it is characterised in that the processing module specifically for:
The all images in the bayonet socket database are obtained as sample image and the vehicle body region of the sample image is determined;
The characteristic information of the sample image is extracted from the vehicle body region in the sample image, the characteristic information includes institute
State yardstick, principal direction and the relative position of the characteristic point of sample image and the characteristic point of the sample image;
The word that the characteristic point of the sample image is mapped in the visual dictionary and the characteristic point for determining the sample image
Word numbering;
By the yardstick of the word of the characteristic point of sample image numbering and the characteristic point of the sample image, principal direction and relative
Position is preserved into the property data base.
9. device as claimed in claim 8, it is characterised in that the processing module specifically for:
Word that the characteristic point of the target vehicle image is mapped in the visual dictionary simultaneously determines the target vehicle image
Characteristic point word numbering;
The word numbering of the characteristic point according to the target vehicle image is filtered out and the target carriage from the property data base
The word numbering identical characteristic point of the characteristic point of image;
By figure corresponding with the word numbering identical characteristic point of the characteristic point of the target vehicle image in the property data base
As image as a comparison;
The vision word numbering of vision word numbering and the contrast images according to the target vehicle image determines the mesh
The matching characteristic point of mark vehicle image and the contrast images, the vision word numbering is the chi according to word numbering, characteristic point
What degree, principal direction and relative position determined;
Characteristic point, the characteristic point of the contrast images and the target vehicle image and institute according to the target vehicle image
The matching characteristic point for stating contrast images determines the similarity of the target vehicle image and the contrast images;
The contrast images that the similarity meets given threshold are defined as the image similar to the target vehicle image.
10. device as claimed in claim 9, it is characterised in that the processing module specifically for:
It is described to determine that vision word numbering meets following formula according to word numbering, the yardstick of characteristic point, principal direction and relative position
(1):
M=Wid+Tw×(Pid+Tp×(Did+Td×Sid))………………………………(1)
Wherein, M is vision word numbering, WidIt is the word numbering in visual dictionary, Tw is the sum of word in visual dictionary, PidIt is phase
To the numbering of position, TpIt is the sum of relative position, DidIt is the numbering of principal direction, TdIt is the sum of principal direction, SidIt is yardstick
Numbering;
The characteristic point according to the target vehicle image, the characteristic point of the contrast images and the target vehicle image
Determine that the target vehicle image and the similarity of the contrast images meet following with the matching characteristic of contrast images point
Formula (2):
Wherein, F (A, B) is the similarity of target vehicle image and contrast images, and P (A ∩ B) is target vehicle image and comparison diagram
The number of the matching characteristic point of picture, P (A) is the number of the characteristic point of target vehicle image, and P (B) is the characteristic point of contrast images
Number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710033558.3A CN106777350B (en) | 2017-01-16 | 2017-01-16 | Method and device for searching pictures with pictures based on bayonet data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710033558.3A CN106777350B (en) | 2017-01-16 | 2017-01-16 | Method and device for searching pictures with pictures based on bayonet data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106777350A true CN106777350A (en) | 2017-05-31 |
CN106777350B CN106777350B (en) | 2020-02-14 |
Family
ID=58946246
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710033558.3A Active CN106777350B (en) | 2017-01-16 | 2017-01-16 | Method and device for searching pictures with pictures based on bayonet data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106777350B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107729502A (en) * | 2017-10-18 | 2018-02-23 | 公安部第三研究所 | A kind of bayonet vehicle individualized feature intelligent retrieval system and method |
CN110334763A (en) * | 2019-07-04 | 2019-10-15 | 北京字节跳动网络技术有限公司 | Model data file generation, image-recognizing method, device, equipment and medium |
CN111127541A (en) * | 2018-10-12 | 2020-05-08 | 杭州海康威视数字技术股份有限公司 | Vehicle size determination method and device and storage medium |
WO2021217924A1 (en) * | 2020-04-29 | 2021-11-04 | 平安国际智慧城市科技股份有限公司 | Method and apparatus for identifying vehicle type at traffic checkpoint, and device and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104199842A (en) * | 2014-08-07 | 2014-12-10 | 同济大学 | Similar image retrieval method based on local feature neighborhood information |
CN104239898A (en) * | 2014-09-05 | 2014-12-24 | 浙江捷尚视觉科技股份有限公司 | Method for carrying out fast vehicle comparison and vehicle type recognition at tollgate |
CN104699726A (en) * | 2013-12-18 | 2015-06-10 | 杭州海康威视数字技术股份有限公司 | Vehicle image retrieval method and device for traffic block port |
CN105354533A (en) * | 2015-09-28 | 2016-02-24 | 江南大学 | Bag-of-word model based vehicle type identification method for unlicensed vehicle at gate |
-
2017
- 2017-01-16 CN CN201710033558.3A patent/CN106777350B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104699726A (en) * | 2013-12-18 | 2015-06-10 | 杭州海康威视数字技术股份有限公司 | Vehicle image retrieval method and device for traffic block port |
CN104199842A (en) * | 2014-08-07 | 2014-12-10 | 同济大学 | Similar image retrieval method based on local feature neighborhood information |
CN104239898A (en) * | 2014-09-05 | 2014-12-24 | 浙江捷尚视觉科技股份有限公司 | Method for carrying out fast vehicle comparison and vehicle type recognition at tollgate |
CN105354533A (en) * | 2015-09-28 | 2016-02-24 | 江南大学 | Bag-of-word model based vehicle type identification method for unlicensed vehicle at gate |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107729502A (en) * | 2017-10-18 | 2018-02-23 | 公安部第三研究所 | A kind of bayonet vehicle individualized feature intelligent retrieval system and method |
CN111127541A (en) * | 2018-10-12 | 2020-05-08 | 杭州海康威视数字技术股份有限公司 | Vehicle size determination method and device and storage medium |
CN111127541B (en) * | 2018-10-12 | 2024-02-27 | 杭州海康威视数字技术股份有限公司 | Method and device for determining vehicle size and storage medium |
CN110334763A (en) * | 2019-07-04 | 2019-10-15 | 北京字节跳动网络技术有限公司 | Model data file generation, image-recognizing method, device, equipment and medium |
CN110334763B (en) * | 2019-07-04 | 2021-07-23 | 北京字节跳动网络技术有限公司 | Model data file generation method, model data file generation device, model data file identification device, model data file generation apparatus, model data file identification apparatus, and model data file identification medium |
WO2021217924A1 (en) * | 2020-04-29 | 2021-11-04 | 平安国际智慧城市科技股份有限公司 | Method and apparatus for identifying vehicle type at traffic checkpoint, and device and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN106777350B (en) | 2020-02-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105160309B (en) | Three lanes detection method based on morphological image segmentation and region growing | |
CN108596277B (en) | Vehicle identity recognition method and device and storage medium | |
CN107451607B (en) | A kind of personal identification method of the typical character based on deep learning | |
CN104063712B (en) | A kind of information of vehicles extracting method and system | |
CN102076531B (en) | Vehicle clear path detection | |
CN104715612B (en) | The suspected vehicles tracking and device realized based on wheelpath matching | |
CN109857889A (en) | A kind of image search method, device, equipment and readable storage medium storing program for executing | |
CN110020651A (en) | Car plate detection localization method based on deep learning network | |
CN106022285A (en) | Vehicle type identification method and vehicle type identification device based on convolutional neural network | |
CN103679191B (en) | An automatic fake-licensed vehicle detection method based on static state pictures | |
CN105844228A (en) | Remote sensing image cloud detection method based on convolution nerve network | |
CN106777350A (en) | It is a kind of based on bayonet socket data scheming to search drawing method and device | |
CN109241349A (en) | A kind of monitor video multiple target classification retrieving method and system based on deep learning | |
CN109271991A (en) | A kind of detection method of license plate based on deep learning | |
CN102054178A (en) | Chinese painting image identifying method based on local semantic concept | |
CN108830254B (en) | Fine-grained vehicle type detection and identification method based on data balance strategy and intensive attention network | |
CN107506368A (en) | The determination method and device of one species case suspected vehicles | |
CN109886147A (en) | A kind of more attribute detection methods of vehicle based on the study of single network multiple-task | |
CN104156734A (en) | Fully-autonomous on-line study method based on random fern classifier | |
CN108647595A (en) | Vehicle recognition methods again based on more attribute depth characteristics | |
CN109033175A (en) | A kind of method and system to scheme to search vehicle | |
CN109214345A (en) | The method for searching the driving trace for changing board vehicle is compared based on similarity | |
CN106897681A (en) | A kind of remote sensing images comparative analysis method and system | |
CN107273832A (en) | Licence plate recognition method and system based on integrating channel feature and convolutional neural networks | |
CN107203775A (en) | A kind of method of image classification, device and equipment |
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 |