CN106577350A - Method and device for recognizing pet type - Google Patents
Method and device for recognizing pet type Download PDFInfo
- Publication number
- CN106577350A CN106577350A CN201611032333.8A CN201611032333A CN106577350A CN 106577350 A CN106577350 A CN 106577350A CN 201611032333 A CN201611032333 A CN 201611032333A CN 106577350 A CN106577350 A CN 106577350A
- Authority
- CN
- China
- Prior art keywords
- pet
- image
- feature
- matching
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012216 screening Methods 0.000 claims description 38
- 238000006073 displacement reaction Methods 0.000 claims description 28
- 230000007246 mechanism Effects 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 12
- 238000012546 transfer Methods 0.000 claims description 7
- 241000282326 Felis catus Species 0.000 description 18
- 238000010586 diagram Methods 0.000 description 14
- 210000003128 head Anatomy 0.000 description 12
- 230000006870 function Effects 0.000 description 11
- 230000006854 communication Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 241000282472 Canis lupus familiaris Species 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 5
- 241001465754 Metazoa Species 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 4
- 230000036651 mood Effects 0.000 description 4
- 210000004556 brain Anatomy 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000001815 facial effect Effects 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 206010037180 Psychiatric symptoms Diseases 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000003705 background correction Methods 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000008676 import Effects 0.000 description 2
- 238000007873 sieving Methods 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 241001580935 Aglossa pinguinalis Species 0.000 description 1
- 241000208340 Araliaceae Species 0.000 description 1
- 241000271566 Aves Species 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 241000879777 Lynx rufus Species 0.000 description 1
- 241000124008 Mammalia Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007175 bidirectional communication Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K29/00—Other apparatus for animal husbandry
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Environmental Sciences (AREA)
- Animal Husbandry (AREA)
- Biodiversity & Conservation Biology (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a method and device for recognizing a pet type. The method includes the steps that an intelligent terminal is used for acquiring a pet image; a plurality of body features of a pet are extracted according to the pet image; and the body features of the pet are matched with data in a pet body feature database, so that the matched pet type is obtained. A user can know the pet type quickly and enhance the understanding of the pet, and the method and device can help the user to select the pet that he or she loves.
Description
Technical field
The present invention relates to image identification technical field, especially relates to the method and dress of a kind of pet type identification
Put.
Background technology
Pet refers generally to people for the animal for eliminating loneliness or feed for amusement purpose.Pet is for spirit
Purpose and the animals and plants fed.It is general in order to eliminate loneliness, or amusement and feed, the usually animal of Mammalia or Aves, because
It is more flourishing for these animal brains, easily exchange with people.
Pet breeds are various, and common pet has cat, dog, fish, bird etc..The kind of pet cat is broadly divided into Garfield,
Folding ear cat, shorthair, Persian cat, Siamese cat, tabby cat, Jin Jila, Bali cat, Birman, Bombay cat, Birman, Egyptian Mau,
Maine Coon, Europe cat, cloth idol cat, frizzle cat, Singapura Cat, Norwegian Forest Cat, fluorescence cat, Somalia cat, Turkey Buddhist cat is beautiful
42 kinds of state bobcat etc. (adopts U.S.'s CAF standards).And《AKC dog registers》In record, whole world kind of dog kind more than 300, AKC recognizes
Can 149 kinds, including A Fenpingjia dogs, afghan hound, omnipotent stalk, autumn fields dog, Alaskan Malamute etc..
Obviously, it is a very difficult thing to go identification so miscellaneous pet kind by personal experience.And it is right
The user of little pet is liked in one, it is seen that a lovely little pet, but it is known nothing, be a thing unfortunately
Feelings.
The content of the invention
The main object of the present invention is to provide a kind of method and apparatus of pet type identification, helps user's identification pet
Type.
The present invention proposes a kind of method of pet type identification, including:
Obtain pet image;
According to multiple physical traits of the pet image zooming-out pet;
Multiple physical traits of the pet are matched with the data in pet physical trait database, is matched
Pet type.
Preferably, it is described that the pet image is shot by camera, including:
Pet is shot by camera, initial pet image is obtained;
Described in records photographing during initial pet image, the displacement information of camera;
Judge that institute's displacement information whether in the range of the displacement threshold value of the permission identification of setting, is judging the displacement
When information is in the range of the displacement threshold value, using the initial pet image as pet image.
Preferably, the multiple physical traits according to the pet image zooming-out pet, including:
The overall profile of pet is extracted from pet image;
The band of position of face, trunk, tail and four limbs is positioned from the overall profile of the pet;
Face feature, the trunk for extracting pet respectively in the band of position of the face, trunk, tail and four limbs is special
Levy, tail feature and four limbs feature, the face feature include eye feature, face feature, nose feature, ear feature and
Face mask;
By the face feature of the pet, trunk feature, tail feature together with four limbs combinations of features, pet body is formed
Body characteristicses set, each feature includes two parameters of CF.
Preferably, the data in the pet physical trait database include:
The matching image set of multiple pet types, the matching image set of each pet type includes that one or more is default
Pet matches image, wherein the pet matching image comprising a carrying pet type mark, is designated as representing pet image;
From the reference pet physical trait set obtained per a pet matching image zooming-out, the reference pet body
Multiple features of pet described in image are matched in characteristic set comprising the pet;
Distinguishing characteristics set corresponding with the matching image set of each pet type, the distinguishing characteristics set expression is a certain
One or more distinguishing characteristics of type pet, the distinguishing characteristics is to carry out area for the type pet and other types pet
The feature divided.
Preferably, the data by pet physical trait set and pet physical trait database are carried out
Match somebody with somebody, the pet type for obtaining matching includes:
Judge pet physical trait set whether comprising distinguishing characteristics set;
If comprising distinguishing characteristics set, judging whether the distinguishing characteristics set for including is more than one;
If so, then by the distinguishing characteristics set for including be designated as sieve distinguishing characteristics set, using matching mechanisms to described many
Individual screening distinguishing characteristics set is sieved, and obtains the pet type for matching;
If comprising a distinguishing characteristics set, obtaining the matching figure of pet type corresponding with the distinguishing characteristics set
The pet type mark of image set, obtains the pet type for matching.
Preferably, the employing matching mechanisms are sieved to the plurality of screening distinguishing characteristics set, obtain what is matched
Pet type, including:
Obtain the screening matching image set of pet type corresponding with screening distinguishing characteristics set;
Transfer all pet images in the screening matching image set, the corresponding reference pet of every pet image of acquisition
Physical trait set;
Feature in the pet physical trait set is entered with the feature in the reference pet physical trait set transferred
Row is compared one by one, calculates the matching degree of the reference pet physical trait set that pet physical trait set is transferred with each;
By the reference pet transferred with each according to all pet physical traits set in the screening matching image set
The matching degree of physical trait set, calculates the feature and the reference pet physical trait transferred in the pet physical trait set
The matching degree of set, calculates the screening matching image set and the mean value of the pet physical trait sets match degree, obtains
As pet image and the whole matching degree of pet type;
The corresponding pet type of whole matching degree highest screening matching image set reference pet physical trait set is made
To match pet type.
The invention allows for a kind of device of pet type identification, including:
Acquisition module, for obtaining pet image;
Characteristic extracting module, for according to the physical trait of the pet image zooming-out pet;
Matching module, for multiple physical traits of the pet to be carried out with the data in pet physical trait database
Matching, obtains the pet type for matching.
Preferably, the characteristic extracting module includes:
Overall profile extraction unit, for the overall profile for extracting pet from pet image;
Body position positioning unit, for positioning face, trunk, tail and four limbs from the overall profile of the pet
The band of position;
Feature extraction unit, for extracting pet respectively in the band of position of the face, trunk, tail and four limbs
Face feature, trunk feature, tail feature and four limbs feature, the face feature includes that eye feature, face feature, nose are special
Levy, ear feature and face mask;
By the face feature of the pet, trunk feature, tail feature together with four limbs combinations of features, pet body is formed
Body characteristicses set, each feature includes two parameters of CF.
Preferably, the pet physical trait database includes:
The matching image set of multiple pet types, the matching image set of each pet type includes that one or more is default
Pet matches image, wherein the pet matching image comprising a carrying pet type mark, is designated as representing pet image;
From the reference pet physical trait set obtained per a pet matching image zooming-out, the reference pet body
Multiple features of pet described in image are matched in characteristic set comprising the pet;
Distinguishing characteristics set corresponding with the matching image set of each pet type, the distinguishing characteristics set expression is a certain
One or more distinguishing characteristics of type pet, the distinguishing characteristics is to carry out area for the type pet and other types pet
The feature divided.
Preferably, the matching module includes:
Matching unit, for judging pet physical trait set whether comprising distinguishing characteristics set;
Matching unit, for judging pet physical trait set whether comprising distinguishing characteristics set;
Interpretation of result unit, if for analyzing comprising distinguishing characteristics set, judging whether is the distinguishing characteristics set that includes
It is more than one;
Many result treatment units, when there are multiple results for interpretation of result unit, then by the distinguishing characteristics set for including
It is designated as sieving distinguishing characteristics set, the plurality of screening distinguishing characteristics set is sieved using matching mechanisms, is matched
Pet type;
Single result treatment unit, during for interpretation of result unit only one of which result, then obtains and the distinguishing characteristics
Gather the pet type mark of the matching image set of corresponding pet type, obtain the pet type for matching.
Pet kind identification method proposed by the present invention and device, in its method, intelligent terminal obtains pet image;According to
Multiple physical traits of the pet image zooming-out pet;By multiple physical traits of the pet and pet physical trait data
Data in storehouse are matched, and obtain the pet type for matching, and user can be made quickly to know the type of pet, are promoted to pet
Understand, help user to select the pet oneself liked.
Description of the drawings
Fig. 1 is the schematic flow sheet of the first embodiment of the method for pet type identification of the present invention;
Fig. 2 is the data type schematic diagram in the method pet physical trait database of pet type identification of the present invention;
Fig. 3 is the structural representation of the first embodiment of the device of pet type identification of the present invention;
Fig. 4 is the structural representation of the second embodiment of the device of pet type identification of the present invention;
Fig. 5 is the structural representation of the 3rd embodiment of the device of pet type identification of the present invention;
Fig. 6 is the structural representation of the 5th embodiment of the device of pet type identification of the present invention;
Fig. 7 is the structural representation of the sixth embodiment of the device of pet type identification of the present invention;
Fig. 8 is the design analysis diagram of pet kind identification method of the present invention and device.
The realization of the object of the invention, functional characteristics and advantage will be described further referring to the drawings in conjunction with the embodiments.
Specific embodiment
Those skilled in the art of the present technique are appreciated that unless expressly stated, singulative " " used herein, "
It is individual ", " described " " above-mentioned " and " being somebody's turn to do " may also comprise plural form.It is to be further understood that making in the specification of the present invention
Wording " including " refers to there is the feature, integer, step, operation, element, unit, module and/or component, but and
Do not preclude the presence or addition of one or more other features, integer, step, operation, element, unit, module, component and/or it
Group.It should be understood that when we claim element to be " connected " or during " coupled " to another element, it can be directly connected to or couple
To other elements, or can also there is intermediary element.Additionally, " connection " used herein or " coupling " can include wirelessly connecting
Connect or wirelessly couple.Wording "and/or" used herein includes one or more associated listing the whole or arbitrary of item
Unit and all combination.
Those skilled in the art of the present technique are appreciated that unless otherwise defined all terms used herein are (including technology art
Language and scientific terminology), with art of the present invention in those of ordinary skill general understanding identical meaning.Should also
It is understood by, those terms defined in such as general dictionary, it should be understood that with the context with prior art
The consistent meaning of meaning, and unless by specific definitions as here, will not otherwise use idealization or excessively formal implication
To explain.
Those skilled in the art of the present technique be appreciated that " terminal " used herein above, " intelligent terminal ", " mobile terminal ",
" terminal device ", both including the equipment of wireless signal receiver, it only possesses setting for the wireless signal receiver of non-emissive ability
It is standby, but including receive and launch hardware equipment, its have can on bidirectional communication link, perform two-way communication reception and
The equipment of transmitting hardware.This equipment can include:Honeycomb or other communication equipments, it has single line display or multi-line
Display or the honeycomb without multi-line display or other communication equipments;PCS(Personal Communications
Service, PCS Personal Communications System), it can be with combine voice, data processing, fax and/or its communication ability;PDA
(Personal Digital Assistant, personal digital assistant), its can include radio frequency receiver, pager, internet/
Intranet access, web browser, notepad, calendar and/or GPS (Global Positioning System, global location
System) receiver;Conventional laptop and/or palmtop computer or other equipment, it has and/or including radio frequency receiver
Conventional laptop and/or palmtop computer or other equipment." terminal " used herein above, " terminal device " can be portable
Formula, can transport, in the vehicles (aviation, sea-freight and/or land), or be suitable for and/or be configured to local
Operation, and/or with distribution form, operate in any other position operation in the earth and/or space.It is used herein above " whole
End ", " intelligent terminal ", " terminal device " can also be communication terminal, access terminals, music/video playback terminal, for example can be with
It is PDA, MID (Mobile Internet Device, mobile internet device) and/or with music/video playing function
The equipment such as mobile phone, or intelligent television, Set Top Box.
It should be appreciated that specific embodiment described herein is not intended to limit the present invention only to explain the present invention.
The present invention proposes a kind of method of pet type identification, and the method for the pet type identification can apply to control eventually
End, control terminal can be but be not limited to panel computer, mobile phone or other smart machines.The present embodiment and following examples are with hand
Illustrate as a example by machine.Image recognition processes are usually that image is uploaded onto the server by mobile phone, are parsed by server and are obtained
As a result result is passed back mobile phone again.With reference to Fig. 1, Fig. 1 is the flow process of the first embodiment of the method for pet type identification of the present invention
Schematic diagram.The method of pet type identification proposed by the present invention, comprises the steps:
S10, acquisition pet image;
S20, according to multiple physical traits of the pet image zooming-out pet;
S30, multiple physical traits of the pet are matched with the data in pet physical trait database, obtained
The pet type of matching.
First it is to obtain pet image as described in step S10, the image can be shot by mobile phone and be obtained, it is also possible to logical
The image crossed on the image or load networks that import interior of mobile phone.
As described in step S20, the physical trait of pet can be extracted by existing technology of identification.
In step s 30, the pet data of a large amount of types are stored in pet physical trait database.A certain type is doted on
Thing also stores substantial amounts of individual pet characteristic information.The statistics of database arrangement feelings of a certain feature of a certain type pet
Condition, if this feature discreteness is larger, illustrates that this feature is not the distinguishing characteristic of the type pet, then this feature is in matching meter
During calculation, weight accounting is negligible or accounting is little;If this feature discreteness is little, illustrate that this feature is the differentiation spy of the type pet
Levy, then this feature is in matching primitives, and weight accounting is big.Specific accounting weight is according to the different and different of pet type.Such as
Pet X includes that distinguishing characteristics A and distinguishing characteristics B, distinguishing characteristics A belong to type-A pet simultaneously, and distinguishing characteristics B belongs to B types and dotes on
Thing.Now, device can analyze other features of pet X and whether fall into the scope of type-A or B type pets, if only type-A symbol
Close and the conjunction of B type mismatch, then preferentially transmit the pet picture and title of type-A.In matching process, the set of pet physical trait with
The mode that pet physical trait database is matched is, each characteristic variable in pet physical trait set and pet body
The pet characteristic of each type is matched one by one in property data base, obtains matching result.
After corresponding pet type is matched, doting on for user's matching can be informed by output matching result to user
Species type.The result of output can include pet typonym, a certain picture of the type pet or pictures.From figure
Piece comes from pet database, and the pet database contains the pet image of character pair in pet physical trait database.
From picture it is typically identical with the action of pet in the image to be recognized.Output result can be single result, but be not limited to
Single result.Judged result can be sequentially output according to the height of matching degree, the high preferential display of matching degree, user to the result not
Satisfaction can switch to the result of relatively low matching degree.
Further, based on pet type identification of the present invention method first embodiment, the invention allows for pet
The second embodiment of the method for type identification, and unlike the first embodiment of the method for pet type identification, step S10 bag
Include:
S101, by camera shoot pet, obtain initial pet image;
Described in S102, records photographing during initial pet image, the displacement information of camera;
S103, institute's displacement information is judged whether in the range of the displacement threshold value of the permission identification of setting, judging
When displacement information is in the range of the displacement threshold value, using the initial pet image as pet image.
In step S101, under this scene, user is to be directed at pet by mobile phone to take pictures, and obtains the image of pet.Identification
Device can read the model and image-forming condition of camera.Image-forming condition includes pixel value, the photosensitive coefficients of ISO, white balance parameter etc..
Device for identifying can be optimized according to the camera model of concrete model and image-forming condition to image, make the pet view data and reality
Actual value closer to.
In step S102, S103, user sees a pet oneself admired, and mood can be more excited, handheld mobile phone
When taking pictures, hand probably due to mood is excited and can not smoothly shoot pet image.The detectable user of the method for the present embodiment
Whether hand shaking excessively causes the pet image for shooting to be not used to recognize pet type when taking pictures.
Specific method is to obtain accessory by three-axis gyroscope or other relevant positions, obtaining and dotes on described in shoot in mobile phone
Device location information during object image, is recorded as front positional information is (x0, y0, z0), the positional information after imaging is (x1,
y1, z1), then can calculate its displacement information Δ X.Specially:
Judge whether institute's displacement information exceedes setting and allow the displacement threshold value of identification.If being not above, the image can
For pet type identification, if exceeding, user is reminded to re-shoot.
Further, based on pet type identification of the present invention method first embodiment, the invention allows for pet
The 3rd embodiment of the method for type identification, and unlike the first embodiment of the method for pet type identification, step S20 bag
Include:
S21, from pet image extract pet overall profile;
S22, the band of position for positioning from the overall profile of the pet face, trunk, tail and four limbs;
S23, face feature, trunk that pet is extracted in the band of position of the face, trunk, tail and four limbs respectively
Feature, tail feature and four limbs feature, the face feature include eye feature, face feature, nose feature, ear feature with
And face mask.
As described in step S21, the identification to pet profile is realized by setting up neural network model.By a large amount of
Picture training, make device to distinguish the difference of pet and background in pet image, and then find out the profile of pet.
Usually, for regard to recognizing pet type, identification face feature is mostly important.Using the training of neural network model
Mechanism, device for identifying may recognize that the face mask of pet.According to the color and the difference of environmental colors of pet face, week is judged
The ambient parameter in collarette border.The overall profile of pet is recognized with reference to color in the profile of the face and the ambient parameter.Than
Environment as residing for recognizing a cat, it is on meadow or on road on sheet or on floor to be.The back of the body is deducted
After scape, the overall profile of pet just can be obtained.
In some complex backgrounds, the pet overall profile obtained after background correction is not real pet wheel
Exterior feature, further comprises part background, now can retain rational outline portion by pet body contour location model.
As described in step S22, on the basis of pet profile is identified, the profile of pet is divided into into four regions:Head
Portion, trunk, four limbs and tail.Because the feature of the angle problem being imaged, four limbs and tail may in the picture not significantly, in system
During meter feature, the feature at two can be ignored according to actual conditions.
In step S23, the feature for recognizing head is mostly important, and the feature of head has specifically included ear shape, ear
Color, ear relative to face size, ear location etc.;Eye shape, eyes relative to face size, eye color,
Eyes are in face location etc.;Nose shape, nose relative to face size, nose color, nose is in position of face etc.;Mouth
Bar shape, face relative to face size, face color, face is in the position of face;Face relative to trunk size, face
Portion's color, facial contours etc..
The feature of trunk mainly includes size of the trunk relative to head, color and distribution of color of trunk etc..
The feature of four limbs mainly includes that leg is long, leg color, thickness of leg etc..
The feature of tail mainly includes tail length, tail color and tail thickness etc..But in actual photographed, tail is usually
Intactly can not come across among image, thus device can intelligent screening can discriminate another characteristic.
The physical trait of above pet, each feature is represented with a variable.Extract all of pet on pet image
All features are constituted a pet physical trait set by feature.But it is not that each feature can be by during extraction
Extract.Now, in the set of pet physical trait, the value for representing the variable of the feature that can not be extracted is designated as 0.
Further, based on pet type identification of the present invention method 3rd embodiment, the invention allows for pet
The fourth embodiment of the method for type identification, and unlike the 3rd embodiment of the method for pet type identification, in step S30
Data in pet physical trait database include:
The matching image set 301 of multiple pet types, the matching image set of each pet type includes that one or more is pre-
If pet matching image, wherein comprising one carrying pet type mark the pet matching image, be designated as representing pet
Image;
From the reference pet physical trait set 302 obtained per a pet matching image zooming-out, the reference pet
Multiple features of pet described in image are matched in physical trait set comprising the pet;
Distinguishing characteristics set 303 corresponding with the matching image set of each pet type, the distinguishing characteristics set expression
One or more distinguishing characteristics of a certain type pet, the distinguishing characteristics is to enter with other types pet for the type pet
The feature that row is distinguished.
As shown in Fig. 2 pet physical trait database includes the characteristic of various pet types, every kind of pet type
Characteristic be divided into three kinds of data types again, respectively match image set, reference pet physical trait set and distinguishing characteristics
Set.
Matching image set 301 refers to the image set of a certain pet type, such as Persian cat, pet physical trait data
Storehouse stores the image set of Persian cat.In order that data are not excessively fitted with specific individual pets, certain individual pets is doted on
Object image is less than three.
Reference pet physical trait set 302 refers to pet physical trait collection corresponding with the picture of every database storage
Close.The method that the extraction of pet physical trait can refer to embodiment two and embodiment three.
Distinguishing characteristics set 303 refers to that a certain type is used for the feature that the type pet and other types pet make a distinction
Set.The pet image for collecting the type can be passed through, the physical trait of every pet image is extracted, each physical trait is counted
Arrangement situation.In distinguishing characteristics set 303, important difference feature, general distinguishing characteristics can be distinguished.Definable in practical situations both
The pet that pet image more than 80% has is characterized as important difference feature, the pet image tool more than 60% but less than 80%
Some pet features are set to general distinguishing characteristics.In single distinguishing characteristics, multiple data areas can be set, and define it is right
Different data areas arranges different weights, for example, a certain feature is divided into into three data areas [0,1], and (1,2], (2,
3], the weight of [0,1] is set to 0.8, (1,2] weight be set to 0.6, (2,3] weight be set to 0.4.Weight is 0.8
Represent the type>80% pet image has this feature.
Further, based on pet type identification of the present invention method fourth embodiment, the invention allows for pet
5th embodiment of the method for type identification, from unlike the fourth embodiment of the method for pet type identification, step S30 is also
Including:
S31, judge pet physical trait set whether comprising distinguishing characteristics set;
If S32, comprising distinguishing characteristics set, judge whether the distinguishing characteristics set for including is more than one;
S33, if so, then by the distinguishing characteristics set for including be designated as sieve distinguishing characteristics set, using matching mechanisms to institute
State multiple screening distinguishing characteristics set to be sieved, obtain the pet type for matching;
If S34, comprising a distinguishing characteristics set, obtain pet type corresponding with the distinguishing characteristics set
Pet type mark with image set, obtains the pet type for matching.
As described in step S31, it is assumed that pet physical trait collection is combined into { a1,a2,a3…an, it is designated as set A.A certain type
Distinguishing characteristics collection be combined into { { X1, { X2, { X3, it is designated as set X, it is herein assumed that distinguishing characteristics set X only includes three differences
Feature.Another type of distinguishing characteristics set Q is { { Q1, { Q2, it is herein assumed that distinguishing characteristics set Q only includes two differences
Feature.Setting a1,a2,a3Respectively with { X1, { X2, { X3Correspondence, a4,a5Respectively with { Q1, { Q2Correspondence.If a1∈{X1, a2
∈{X2, a3∈{X3, then can determine whether that set A includes distinguishing characteristics set X.If a4∈{Q1, a5∈{Q2, then judge set A
Comprising distinguishing characteristics set Q.If the subset X in distinguishing characteristics set XnCharacter pair element a not comprising set An, then can sentence
Surely set A does not include distinguishing characteristics set X.
As described in step S32, S33, S34, this step is the processing method with regard to judging Different Results occur.If set A
Not comprising any distinguishing characteristics set, then unsuccessful, the result of output None- identified is matched.If set A is special comprising a difference
Collection is closed, then export the pet representative image of pet type corresponding with the distinguishing characteristics set.Pet representative image is clicked on, can
The corresponding matching pet image set of loading, the characteristics of facilitate user to check the Different Individual of the type pet, increases the type
The understanding of pet.If set A includes multiple distinguishing characteristics set, using screening mechanism, the result to obtaining further is sieved,
The matching degree of each distinguishing characteristics set is obtained, by the height output matching result of matching degree.
Further, based on pet type identification of the present invention method the 5th embodiment, the invention allows for pet
The sixth embodiment of the method for type identification, and unlike the 5th embodiment of the method for pet type identification, step S33
In, the employing matching mechanisms are sieved to the plurality of screening distinguishing characteristics set, obtain the pet type for matching, bag
Include:
S331, acquisition match image set with the screening of the multiple corresponding pet type of screening distinguishing characteristics set;
S332, all pet images matched with the screening of pet type each described in image set are transferred, obtained
The corresponding reference pet physical trait set of every pet image;
S333, by the spy in the feature in the pet physical trait set and the reference pet physical trait set transferred
Levy and compared one by one, calculate matching for the reference pet physical trait set that pet physical trait set is transferred with each
Degree;
S334, by according to the set of all pet physical traits and reference that each is transferred in the screening matching image set
The matching degree of pet physical trait set, calculates the feature and the reference pet body transferred in the pet physical trait set
The matching degree of characteristic set, calculates the mean value of the screening matching image set and the pet physical trait sets match degree,
Obtain the whole matching degree as pet image and pet type;
S335, using the corresponding pet type of whole matching degree highest reference pet physical trait set as matching pet
Type.
In step S331, S332, it is assumed that the corresponding pet type of screening distinguishing characteristics set is B, C, D, then transfer pet
Type B, the corresponding pet image set of C, D, and transfer the corresponding reference pet of all of pet image in B, C, D pet image set
Physical trait set.By taking pet type B as an example, there is n pet image in pet type B, then its corresponding reference pet body
Characteristic set also has n, is designated as B1, B2... Bn.Reference pet physical trait set Bt={ b1, b2... bn, t is positive integer,
Span is [1, n].
In step S333, set A respectively with B1, B2... BnRelatively, corresponding matching degree η is calculated1, η2... ηn.Then
Pet type B whole matching degree ηBCan be by solution η1, η2... ηnBe averagely worth to.The entirety of pet Type C can be obtained in the same manner
Matching degree ηCWith the whole matching degree η of pet type DD。
In step S334, S335, according to the size of whole matching degree, corresponding pet representative image is exported in order.
Whole matching degree highest pet type priority shows that click on pet representative image, can load match accordingly pet image set.
It is assumed that whole matching degree ηBHighest, user can check multiple B type pet images in pet image set B, intuitively with reality
In pet contrast, draw the judged result of oneself.
With reference to Fig. 3, Fig. 3 is the structural representation of the first embodiment of the device of pet type identification of the present invention.The present invention
A kind of device of pet type identification is proposed, including:
Acquisition module 10, for obtaining pet image;
Characteristic extracting module 20, for according to the physical trait of the pet image zooming-out pet;
Matching module 30, for multiple physical traits of the pet to be entered with the data in pet physical trait database
Row matching, obtains the pet type for matching.
First it is to obtain pet image in acquisition module 10, the image can be shot by mobile phone and be obtained, it is also possible to be passed through
Import the image on the image or load networks of interior of mobile phone.
Characteristic extracting module 20, is realized by setting up neural network model.By substantial amounts of picture training, device is made
The difference of pet and background in pet image can be distinguished, and then finds out the profile of pet.
On the basis of pet profile is identified, the profile of pet is divided into into four regions:Head, trunk, four limbs and
Tail.Because the feature of the angle problem being imaged, four limbs and tail may in the picture not significantly, in statistical nature, can root
Ignore the feature at two according to actual conditions.
The feature of identification head is mostly important, and the feature of head has specifically included ear shape, and ear color, ear is relative
In the size of face, ear location;Eye shape, eyes relative to face size, eye color, eyes are in face location;
Nose shape, nose relative to face size, nose color, nose is in the position of face;Face shape, face is relative to face
The size in portion, face color, face is in the position of face;Face relative to trunk size, face color, facial contours.
The feature of trunk mainly includes size of the trunk relative to head, the color and distribution of color of trunk.
The feature of four limbs mainly includes that leg is long, leg color, the thickness of leg.
The feature of tail mainly includes tail length, tail color and tail thickness.But in actual photographed, tail is usually not
Intactly can come across among image, thus device can intelligent screening can discriminate another characteristic.
In matching module 30, the pet data of a large amount of types are stored in pet physical trait database.A certain type
Pet also store substantial amounts of individual pet characteristic information.The statistics of database arrangement of a certain feature of a certain type pet
Situation, if this feature discreteness is larger, illustrates that this feature is not the distinguishing characteristic of the type pet, then this feature is in matching
During calculating, weight accounting is negligible or accounting is little;If this feature discreteness is little, illustrate that this feature is the differentiation of the type pet
Feature, then this feature is in matching primitives, and weight accounting is big.Specific accounting weight is according to the different and different of pet type.
As simultaneously pet X includes that distinguishing characteristics A and distinguishing characteristics B, distinguishing characteristics A belong to type-A pet, distinguishing characteristics B belongs to B types
Pet.Now, device can analyze other features of pet X and whether fall into the scope of type-A or B type pets, if only type-A
Meet and the conjunction of B type mismatch, then preferentially transmit the pet picture and title of type-A.
Output result is to include pet typonym, a certain picture of the type pet or pictures.From picture come
From in pet database, the pet database contains the pet image of character pair in pet physical trait database.From
Picture typically with identification image pet action it is identical.Output result can be single result, but be not limited to single result.Can
Judged result is sequentially output according to the height of matching degree, the high preferential display of matching degree, user is unsatisfied with changeable to the result
To the result of relatively low matching degree.
As shown in figure 4, Fig. 4 is the structural representation of the second embodiment of the device of pet type identification of the present invention.Enter one
Step, based on the first embodiment of the device of pet type identification of the present invention, the invention allows for the dress of pet type identification
The second embodiment put.From unlike the first embodiment of the device of pet type identification, acquisition module 10 also includes:
Image acquisition unit 101, for shooting pet by camera, obtains initial pet image.
Displacement record unit 102, for the equipment displacement information described in records photographing during pet image;
Image definition judging unit 103, for judging, whether displacement information is in the displacement for allowing to recognize for setting
In threshold range, when judging, displacement information is in the range of the displacement threshold value, using the initial pet image as
Pet image.
In image acquisition unit 101, user is to be directed at pet by mobile phone to take pictures, and obtains the image of pet.Device for identifying
Can read the model and image-forming condition of camera.Image-forming condition includes pixel value, the photosensitive coefficients of ISO, white balance parameter etc..Identification
Device can be optimized according to the camera model of concrete model and image-forming condition to image, make the pet view data and actual value
Closer to.
In displacement record unit 102, user sees a pet oneself admired, and mood can be more excited, handheld mobile phone
When taking pictures, hand probably due to mood is excited and can not smoothly shoot pet image.The detectable user of the device of the present embodiment
Whether hand shaking excessively causes the pet image for shooting to be not used to recognize pet type when taking pictures.
In image definition judging unit 103, specific device is obtained by three-axis gyroscope or other relevant positions
Accessory, obtains the device location information when pet image is shot in mobile phone, is recorded as front positional information is (x0, y0,
z0), the positional information after imaging is (x1, y1, z1), then can calculate its displacement information Δ X.Specially:
Judge whether institute's displacement information exceedes setting and allow in the range of the displacement threshold value of identification, wherein, shift value scope
To pre-set.If being not above, the image can be used for pet type identification, if exceeding, remind user to re-shoot.
As shown in figure 5, Fig. 5 is the structural representation of the 3rd embodiment of the device of pet type identification of the present invention.Enter one
Step, based on the first embodiment of the device of pet type identification of the present invention, the invention allows for the dress of pet type identification
The 3rd embodiment put, from unlike the first embodiment of the device of pet type identification, the characteristic extracting module 20 is wrapped
Include:
Overall profile extraction unit 21, for the overall profile for extracting pet from pet image;
Body position positioning unit 22, for positioning face, trunk, tail and four limbs from the overall profile of the pet
The band of position;
Feature extraction unit 23, for extracting pet respectively in the band of position of the face, trunk, tail and four limbs
Face feature, trunk feature, tail feature and four limbs feature, the face feature includes eye feature, face feature, nose
Feature, ear feature and face mask.
In overall profile extraction unit 21, the identification to pet profile is realized by setting up neural network model.It is logical
Substantial amounts of picture training is crossed, makes device to distinguish the difference of pet and background in pet image, and then find out the profile of pet.
Usually, for regard to recognizing pet type, identification face feature is mostly important.Using the training of neural network model
Mechanism, device for identifying may recognize that the face mask of pet.According to the color and the difference of environmental colors of pet face, week is judged
The ambient parameter in collarette border.The overall profile of pet is recognized with reference to color in the profile of the face and the ambient parameter.Than
Environment as residing for recognizing a cat, it is on meadow or on road on sheet or on floor to be.The back of the body is deducted
After scape, the overall profile of pet just can be obtained.
In some complex backgrounds, the pet overall profile obtained after background correction is not real pet wheel
Exterior feature, further comprises part background, now can retain rational outline portion by pet body contour location model.
In body position positioning unit 22, on the basis of pet profile is identified, the profile of pet is divided into into four
Region:Head, trunk, four limbs and tail.Because the feature of the angle problem being imaged, four limbs and tail may not show in the picture
Write, in statistical nature, the feature at two can be ignored according to actual conditions.
In feature extraction unit 23, the feature for recognizing head is mostly important, and the feature of head has specifically included ear shape,
Ear color, ear relative to face size, ear location etc.;Eye shape, eyes relative to face size, eyes face
Color, eyes are in face location etc.;Nose shape, nose relative to face size, nose color, nose is in the position of face
Deng;Face shape, face relative to face size, face color, face is in the position of face;Face is relative to the big of trunk
It is little, face color, facial contours etc..
The feature of trunk mainly includes size of the trunk relative to head, color and distribution of color of trunk etc..
The feature of four limbs mainly includes that leg is long, leg color, thickness of leg etc..
The feature of tail mainly includes tail length, tail color and tail thickness etc..But in actual photographed, tail is usually
Intactly can not come across among image, thus device can intelligent screening can discriminate another characteristic.
The physical trait of above pet, each feature is represented with a variable.Extract all of pet on pet image
All features are constituted a pet physical trait set by feature.But it is not that each feature can be by during extraction
Extract.Now, in the set of pet physical trait, the value for representing the variable of the feature that can not be extracted is designated as 0.
By the face feature of the pet, trunk feature, tail feature together with four limbs combinations of features, pet body is formed
Body characteristicses set, each feature includes two parameters of CF.
Further, based on pet type identification of the present invention device 3rd embodiment, the invention allows for pet
The fourth embodiment of the device of type identification, and unlike the 3rd embodiment of the device of pet type identification, the pet
Physical trait database includes:
The matching image set 301 of multiple pet types, the matching image set of each pet type includes that one or more is pre-
If pet matching image, wherein comprising one carrying pet type mark the pet matching image, be designated as representing pet
Image;
From the reference pet physical trait set 302 obtained per a pet matching image zooming-out, the reference pet
Multiple features of pet described in image are matched in physical trait set comprising the pet;
Distinguishing characteristics set 303 corresponding with the matching image set of each pet type, the distinguishing characteristics set expression
One or more distinguishing characteristics of a certain type pet, the distinguishing characteristics is to enter with other types pet for the type pet
The feature that row is distinguished.
Matching image set 301 refers to the image set of a certain pet type, such as Persian cat, pet physical trait data
Storehouse stores the image set of Persian cat.In order that data are not excessively fitted with specific individual pets, certain individual pets is doted on
Object image is less than three.
Reference pet physical trait set 302 refers to pet physical trait collection corresponding with the picture of every database storage
Close.The method that the extraction of pet physical trait can refer to embodiment 2 and embodiment 3.
Distinguishing characteristics set 303 refers to that a certain type is used for the feature that the type pet and other types pet make a distinction
Set.The pet image for collecting the type can be passed through, the physical trait of every pet image is extracted, each physical trait is counted
Arrangement situation.In distinguishing characteristics set 303, important difference feature, general distinguishing characteristics can be distinguished.Definable in practical situations both
The pet that pet image more than 80% has is characterized as important difference feature, the pet image tool more than 60% but less than 80%
Some pet features are set to general distinguishing characteristics.In single distinguishing characteristics, multiple data areas can be set, and define it is right
Different data areas arranges different weights, for example, a certain feature is divided into into three data areas [0,1], and (1,2], (2,
3], the weight of [0,1] is set to 0.8, (1,2] weight be set to 0.6, (2,3] weight be set to 0.4.Weight is 0.8 table
Show the type>80% pet image has this feature.
As shown in fig. 6, Fig. 6 is the structural representation of the 5th embodiment of the device of pet type identification of the present invention.Enter one
Step, based on the fourth embodiment of the device of pet type identification of the present invention, the invention allows for the dress of pet type identification
The 5th embodiment put, from unlike the fourth embodiment of the device of pet type identification, the matching module 30 includes:
Matching unit 31, for judging pet physical trait set whether comprising distinguishing characteristics set;
Interpretation of result unit 32, if for analyzing comprising distinguishing characteristics set, judging that the distinguishing characteristics set for including is
It is no more than one;
Many result treatment units 33, when there are multiple results for interpretation of result unit, then by the distinguishing characteristics collection for including
Conjunction is designated as sieving distinguishing characteristics set, and the plurality of screening distinguishing characteristics set is sieved using matching mechanisms, obtains
The pet type matched somebody with somebody;
Single result treatment unit 34, during for interpretation of result unit only one of which result, then obtains special with the difference
The pet type mark of the matching image set of corresponding pet type is closed in collection, obtains the pet type for matching.
In matching unit 31, it is assumed that pet physical trait collection is combined into { a1,a2,a3…an, it is designated as set A.A certain type
Distinguishing characteristics collection is combined into { { X1, { X2, { X3, it is designated as set X, it is herein assumed that it is special that distinguishing characteristics set X only includes three differences
Levy.Another type of distinguishing characteristics set Q is { { Q1, { Q2, it is herein assumed that it is special that distinguishing characteristics set Q only includes two differences
Levy.Setting a1,a2,a3Respectively with { X1, { X2, { X3Correspondence, a4,a5Respectively with { Q1, { Q2Correspondence.If a1∈{X1, a2∈
{X2, a3∈{X3, then can determine whether that set A includes distinguishing characteristics set X.If a4∈{Q1, a5∈{Q2, then judge set A bags
The Q of set containing distinguishing characteristics.If the subset X in distinguishing characteristics set XnCharacter pair element a not comprising set An, then can determine that
Set A does not include distinguishing characteristics set X.
In interpretation of result unit 32, if set A does not include any distinguishing characteristics set, the result of None- identified is exported.
In single result treatment unit 34, set A only includes a distinguishing characteristics set, then export corresponding with the distinguishing characteristics set
The pet representative image of pet type.Pet representative image is clicked on, corresponding matching pet image set can be loaded, facilitate user to look into
The characteristics of seeing the Different Individual of the type pet, increases the understanding of the pet of the type.In many result treatment units 33, set A
Comprising multiple distinguishing characteristics set, then using screening mechanism, the result to obtaining further is sieved, and obtains each distinguishing characteristics collection
The matching degree of conjunction, by the height output matching result of matching degree.
As shown in fig. 7, Fig. 7 is the structural representation of the sixth embodiment of the device of pet type identification of the present invention.Enter one
Step, based on the 5th embodiment of the device of pet type identification of the present invention, the invention allows for the dress of pet type identification
The sixth embodiment put, and unlike the 5th embodiment of the device of pet type identification, in interpretation of result unit 33
Include with mechanism:
Unit 331 is transferred, the screening for obtaining pet type corresponding with distinguishing characteristics set is sieved matches image set;
Transfer all pet images in the screening matching image set, the corresponding reference pet physical trait of every pet image of acquisition
Set;
Matching unit 332, for the feature in the pet physical trait set is special with the reference pet body transferred
Feature during collection is closed is compared one by one, calculates the reference pet body spy that pet physical trait set is transferred with each
The matching degree that collection is closed;By the ginseng transferred with each according to all pet physical traits set in the screening matching image set
Than the matching degree of pet physical trait set, the feature and the reference pet body transferred in the pet physical trait set is calculated
The matching degree of body characteristicses set, calculates the screening matching image set average with the pet physical trait sets match degree
Value, obtains the whole matching degree as pet image and pet type;
Adaptation unit 333, for the screening of whole matching degree highest to be matched into image set reference pet physical trait set
Corresponding pet type is used as matching pet type.
In transferring unit 331, it is assumed that the corresponding pet type of screening distinguishing characteristics set is B, C, D, then transfer pets
The corresponding pet image set of type B, C, D, and transfer the corresponding reference pet body of all of pet image in B, C, D pet image set
Body characteristicses set.By taking pet type B as an example, there is n pet image in pet type B, then its corresponding reference pet body is special
Collection is closed also n, is designated as B1, B2... Bn.Reference pet physical trait set Bt={ b1, b2... bn, t is positive integer, is taken
Value scope is [1, n].
In matching unit 332, set A respectively with B1, B2... BnRelatively, corresponding matching degree η is calculated1, η2... ηn.Then
Pet type B whole matching degree ηBCan be by solution η1, η2... ηnBe averagely worth to.The entirety of pet Type C can be obtained in the same manner
Matching degree ηCWith the whole matching degree η of pet type DD。
In adaptation unit 333, according to the size of whole matching degree, corresponding pet representative image is exported in order.It is overall
Matching degree highest pet type priority shows that click on pet representative image, can load match accordingly pet image set.It is assumed that
Whole matching degree ηBHighest, user can check multiple B type pet images in pet image set B, intuitively with reality in
Pet contrasts, and draws the judged result of oneself.
As shown in figure 8, Fig. 8 is the design analysis diagram of pet kind identification method of the present invention and device.It is proposed by the present invention
Pet kind identification method, is the position (i.e. the profile of pet) that pet is judged by training mechanism first, then by the wheel of pet
Exterior feature is divided into head, trunk, four limbs, and the feature at each position is extracted, in pet physical trait database in four positions of afterbody
Data matched, output matching result.
Pet kind identification method proposed by the present invention and device, in its method, intelligent terminal obtains pet image;According to
Multiple physical traits of the pet image zooming-out pet;By multiple physical traits of the pet and pet physical trait data
Data in storehouse are matched, and obtain the pet type for matching, and user can be made quickly to know the type of pet, are promoted to pet
Understand, help user to select the pet oneself liked.
It should be noted that the pet kind identification method of the present invention and pet type identifying device can be embodied as computer journey
The computer readable media of sequence product.In further, the present invention provides a kind of pet type identification readable media, wherein pets
Type identification readable media storage pet kind identification method or pet type identifying device, and pet type identification readable media tool
Have functions that to recognize image and measurement presentation content.
In other words, pet kind identification method and pet type identifying device may be embodied as various forms, for example:Firmly
Part embodiment, software implementation (comprising firmware, resident software or micro program code etc.).Additionally, pet kind identification method and doting on
Species type identifying device may be embodied as the embodiment of software and hardware, for example:Pet type identification electronic module, pet type
Identification embedded equipment, but be not limited.In actual applications, pet kind identification method of the invention and pet type are known
Other device can be embodied as computer program product by tangible media format, and wherein computer program product has a plurality of image journeys
Sequence code, and this etc. imaging application code comprising above-mentioned pet kind identification method or above-mentioned pet type identifying device.
One or more computers can use or can read the combination of media and may be by.For example, computer can make
With or can read media can be (but being not limited to) electronics, magnetic, optical, electromagnetism, ultrared or semiconductor
System, device, equipment or communications media.More specifically computer readable media embodiment can include following shown (non-limiting
Illustration):The electrical connection that is made up of one or more connecting lines, portable computer magnetic sheet, Hard Disk Drive, arbitrary access are deposited
Reservoir (RAM), read-only storage (ROM), can erasing procedure read-only storage (EPROM or flash memory), optical fiber, can take
Formula CD-RW discsCD-RW (CD-ROM), optical disc drive, transmission media (such as world-wide web (Internet) or internal network
(intranet) basis connection) or magnetic storage device.It is noted that computer can be used or can read media can also be
Paper or it is any can be used for by program print thereon and allow the program once again by electronization appropriate media, for example lead to
The optical scanner paper or other media are crossed, then compiled again, interpreted or other suitable necessity processing modes, then can be once again
In being stored in computer memory.Herein, computer can use or can read media can be it is any for keeping, storing,
The media of transmission, propagation or transmission procedure code, are processed for coupled instruction execution system, device or equipment.Electricity
Brain can may include that wherein storing computer can use the propagation data signal of procedure code using media, whether with fundamental frequency
(baseband) or partial carrier wave kenel.Computer can use the transmission of procedure code to use any fit media, bag
Include wireless, wired (but being not limited to), Connectorized fiber optic cabling, radio frequency (RF) etc..
Computer program code for performing present invention operation can be write using the combination of one or more program language,
Including OO program language (such as Java, Smalltalk, C++ or other fellows) and traditional program language is (for example
C programming language or other similar program languages).Procedure code can be complete in the form of stand alone software external member the electricity in user
Perform on brain or part performs on the computer of user, or partly in user's computer partly in far-end computer, Huo Zhequan
Portion performs on far-end computer or server.In situation below, far-end computer can be with any type of network connection to using
The computer of person, including Local Area Network (LAN) or Wide Area Network (WAN), or be possible with external computer and link (such as profit doing
World-wide web is connected to Internet service provider).
The related narration of the present invention can be with reference to the system according to the specific embodiment of the invention, device, method and computer program
The flow chart and/or block diagram of product is illustrating.Each side in each flow chart and/or block diagram is appreciated that
Any combinations of square in block, and flow graph and/or block diagram, it is possible to use computer program instructions are implementing.These computer journeys
Sequence instruction is available for the machine that the processor or other programmable data processing equipments of universal computer or special computer are constituted
To perform, and instruct and processed so as to implementing procedure figure and/or block diagram via computer or other programmable data processing equipments
In illustrated function or operation.
These computer program instructions also can be stored on computer readable media, so as to indicate computer or other can program
Change data processing equipment to carry out specific function, and the instruction that these are stored on computer readable media constitutes one and makes
Product, function or operation that the instruction included in it can be illustrated in implementing procedure figure and/or block diagram.
Computer program instructions also can be loaded on computer or other programmable data processing equipments, so as in computer or
A system operation procedures are carried out on other programmable devices, and the instruction is performed on the computer or other programmable devices
When produce computer implementation procedure with function illustrated in reaching flow chart and/or block diagram or operation.
Show that the system according to various embodiments of the invention, device, method and computer program product can be implemented in the drawings
Framework, function and operation flow chart and block diagram.Therefore, each square in flow chart or block diagram can represent a mould
Block, section or partial procedure code, it includes one or more executable instructions, with the logic function for implementing to specify.Separately when note
Meaning, in some other embodiments, the function described in square can disobey the order shown in figure and carry out.For example,
Two illustrate the square that is connected and in fact while perform, or can also in some cases can also according to the function of being involved
Perform according to contrary order is illustrated.In addition do not note too that while person, the square of each block diagram and/or flow chart, and block diagram and/
Or in flow chart square combination, can be by being implemented based on the system of specific purposes hardware, or by specific purposes hardware
Specific function or operation are performed with the combination of computer instruction.
The preferred embodiments of the present invention are the foregoing is only, the scope of the claims of the present invention, every utilization is not thereby limited
Equivalent structure or equivalent flow conversion that description of the invention and accompanying drawing content are made, or directly or indirectly it is used in other correlations
Technical field, be included within the scope of the present invention.
Claims (10)
1. a kind of method of pet type identification, it is characterised in that include:
Obtain pet image;
According to multiple physical traits of the pet image zooming-out pet;
Multiple physical traits of the pet are matched with the data in pet physical trait database, doting on for matching is obtained
Species type.
2. the method for pet type identification according to claim 1, it is characterised in that it is described by camera shoot described in dote on
Object image, including:
Pet is shot by camera, initial pet image is obtained;
Described in records photographing during initial pet image, the displacement information of camera;
Judge whether institute's displacement information is allowed in the range of the displacement threshold value of identification what is set, the displacement information judging
When in the range of the displacement threshold value, using the initial pet image as pet image.
3. the method for pet type identification according to claim 1, it is characterised in that described to be carried according to the pet image
Obtain favour multiple physical traits of thing, including:
The overall profile of pet is extracted from pet image;
The band of position of face, trunk, tail and four limbs is positioned from the overall profile of the pet;
Extract face feature, trunk feature, the tail of pet respectively in the band of position of the face, trunk, tail and four limbs
Bart seeks peace four limbs feature, and the face feature includes eye feature, face feature, nose feature, ear feature and face
Profile;
By the face feature of the pet, trunk feature, tail feature together with four limbs combinations of features, pet body is formed special
Collection is closed, and each feature includes two parameters of CF.
4. the method for pet type identification according to claim 3, it is characterised in that the pet physical trait database
In data include:
The matching image set of multiple pet types, the matching image set of each pet type includes one or more default pet
Matching image, wherein the pet matching image comprising a carrying pet type mark, is designated as representing pet image;
From the reference pet physical trait set obtained per a pet matching image zooming-out, the reference pet physical trait
Multiple features of pet described in image are matched in set comprising the pet;
Distinguishing characteristics set corresponding with the matching image set of each pet type, a certain type of the distinguishing characteristics set expression
One or more distinguishing characteristics of pet, the distinguishing characteristics makes a distinction for the type pet with other types pet
Feature.
5. the method for pet type identification according to claim 4, it is characterised in that described by the pet physical trait
Set is matched with the data in pet physical trait database, and the pet type for obtaining matching includes:
Judge pet physical trait set whether comprising distinguishing characteristics set;
If comprising distinguishing characteristics set, judging whether the distinguishing characteristics set for including is more than one;
If so, then by the distinguishing characteristics set for including be designated as sieve distinguishing characteristics set, using matching mechanisms to the plurality of sieve
The other characteristic set of subregion is sieved, and obtains the pet type for matching;
If comprising a distinguishing characteristics set, obtaining the matching image set of pet type corresponding with the distinguishing characteristics set
Pet type mark, obtain match pet type.
6. the method for pet type identification according to claim 5, it is characterised in that the employing matching mechanisms are to described
Multiple screening distinguishing characteristics set are sieved, and obtain the pet type for matching, including:
Obtain the screening matching image set of pet type corresponding with screening distinguishing characteristics set;
Transfer all pet images in the screening matching image set, the corresponding reference pet body of every pet image of acquisition
Characteristic set;
By the feature in the feature in the pet physical trait set and the reference pet physical trait set transferred carry out by
One compares, and calculates the matching degree of the reference pet physical trait set that pet physical trait set is transferred with each;
According to the matching degree of all of reference pet physical trait set in the screening matching image set, the screening is calculated
Mean value with image set with the pet physical trait sets match degree, as pet image and the whole matching of pet type
Degree;
Using the corresponding pet type of whole matching degree highest screening matching image set as matching pet type.
7. a kind of device of pet type identification, it is characterised in that include:
Acquisition module, for obtaining pet image;
Characteristic extracting module, for according to the physical trait of the pet image zooming-out pet;
A matching module, for the data in multiple physical traits of the pet and pet physical trait database to be carried out
Match somebody with somebody, obtain the pet type for matching.
8. the device of pet type identification according to claim 7, it is characterised in that the characteristic extracting module includes:
Overall profile extraction unit, for the overall profile for extracting pet from pet image;
Body position positioning unit, for the position for positioning face, trunk, tail and four limbs from the overall profile of the pet
Region;
Feature extraction unit, for the face for extracting pet respectively in the band of position of the face, trunk, tail and four limbs
Feature, trunk feature, tail feature and four limbs feature, the face feature include eye feature, face feature, nose feature,
Ear feature and face mask;
By the face feature of the pet, trunk feature, tail feature together with four limbs combinations of features, pet body is formed special
Collection is closed, and each feature includes two parameters of CF.
9. the device of pet type identification according to claim 8, it is characterised in that the pet physical trait database
Including:
The matching image set of multiple pet types, the matching image set of each pet type includes one or more default pet
Matching image, wherein the pet matching image comprising a carrying pet type mark, is designated as representing pet image;
From the reference pet physical trait set obtained per a pet matching image zooming-out, the reference pet physical trait
Multiple features of pet described in image are matched in set comprising the pet;
Distinguishing characteristics set corresponding with the matching image set of each pet type, a certain type of the distinguishing characteristics set expression
One or more distinguishing characteristics of pet, the distinguishing characteristics makes a distinction for the type pet with other types pet
Feature.
10. the device of pet type identification according to claim 9, it is characterised in that the matching module includes:
Matching unit, for judging pet physical trait set whether comprising distinguishing characteristics set;
Interpretation of result unit, if for analyzing comprising distinguishing characteristics set, judging whether the distinguishing characteristics set for including is more than
One;
Many result treatment units, when there are multiple results for interpretation of result unit, are then designated as the distinguishing characteristics set for including
Screening distinguishing characteristics set, is sieved using matching mechanisms to the plurality of screening distinguishing characteristics set, obtains doting on for matching
Species type;
Single result treatment unit, during for interpretation of result unit only one of which result, then obtains and the distinguishing characteristics set
The pet type mark of the matching image set of corresponding pet type, obtains the pet type for matching.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611032333.8A CN106577350B (en) | 2016-11-22 | 2016-11-22 | Pet type identification method and device |
PCT/CN2017/074840 WO2018094892A1 (en) | 2016-11-22 | 2017-02-24 | Pet type recognition method and device, and terminal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611032333.8A CN106577350B (en) | 2016-11-22 | 2016-11-22 | Pet type identification method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106577350A true CN106577350A (en) | 2017-04-26 |
CN106577350B CN106577350B (en) | 2020-10-09 |
Family
ID=58591589
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611032333.8A Active CN106577350B (en) | 2016-11-22 | 2016-11-22 | Pet type identification method and device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN106577350B (en) |
WO (1) | WO2018094892A1 (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108875564A (en) * | 2018-05-02 | 2018-11-23 | 浙江工业大学 | A kind of pet face recognition method |
CN109274891A (en) * | 2018-11-07 | 2019-01-25 | 北京旷视科技有限公司 | A kind of image processing method, device and its storage medium |
CN109409319A (en) * | 2018-11-07 | 2019-03-01 | 北京旷视科技有限公司 | A kind of pet image beautification method, device and its storage medium |
CN110704646A (en) * | 2019-10-16 | 2020-01-17 | 支付宝(杭州)信息技术有限公司 | Method and device for establishing stored material file |
CN112101070A (en) * | 2019-06-18 | 2020-12-18 | 财团法人农业科技研究院 | Animal identity recognition system and method for improving recognition rate by nasal print |
WO2021212760A1 (en) * | 2020-04-24 | 2021-10-28 | 北京旷视科技有限公司 | Method and apparatus for determining identity type of person, and electronic system |
CN113657318A (en) * | 2021-08-23 | 2021-11-16 | 平安科技(深圳)有限公司 | Pet classification method, device, equipment and storage medium based on artificial intelligence |
WO2022206291A1 (en) * | 2021-03-29 | 2022-10-06 | 青岛海尔空调器有限总公司 | Air conditioner control method and apparatus, and device and storage medium |
CN115546838A (en) * | 2022-10-20 | 2022-12-30 | 星宠王国(北京)科技有限公司 | Wedding system and method based on dog face image recognition technology |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP7377876B2 (en) | 2019-02-01 | 2023-11-10 | ソシエテ・デ・プロデュイ・ネスレ・エス・アー | Apparatus and method for recommending pet food |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5134277A (en) * | 1983-11-07 | 1992-07-28 | Australian Meat And Live-Stock Corporation | Remote data transfer system with ambient light insensitive circuitry |
EP0821912A2 (en) * | 1996-07-25 | 1998-02-04 | Oki Electric Industry Co., Ltd. | Animal body identifying device and body identifying system |
CN101309357A (en) * | 2007-05-18 | 2008-11-19 | 卡西欧计算机株式会社 | Image pickup apparatus equipped with function of detecting image shaking |
CN201409150Y (en) * | 2008-12-19 | 2010-02-17 | 康佳集团股份有限公司 | Mobile phone capable of identifying pet |
CN102523380A (en) * | 2011-11-10 | 2012-06-27 | 深圳市同洲电子股份有限公司 | Method for preventing camera of mobile terminal from shaking, and mobile terminal |
JP2012129933A (en) * | 2010-12-17 | 2012-07-05 | Casio Comput Co Ltd | Imaging apparatus and program |
CN102594857A (en) * | 2010-10-11 | 2012-07-18 | 微软公司 | Image identification and sharing on mobile devices |
CN103124311A (en) * | 2013-01-24 | 2013-05-29 | 广东欧珀移动通信有限公司 | Mobile phone photographing method |
WO2014018531A1 (en) * | 2012-07-23 | 2014-01-30 | Clicrweight, LLC | Body condition score determination for an animal |
CN105744164A (en) * | 2016-02-24 | 2016-07-06 | 惠州Tcl移动通信有限公司 | Photographing method and system for mobile terminal |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101320319B (en) * | 2008-05-14 | 2010-06-09 | 山东大学 | Apparatus for timely recognizing aquatic animal and working method thereof |
US9104906B2 (en) * | 2011-05-09 | 2015-08-11 | Catherine Grace McVey | Image analysis for determining characteristics of animals |
CN104573745B (en) * | 2015-01-21 | 2017-10-03 | 中国计量学院 | Fruit-fly classified method based on magnetic resonance imaging |
CN105954281B (en) * | 2016-04-21 | 2018-07-27 | 南京农业大学 | A kind of paddy goes mouldy the method for fungus colony non-damage drive |
-
2016
- 2016-11-22 CN CN201611032333.8A patent/CN106577350B/en active Active
-
2017
- 2017-02-24 WO PCT/CN2017/074840 patent/WO2018094892A1/en active Application Filing
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5134277A (en) * | 1983-11-07 | 1992-07-28 | Australian Meat And Live-Stock Corporation | Remote data transfer system with ambient light insensitive circuitry |
EP0821912A2 (en) * | 1996-07-25 | 1998-02-04 | Oki Electric Industry Co., Ltd. | Animal body identifying device and body identifying system |
CN101309357A (en) * | 2007-05-18 | 2008-11-19 | 卡西欧计算机株式会社 | Image pickup apparatus equipped with function of detecting image shaking |
CN201409150Y (en) * | 2008-12-19 | 2010-02-17 | 康佳集团股份有限公司 | Mobile phone capable of identifying pet |
CN102594857A (en) * | 2010-10-11 | 2012-07-18 | 微软公司 | Image identification and sharing on mobile devices |
JP2012129933A (en) * | 2010-12-17 | 2012-07-05 | Casio Comput Co Ltd | Imaging apparatus and program |
CN102523380A (en) * | 2011-11-10 | 2012-06-27 | 深圳市同洲电子股份有限公司 | Method for preventing camera of mobile terminal from shaking, and mobile terminal |
WO2014018531A1 (en) * | 2012-07-23 | 2014-01-30 | Clicrweight, LLC | Body condition score determination for an animal |
CN103124311A (en) * | 2013-01-24 | 2013-05-29 | 广东欧珀移动通信有限公司 | Mobile phone photographing method |
CN105744164A (en) * | 2016-02-24 | 2016-07-06 | 惠州Tcl移动通信有限公司 | Photographing method and system for mobile terminal |
Non-Patent Citations (6)
Title |
---|
关锦文: "《移动互联网技术应用基础》", 31 January 2015, 华南理工大学出版社 * |
刘子山: "人脸信息识别技术研究及在虚拟宠物上的应用", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
曾陈颖: "面向珍稀野生动物保护的图像监测与识别技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
杨震 等: "《物联网的技术体系》", 30 September 2013, 北京邮电大学出版社 * |
谢素仪: "宠物猫脸检测的方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
高旭 等: "基于图像识别的濒危动物信息采集系统设计", 《电子世界》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108875564A (en) * | 2018-05-02 | 2018-11-23 | 浙江工业大学 | A kind of pet face recognition method |
CN109274891A (en) * | 2018-11-07 | 2019-01-25 | 北京旷视科技有限公司 | A kind of image processing method, device and its storage medium |
CN109409319A (en) * | 2018-11-07 | 2019-03-01 | 北京旷视科技有限公司 | A kind of pet image beautification method, device and its storage medium |
CN109274891B (en) * | 2018-11-07 | 2021-06-22 | 北京旷视科技有限公司 | Image processing method, device and storage medium thereof |
CN112101070A (en) * | 2019-06-18 | 2020-12-18 | 财团法人农业科技研究院 | Animal identity recognition system and method for improving recognition rate by nasal print |
CN110704646A (en) * | 2019-10-16 | 2020-01-17 | 支付宝(杭州)信息技术有限公司 | Method and device for establishing stored material file |
WO2021212760A1 (en) * | 2020-04-24 | 2021-10-28 | 北京旷视科技有限公司 | Method and apparatus for determining identity type of person, and electronic system |
WO2022206291A1 (en) * | 2021-03-29 | 2022-10-06 | 青岛海尔空调器有限总公司 | Air conditioner control method and apparatus, and device and storage medium |
CN113657318A (en) * | 2021-08-23 | 2021-11-16 | 平安科技(深圳)有限公司 | Pet classification method, device, equipment and storage medium based on artificial intelligence |
CN113657318B (en) * | 2021-08-23 | 2024-05-07 | 平安科技(深圳)有限公司 | Pet classification method, device, equipment and storage medium based on artificial intelligence |
CN115546838A (en) * | 2022-10-20 | 2022-12-30 | 星宠王国(北京)科技有限公司 | Wedding system and method based on dog face image recognition technology |
CN115546838B (en) * | 2022-10-20 | 2023-11-24 | 星宠王国(北京)科技有限公司 | Wedding system and method based on dog face image recognition technology |
Also Published As
Publication number | Publication date |
---|---|
CN106577350B (en) | 2020-10-09 |
WO2018094892A1 (en) | 2018-05-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106577350A (en) | Method and device for recognizing pet type | |
CN110348387B (en) | Image data processing method, device and computer readable storage medium | |
US20210241607A1 (en) | Methods and systems for content processing | |
CN108764372B (en) | Construction method and device, mobile terminal, the readable storage medium storing program for executing of data set | |
KR102102161B1 (en) | Method, apparatus and computer program for extracting representative feature of object in image | |
US9918183B2 (en) | Methods and systems for content processing | |
US9271133B2 (en) | Methods and systems for image or audio recognition processing | |
CA2734613C (en) | Methods and systems for content processing | |
CN110222718A (en) | The method and device of image procossing | |
CN111753697B (en) | Intelligent pet management system and management method thereof | |
CN110826371A (en) | Animal identification method, device, medium and electronic equipment | |
JPWO2021084590A5 (en) | ||
CN114596584A (en) | Intelligent detection and identification method for marine organisms | |
CN117079195B (en) | Wild animal identification method and system based on image video | |
CN108270958A (en) | Camera module Adaptable System and its adaptive approach | |
CN112507770B (en) | Rice disease and insect pest identification method and system | |
JP5351084B2 (en) | Image recognition apparatus and image recognition method | |
CN108664921B (en) | Image identification method and system based on bag-of-words model under android platform | |
CN117133014A (en) | Live pig face key point detection method | |
Hitelman et al. | The effect of age on young sheep biometric identification | |
CN110489634A (en) | A kind of build information recommended method, device, system and terminal device | |
CN115114467B (en) | Training method and device for picture neural network model | |
CN110620877B (en) | Position information generation method, device, terminal and computer readable storage medium | |
CN114511877A (en) | Behavior recognition method and device, storage medium and terminal | |
CN110796197A (en) | Fish seed identification method and system |
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 | ||
TR01 | Transfer of patent right |
Effective date of registration: 20210915 Address after: 518000 201, No.26, yifenghua Innovation Industrial Park, Xinshi community, Dalang street, Longhua District, Shenzhen City, Guangdong Province Patentee after: Shenzhen waterward Information Co.,Ltd. Address before: 5 / F, block B, huayuancheng digital building, 1079 Nanhai Avenue, Nanshan District, Shenzhen City, Guangdong Province Patentee before: SHENZHEN WATER WORLD Co.,Ltd. |
|
TR01 | Transfer of patent right |