CN109858318A - The classification recognition methods of landscape image and device - Google Patents
The classification recognition methods of landscape image and device Download PDFInfo
- Publication number
- CN109858318A CN109858318A CN201811369125.6A CN201811369125A CN109858318A CN 109858318 A CN109858318 A CN 109858318A CN 201811369125 A CN201811369125 A CN 201811369125A CN 109858318 A CN109858318 A CN 109858318A
- Authority
- CN
- China
- Prior art keywords
- image
- landscape
- landscape image
- classification
- attribute
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The present invention provides classification recognition methods and the device of a kind of landscape image, this method comprises: obtain include target image and background landscape image, target image is partitioned into from the landscape image, as segmented image；By in the class prediction model of segmented image input built in advance, category attribute is obtained；In the convolutional neural networks structural model that target image input is pre-established, the convolutional neural networks structural model is used to characterize the incidence relation of segmented image and category attribute；Obtain the corresponding characteristic attribute of the category attribute；Using preset landscape classifying rules, the classification of the landscape image is confirmed according to the characteristic attribute.The classification recognition methods of landscape image of the present invention realizes the quick of landscape image and accurately identifies.
Description
Technical field
The present invention relates to the classification recognition methods of field of image processing more particularly to a kind of landscape image and devices.
Background technique
With the continuous development of internet, tourist attractions picture all over the world is shared on network, with various shapes
The landscape image of shape and color color is covered the sky and the earth, and when for some nameless buildings or sight spot, people do not know about it usually
Source, such as location message.
Therefore, it is necessary to a kind of classification recognition methods of landscape image be provided, to realize the quick identification of landscape image.
Summary of the invention
To solve the above problems, especially the inconvenient problem of landscape image identification, the present invention use as follows in the prior art
The technical solution of various aspects:
In a first aspect, the present invention provides a kind of classification recognition methods of landscape image, comprising:
The landscape image comprising target image and background is obtained, is partitioned into target image from the landscape image, as
Segmented image；
By in the class prediction model of segmented image input built in advance, category attribute is obtained；Target image input is pre-
In the convolutional neural networks structural model first established, the convolutional neural networks structural model is for characterizing segmented image and classification
The incidence relation of attribute；
Obtain the corresponding characteristic attribute of the category attribute；Using preset landscape classifying rules, according to the feature category
Property confirms the classification of the landscape image.
Preferably, the class prediction model generates by the following method:
Obtain several sample landscape images and the corresponding classification category of the target image comprising target image and background
Property；
It is partitioned into target image from the sample landscape image, as segmented image；
Using segmented image and category attribute, convolutional neural networks training is inputted, class prediction model is obtained.
Preferably, using preset landscape classifying rules, the classification of the landscape image is confirmed according to the characteristic attribute,
Include:
Using the characteristic attribute, the number of the corresponding relationship of the classification of preset characteristic feature attribute and landscape image is inquired
According to library, the classification of the landscape image is obtained.
Preferably, it is partitioned into target image from the landscape image, as segmented image, comprising:
Feature extraction is carried out to the landscape image by semantic segmentation model, identifies at least one in the landscape image
Region, and generate target image；
The semantic segmentation model carries out sample training, trained semantic segmentation model by mark characteristic area in advance
For generating the target image comprising at least one region.
It preferably, will be in the class prediction model of segmented image input built in advance, comprising:
The maximum segmented image of image scaled is chosen as the first segmented image, first segmented image is inputted into built in advance
Class prediction model in；
If being more than present count according to the corresponding category result quantity of the landscape image that first segmented image obtains
Amount；
Remaining segmented image is sequentially input in the class prediction model of built in advance according to preset recognition sequence then and is carried out
Identification；The recognition sequence presets the influence degree of recognition result according to the attribute of segmented image.
Preferably, after the classification that the landscape image is confirmed according to the characteristic attribute, further includes:
The corresponding summary info of the landscape image is generated according to the classification of the landscape image.
Preferably, the landscape image is street view image, generates the landscape image according to the classification of the landscape image
Corresponding summary info, comprising:
The corresponding location information of the street view image is generated according to the classification of the street view image；
Route guidance information is formed according to the positional information.
Second aspect, the present invention provide a kind of classification identification device of landscape image, comprising:
Segmentation module is partitioned into from the landscape image for obtaining the landscape image comprising target image and background
Target image, as segmented image；
Class prediction module, for obtaining category attribute in the class prediction model of segmented image input built in advance；
In the convolutional neural networks structural model that target image input is pre-established, the convolutional neural networks structural model is used for table
Levy the incidence relation of segmented image and category attribute；
Module is obtained, for obtaining the corresponding characteristic attribute of the category attribute；
Confirmation module confirms the landscape image according to the characteristic attribute for utilizing preset landscape classifying rules
Classification.
The third aspect, the present invention provide a kind of computer equipment, including storage medium and processor, in the storage medium
It is stored with computer-readable instruction, when the computer-readable instruction is executed by the processor, so that the processor executes
As described in above-mentioned first aspect the step of the classification recognition methods of landscape image.
Fourth aspect, the present invention provide a kind of computer readable storage medium, the computer-readable instruction by one or
When multiple processors execute, so that the classification that one or more processors execute the landscape image as described in above-mentioned first aspect is known
The step of other method.
Compared with the existing technology, technical solution of the present invention at least has following advantage:
1. the classification recognition methods of landscape image of the invention is obtained by training to identify corresponding to segmented image
The class prediction model of category attribute, by tentatively identifying the corresponding category attribute of landscape image, based on trained identification
Model identifies the corresponding target image of the category attribute, and establishes the class of characteristic feature attribute and landscape image simultaneously
The category database of other corresponding relationship realizes and carries out gradually fine identification to landscape image, guarantees final identification knot
The accuracy of fruit.Class prediction model, identification model etc. pass through deep learning method, additionally it is possible to recognition result is constantly improve, it is real
Show the quick of landscape image and accurately identifies.
2. the classification recognition methods of landscape image of the invention, after identifying the classification of landscape image, additionally it is possible to generate
The corresponding summary info of the landscape image, particularly, additionally it is possible to generate the corresponding location information of street view image, and according to described
Location information forms route guidance information, to guide user to reach the correspondence location of landscape image, without being tied according to identification
The manually opened application program with navigation feature of fruit, to realize quickly navigation.
The additional aspect of the present invention and advantage will be set forth in part in the description, these will become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
Fig. 1 is a kind of embodiment flow diagram of classification recognition methods of landscape image of the present invention；
A kind of embodiment flow diagram of the classification identification device of Fig. 2 landscape image of the present invention；
Fig. 3 is the internal structure block diagram of computer equipment in one embodiment of the invention.
The object of the invention is realized, the embodiments will be further described with reference to the accompanying drawings for functional characteristics and advantage.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described.
In some processes of the description in description and claims of this specification and above-mentioned attached drawing, contain according to
Multiple operations that particular order occurs, but it should be clearly understood that these operations can not be what appears in this article suitable according to its
Sequence is executed or is executed parallel, and the serial number of operation such as S11, S12 etc. be only used for distinguishing each different operation, serial number
It itself does not represent and any executes sequence.In addition, these processes may include more or fewer operations, and these operations can
To execute or execute parallel in order.It should be noted that the description such as " first " herein, " second ", is for distinguishing not
Same message, equipment, module etc., does not represent sequencing, does not also limit " first " and " second " and be different type.
It will appreciated by the skilled person that unless expressly stated, singular " one " used herein, " one
It is a ", " described " and "the" may also comprise plural form.It is to be further understood that being arranged used in specification of the invention
Diction " comprising " refer to that there are the feature, integer, step, operation, element and/or component, but it is not excluded that in the presence of or addition
Other one or more features, integer, step, operation, element, component and/or their group.It should be understood that when we claim member
Part is " connected " or when " coupled " to another element, it can be directly connected or coupled to other elements, or there may also be
Intermediary element.In addition, " connection " used herein or " coupling " may include being wirelessly connected or wirelessly coupling.It is used herein to arrange
Diction "and/or" includes one or more associated wholes for listing item or any cell and all combinations.
It will appreciated by the skilled person that unless otherwise defined, all terms used herein (including technology art
Language and scientific term), there is meaning identical with the general understanding of those of ordinary skill in fields of the present invention.Should also
Understand, those terms such as defined in the general dictionary, it should be understood that have in the context of the prior art
The consistent meaning of meaning, and unless idealization or meaning too formal otherwise will not be used by specific definitions as here
To explain.
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description in which the same or similar labels are throughly indicated same or similar element or has same or like function
Element.Obviously, described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, those skilled in the art's every other implementation obtained without creative efforts
Example, shall fall within the protection scope of the present invention.
Referring to Fig. 1, a kind of classification recognition methods of landscape image provided by the present invention, wherein a kind of specific reality
It applies in mode, includes the following steps:
S11, the landscape image comprising target image and background is obtained, is partitioned into target image from the landscape image,
As segmented image；
Terminal or server, which can be extracted directly, is stored in local landscape image, and it is logical also to can receive other electronic equipments
Cross the landscape image of wired connection mode or radio connection transmission.Wherein, in landscape image include target image and background,
Target image can be the element image of mark action, such as landmark, road, the text information with labeled
Nameplate, guideboard, bus stop board etc..
Preferably, it is partitioned into target image from the landscape image, as segmented image, comprising:
Feature extraction is carried out to the landscape image by semantic segmentation model, identifies at least one in the landscape image
Region, and generate target image；
The semantic segmentation model carries out sample training, trained semantic segmentation model by mark characteristic area in advance
For generating the target image comprising at least one region.
It is preferably based on convolutional neural networks unit (CNN), constructs a U-shaped neural network structure model, and using prison
Educational inspector's learning and training method, to realize detection and the building semantic segmentation model of target image.
In addition, landscape image is by taking street view image as an example, by semantic segmentation model, such as ENet, the streetscape figure that will test
The target images such as building, road, nameplate as in extract, and are respectively formed the area comprising building, road, nameplate
Domain, as segmented image.The sample training ENet for having marked street view image region is first passed through in advance, when performing image processing, benefit
The target image comprising at least one region is generated with trained ENet.
S12, the segmented image is inputted in the class prediction model of built in advance, obtains category attribute；Target image is defeated
Enter in the convolutional neural networks structural model pre-established, the convolutional neural networks structural model for characterize segmented image with
The incidence relation of category attribute.
Segmented image is identified based on neural network, can correspond to and obtain the corresponding category attribute of the segmented image.
By taking street view image as an example, the category attribute of street view image refers to streetscape type, such as the type of building, road, specifically,
Building includes the different building types such as school, mansion, church, villa.Road includes highway, drive, country lane etc.
Road type.
Preferably, the class prediction model generates by the following method:
Obtain several sample landscape images and the corresponding classification category of the target image comprising target image and background
Property；
It is partitioned into target image from the sample landscape image, as segmented image；
Using segmented image and category attribute, convolutional neural networks training is inputted, class prediction model is obtained.
Wherein, sample landscape image derives from landscape image disclosed in network and society, will by using crowdsourcing mode
The object region of building, road, nameplate in landscape image etc. is identified, while by the category attribute of target image
It is identified.Target image is inputted into training in deep layer convolutional neural networks model, and is based on the corresponding classification of the target image
Attribute completes supervised learning training, realizes the class prediction of segmented image.Preferably, by segmented image input built in advance
In class prediction model, comprising:
The maximum segmented image of image scaled is chosen as the first segmented image, first segmented image is inputted into built in advance
Class prediction model in；
If being more than present count according to the corresponding category result quantity of the landscape image that first segmented image obtains
Amount；
Remaining segmented image is sequentially input in the class prediction model of built in advance according to preset recognition sequence then and is carried out
Identification；The recognition sequence presets the influence degree of recognition result according to the attribute of segmented image.For example, segmented
To building, road and nameplate totally three independent segmented images, wherein the image scaled built is maximum, then building is made first
It is identified for the first segmented image, obtains the corresponding category result quantity of the landscape image when the first segmented image is identified
More than preset quantity, that is, when being unable to judge accurately the corresponding classification of the landscape image, by remaining road and nameplate this two
A segmented image sequentially inputs in the class prediction model of built in advance according to preset recognition sequence and is identified.Wherein, it identifies suitable
Sequence presets the influence degree of recognition result according to the attribute of different segmented images, for example, eye-catching when containing in direction board
Words identification, pattern identification when, be beneficial to the accurate judgement of recognition result, then recognition sequence be set as nameplate, road
Road.
By in the class prediction model after segmented image input training, the corresponding classification category of landscape image can be obtained
Property, thus tentatively identify the corresponding type of landscape image, it can be in specific category attribute based on the category attribute after identification
It is further identified in range, greatly reduces the range of landscape image identification, be conducive to the subsequent accurate knowledge of landscape image
Not.
S13, the corresponding characteristic attribute of the category attribute is obtained；Using preset landscape classifying rules, according to the spy
Sign attribute confirms the classification of the landscape image.
Characteristic attribute refers to the features such as shape corresponding to the segmented image of attribute of all categories, text, pattern, color.To build
For building object, the characteristic attribute of building includes that monnolithic case, color, floor quantity etc. can also include further building
Clock, statue in additional structural features on object, such as building.Specifically, based on trained identification model to the class
The corresponding target image of other attribute is identified, the corresponding characteristic attribute of the category attribute is obtained.For example, after for segmentation
Building A is identified that obtain the corresponding characteristic attribute of building A, characteristic attribute includes shape a, pattern b and color c.
Preferably, using preset landscape classifying rules, the classification of the landscape image is confirmed according to the characteristic attribute,
Include:
Using the characteristic attribute, the class of the corresponding relationship of the classification of preset characteristic feature attribute and landscape image is inquired
Other database, obtains the classification of the landscape image.
Specifically, the category database of the corresponding relationship of the classification of the characteristic feature attribute and landscape image, is to be directed to
The characteristic attribute set that landscape image of all categories is established, wherein each landscape image corresponds to one or more feature category
Property.For example, the characteristic attribute of " big ben " includes shape a, pattern b, color c and structure d.
After shape a, pattern b and color c after being identified based on identification model, preset classification number is further inquired
According in library, the landscape image comprising shape a, pattern b and color c obtains the classification of landscape image to confirm.
Specifically, it if being confirmed to obtain the category result of multiple landscape images according to the characteristic attribute, shows respectively more
True picture corresponding to the classification of a landscape image.Preferably, at least show that the classification of the landscape image is corresponding
Details picture and whole picture further can also show the picture of the corresponding different angle of the classification of the landscape image.
For example, there are three types of as a result, then showing that three kinds of results are corresponding respectively for the landscape image comprising shape a, pattern b and color c
Landscape image.
Preferably, after the classification that the landscape image is confirmed according to the characteristic attribute, further includes:
The corresponding summary info of the landscape image is generated according to the classification of the landscape image.
Wherein, summary info may include building title, road name, sight name etc., also may include landscape image
Brief introduction, location information etc..For example, the landscape image that characteristic attribute includes shape a, pattern b and color c has " big ben ", then
Generate the summary info comprising " big ben ".
Preferably, the landscape image is street view image, generates the landscape image according to the classification of the landscape image
Corresponding summary info, comprising:
The corresponding location information of the street view image is generated according to the classification of the street view image；
Route guidance information is formed according to the positional information.
Specifically, after generating the summary info comprising " big ben ", the position of " big ben " can further be generated
Information.
By training obtain to identify category attribute corresponding to segmented image class prediction model, by tentatively knowing
Not Chu the corresponding category attribute of landscape image, based on trained identification model to the corresponding target image of the category attribute into
Row identification, and establishes the category database of the corresponding relationship of the classification of characteristic feature attribute and landscape image simultaneously, realizes pair
Landscape image carries out gradually fine identification, guarantees the accuracy of final recognition result.For example, by category attribute range
Interior progress characteristic attribute identification, can be greatly improved the recognition efficiency of landscape image, while guaranteeing the identification knot of landscape image
The accuracy of fruit.In addition, class prediction model, identification model etc. pass through deep learning method, additionally it is possible to constantly improve identification knot
Fruit realizes the quick of landscape image and accurately identifies.
It, can be according to the class of street view image after the classification for identifying street view image when landscape image is street view image
The mapping relations of location information not corresponding with the street view image generate the corresponding location information of street view image, to guide user
The correspondence location for reaching landscape image, without according to the manually opened application program with navigation feature of recognition result, from
And realize quickly navigation.
Referring to FIG. 2, the embodiment of the present invention also provides a kind of classification identification device of landscape image, including segmentation module
11, class prediction module 12, acquisition module 13 and confirmation module 14, in which:
Segmentation module 11 is divided from the landscape image for obtaining the landscape image comprising target image and background
Target image out, as segmented image.
Terminal or server, which can be extracted directly, is stored in local landscape image, and it is logical also to can receive other electronic equipments
Cross the landscape image of wired connection mode or radio connection transmission.Wherein, in landscape image include target image and background,
Target image can be the element image of mark action, such as landmark, road, the text information with labeled
Nameplate, guideboard, bus stop board etc..
Preferably, it is partitioned into target image from the landscape image, as segmented image, comprising:
Feature extraction is carried out to the landscape image by semantic segmentation model, identifies at least one in the landscape image
Region, and generate target image；
The semantic segmentation model carries out sample training, trained semantic segmentation model by mark characteristic area in advance
For generating the target image comprising at least one region.
It is preferably based on convolutional neural networks unit (CNN), constructs a U-shaped neural network structure model, and using prison
Educational inspector's learning and training method, to realize detection and the building semantic segmentation model of target image.
In addition, landscape image is by taking street view image as an example, by semantic segmentation model, such as ENet, the streetscape figure that will test
The target images such as building, road, nameplate as in extract, and are respectively formed the area comprising building, road, nameplate
Domain, as segmented image.The sample training ENet for having marked street view image region is first passed through in advance, when performing image processing, benefit
The target image comprising at least one region is generated with trained ENet.
Class prediction module 12, for obtaining classification category in the class prediction model of segmented image input built in advance
Property；In the convolutional neural networks structural model that target image input is pre-established, the convolutional neural networks structural model is used
In the incidence relation of characterization segmented image and category attribute.
Segmented image is identified based on neural network, can correspond to and obtain the corresponding category attribute of the segmented image.
By taking street view image as an example, the category attribute of street view image refers to streetscape type, such as the type of building, road, specifically,
Building includes the different building types such as school, mansion, church, villa.Road includes highway, drive, country lane etc.
Road type.
Preferably, the class prediction model generates by the following method:
Obtain several sample landscape images and the corresponding classification category of the target image comprising target image and background
Property；
It is partitioned into target image from the sample landscape image, as segmented image；
Using segmented image and category attribute, convolutional neural networks training is inputted, class prediction model is obtained.
Wherein, sample landscape image derives from landscape image disclosed in network and society, will by using crowdsourcing mode
The object region of building, road, nameplate in landscape image etc. is identified, while by the category attribute of target image
It is identified.Target image is inputted into training in deep layer convolutional neural networks model, and is based on the corresponding classification of the target image
Attribute completes supervised learning training, realizes the class prediction of segmented image.Preferably, by segmented image input built in advance
In class prediction model, comprising:
The maximum segmented image of image scaled is chosen as the first segmented image, first segmented image is inputted into built in advance
Class prediction model in；
If being more than present count according to the corresponding category result quantity of the landscape image that first segmented image obtains
Amount；
Remaining segmented image is sequentially input in the class prediction model of built in advance according to preset recognition sequence then and is carried out
Identification；The recognition sequence presets the influence degree of recognition result according to the attribute of segmented image.For example, segmented
To building, road and nameplate totally three independent segmented images, wherein the image scaled built is maximum, then building is made first
It is identified for the first segmented image, obtains the corresponding category result quantity of the landscape image when the first segmented image is identified
More than preset quantity, that is, when being unable to judge accurately the corresponding classification of the landscape image, by remaining road and nameplate this two
A segmented image sequentially inputs in the class prediction model of built in advance according to preset recognition sequence and is identified.Wherein, it identifies suitable
Sequence presets the influence degree of recognition result according to the attribute of different segmented images, for example, eye-catching when containing in direction board
Words identification, pattern identification when, be beneficial to the accurate judgement of recognition result, then recognition sequence be set as nameplate, road
Road.
By in the class prediction model after segmented image input training, the corresponding classification category of landscape image can be obtained
Property, thus tentatively identify the corresponding type of landscape image, it can be in specific category attribute based on the category attribute after identification
It is further identified in range, greatly reduces the range of landscape image identification, be conducive to the subsequent accurate knowledge of landscape image
Not.
Module 13 is obtained, for obtaining the corresponding characteristic attribute of the category attribute；
Confirmation module 14 confirms the landscape figure according to the characteristic attribute for utilizing preset landscape classifying rules
The classification of picture.
Characteristic attribute refers to the features such as shape corresponding to the segmented image of attribute of all categories, text, pattern, color.To build
For building object, the characteristic attribute of building includes that monnolithic case, color, floor quantity etc. can also include further building
Clock, statue in additional structural features on object, such as building.Specifically, based on trained identification model to the class
The corresponding target image of other attribute is identified, the corresponding characteristic attribute of the category attribute is obtained.For example, after for segmentation
Building A is identified that obtain the corresponding characteristic attribute of building A, characteristic attribute includes shape a, pattern b and color c.
Preferably, using preset landscape classifying rules, the classification of the landscape image is confirmed according to the characteristic attribute,
Include:
Using the characteristic attribute, the class of the corresponding relationship of the classification of preset characteristic feature attribute and landscape image is inquired
Other database, obtains the classification of the landscape image.
Specifically, the category database of the corresponding relationship of the classification of the characteristic feature attribute and landscape image, is to be directed to
The characteristic attribute set that landscape image of all categories is established, wherein each landscape image corresponds to one or more feature category
Property.For example, the characteristic attribute of " big ben " includes shape a, pattern b, color c and structure d.
After shape a, pattern b and color c after being identified based on identification model, preset classification number is further inquired
According in library, the landscape image comprising shape a, pattern b and color c obtains the classification of landscape image to confirm.
Specifically, it if being confirmed to obtain the category result of multiple landscape images according to the characteristic attribute, shows respectively more
True picture corresponding to the classification of a landscape image.Preferably, at least show that the classification of the landscape image is corresponding
Details picture and whole picture further can also show the picture of the corresponding different angle of the classification of the landscape image.
For example, there are three types of as a result, then showing that three kinds of results are corresponding respectively for the landscape image comprising shape a, pattern b and color c
Landscape image.
Preferably, after the classification that the landscape image is confirmed according to the characteristic attribute, further includes:
The corresponding summary info of the landscape image is generated according to the classification of the landscape image.
Wherein, summary info may include building title, road name, sight name etc., also may include landscape image
Brief introduction, location information etc..For example, the landscape image that characteristic attribute includes shape a, pattern b and color c has " big ben ", then
Generate the summary info comprising " big ben ".
Preferably, the landscape image is street view image, generates the landscape image according to the classification of the landscape image
Corresponding summary info, comprising:
The corresponding location information of the street view image is generated according to the classification of the street view image；
Route guidance information is formed according to the positional information.
Specifically, after generating the summary info comprising " big ben ", the position of " big ben " can further be generated
Information.
By training obtain to identify category attribute corresponding to segmented image class prediction model, by tentatively knowing
Not Chu the corresponding category attribute of landscape image, based on trained identification model to the corresponding target image of the category attribute into
Row identification, and establishes the category database of the corresponding relationship of the classification of characteristic feature attribute and landscape image simultaneously, realizes pair
Landscape image carries out gradually fine identification, guarantees the accuracy of final recognition result.For example, by category attribute range
Interior progress characteristic attribute identification, can be greatly improved the recognition efficiency of landscape image, while guaranteeing the identification knot of landscape image
The accuracy of fruit.In addition, class prediction model, identification model etc. pass through deep learning method, additionally it is possible to constantly improve identification knot
Fruit realizes the quick of landscape image and accurately identifies.
It, can be according to the class of street view image after the classification for identifying street view image when landscape image is street view image
The mapping relations of location information not corresponding with the street view image generate the corresponding location information of street view image, to guide user
The correspondence location for reaching landscape image, without according to the manually opened application program with navigation feature of recognition result, from
And realize quickly navigation.
In one embodiment, the invention also provides a kind of computer equipment, the computer equipment includes that storage is situated between
Matter, processor and storage on said storage and can handle the computer program that run on medium, the processing described
Device performs the steps of when executing the computer program obtains the landscape image comprising target image and background, from the wind
It is partitioned into target image in scape image, as segmented image；By in the class prediction model of segmented image input built in advance, obtain
To category attribute；In the convolutional neural networks structural model that target image input is pre-established, the convolutional neural networks knot
Structure model is used to characterize the incidence relation of segmented image and category attribute；Obtain the corresponding characteristic attribute of the category attribute；Benefit
With preset landscape classifying rules, the classification of the landscape image is confirmed according to the characteristic attribute.
In one embodiment, following steps are also executed when processor executes computer-readable instruction: according to the landscape
The classification of image generates the corresponding summary info of the landscape image.
Referring to FIG. 3, Fig. 3 is the schematic diagram of internal structure of computer equipment in one embodiment.As shown in figure 3, the meter
Calculating machine equipment includes processor 1, storage medium 2, memory 3 and the network interface 4 connected by system bus.Wherein, the meter
The storage medium 2 for calculating machine equipment is stored with operating system, database and computer-readable instruction, can be stored with control in database
Information sequence may make processor 1 to realize a kind of classification of landscape image when the computer-readable instruction is executed by processor 1
Recognition methods, processor 1 are able to achieve the segmentation module in the classification identification device of one of embodiment illustrated in fig. 2 landscape image
11, class prediction module 12, the function of acquisition module 13 and confirmation module 14.The processor 1 of the computer equipment is for providing
Calculating and control ability, support the operation of entire computer equipment.Computer can be stored in the memory 3 of the computer equipment
Readable instruction may make processor 1 to execute a kind of classification of landscape image when the computer-readable instruction is executed by processor 1
Recognition methods.The network interface 4 of the computer equipment is used for and terminal connection communication.It will be understood by those skilled in the art that Fig. 3
Shown in structure, only the block diagram of part-structure relevant to application scheme, does not constitute and is answered application scheme
With the restriction of computer equipment thereon, specific computer equipment may include than more or fewer portions as shown in the figure
Part perhaps combines certain components or with different component layouts.
In addition, the invention also provides a kind of storage medium for being stored with computer-readable instruction, the computer-readable finger
When order is executed by one or more processors, so that one or more processors execute following steps: obtaining includes target image
With the landscape image of background, it is partitioned into target image from the landscape image, as segmented image；The segmented image is defeated
Enter in the class prediction model of built in advance, obtains category attribute；Target image is inputted to the convolutional neural networks structure pre-established
In model, the convolutional neural networks structural model is used to characterize the incidence relation of segmented image and category attribute；Described in acquisition
The corresponding characteristic attribute of category attribute；Using preset landscape classifying rules, the landscape figure is confirmed according to the characteristic attribute
The classification of picture.One or more processors be able to achieve in embodiment illustrated in fig. 2 in the classification identification device of landscape image
Divide module 11, class prediction module 12, the function of obtaining module 13 and confirmation module 14.
In one embodiment, following steps are also executed when processor executes computer-readable instruction: according to the landscape
The classification of image generates the corresponding summary info of the landscape image.
Based on the above embodiments it is found that the maximum beneficial effect of the present invention is:
By training obtain to identify category attribute corresponding to segmented image class prediction model, by tentatively knowing
Not Chu the corresponding category attribute of landscape image, based on trained identification model to the corresponding target image of the category attribute into
Row identification, and establishes the category database of the corresponding relationship of the classification of characteristic feature attribute and landscape image simultaneously, realizes pair
Landscape image carries out gradually fine identification, guarantees the accuracy of final recognition result.For example, by category attribute range
Interior progress characteristic attribute identification, can be greatly improved the recognition efficiency of landscape image, while guaranteeing the identification knot of landscape image
The accuracy of fruit.In addition, class prediction model, identification model etc. pass through deep learning method, additionally it is possible to constantly improve identification knot
Fruit realizes the quick of landscape image and accurately identifies.
It, can be according to the class of street view image after the classification for identifying street view image when landscape image is street view image
The mapping relations of location information not corresponding with the street view image generate the corresponding location information of street view image, to guide user
The correspondence location for reaching landscape image, without according to the manually opened application program with navigation feature of recognition result, from
And realize quickly navigation.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, which can be stored in a computer-readable storage and be situated between
In matter, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, storage medium above-mentioned can be
The non-volatile memory mediums such as magnetic disk, CD, read-only memory (Read-Only Memory, ROM) or random storage note
Recall body (Random Access Memory, RAM) etc..
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
Limitations on the scope of the patent of the present invention therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to guarantor of the invention
Protect range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of classification recognition methods of landscape image, which is characterized in that the described method includes:
The landscape image comprising target image and background is obtained, target image is partitioned into from the landscape image, as segmentation
Image；
By in the class prediction model of segmented image input built in advance, category attribute is obtained；Target image input is built in advance
In vertical convolutional neural networks structural model, the convolutional neural networks structural model is for characterizing segmented image and category attribute
Incidence relation；
Obtain the corresponding characteristic attribute of the category attribute；It is true according to the characteristic attribute using preset landscape classifying rules
Recognize the classification of the landscape image.
2. the classification recognition methods of landscape image according to claim 1, which is characterized in that the class prediction model is logical
Cross following methods generation:
Obtain several sample landscape images and the corresponding category attribute of the target image comprising target image and background；
It is partitioned into target image from the sample landscape image, as segmented image；
Using segmented image and category attribute, convolutional neural networks training is inputted, class prediction model is obtained.
3. the classification recognition methods of landscape image according to claim 1, which is characterized in that classified using preset landscape
Rule confirms the classification of the landscape image according to the characteristic attribute, comprising:
Using the characteristic attribute, the data of the corresponding relationship of the classification of preset characteristic feature attribute and landscape image are inquired
Library obtains the classification of the landscape image.
4. the classification recognition methods of landscape image according to claim 1, which is characterized in that divide from the landscape image
Target image is cut out, as segmented image, comprising:
Feature extraction is carried out to the landscape image by semantic segmentation model, identifies at least one area in the landscape image
Domain, and generate target image；
The semantic segmentation model carries out sample training by mark characteristic area in advance, and trained semantic segmentation model is used for
Generate the target image comprising at least one region.
5. the classification recognition methods of landscape image according to claim 1, which is characterized in that input the segmented image
In the class prediction model of built in advance, comprising:
The maximum segmented image of image scaled is chosen as the first segmented image, by the class of first segmented image input built in advance
In other prediction model；
If being more than preset quantity according to the corresponding category result quantity of the landscape image that first segmented image obtains；Then
Remaining segmented image is sequentially input in the class prediction model of built in advance according to preset recognition sequence and is identified；The knowledge
Sequence does not preset the influence degree of recognition result according to the attribute of segmented image.
6. the classification recognition methods of landscape image according to claim 1, which is characterized in that true according to the characteristic attribute
Recognize after the classification of the landscape image, further includes:
The corresponding summary info of the landscape image is generated according to the classification of the landscape image.
7. the classification recognition methods of landscape image according to claim 6, which is characterized in that the landscape image is streetscape
Image generates the corresponding summary info of the landscape image according to the classification of the landscape image, comprising:
The corresponding location information of the street view image is generated according to the classification of the street view image；
Route guidance information is formed according to the positional information.
8. a kind of classification identification device of landscape image, which is characterized in that described device includes:
Divide module and is partitioned into target from the landscape image for obtaining the landscape image comprising target image and background
Image, as segmented image；
Class prediction module, for obtaining category attribute in the class prediction model of segmented image input built in advance；By mesh
In the convolutional neural networks structural model that logo image input pre-establishes, the convolutional neural networks structural model divides for characterizing
Cut the incidence relation of image and category attribute；
Module is obtained, for obtaining the corresponding characteristic attribute of the category attribute；
Confirmation module confirms the class of the landscape image according to the characteristic attribute for utilizing preset landscape classifying rules
Not.
9. a kind of computer equipment, which is characterized in that including storage medium and processor, calculating is stored in the storage medium
Machine readable instruction, when the computer-readable instruction is executed by the processor, so that the processor executes such as claim 1
To landscape image described in any one of 7 classification recognition methods the step of.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable instruction is by one or more processors
When execution, so that one or more processors execute the classification identification side of the landscape image as described in any one of claims 1 to 7
The step of method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811369125.6A CN109858318A (en) | 2018-11-16 | 2018-11-16 | The classification recognition methods of landscape image and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811369125.6A CN109858318A (en) | 2018-11-16 | 2018-11-16 | The classification recognition methods of landscape image and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109858318A true CN109858318A (en) | 2019-06-07 |
Family
ID=66890112
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811369125.6A Pending CN109858318A (en) | 2018-11-16 | 2018-11-16 | The classification recognition methods of landscape image and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109858318A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334234A (en) * | 2019-07-15 | 2019-10-15 | 深圳市祈锦通信技术有限公司 | A kind of scenery picture classification method and its device |
| WO2021003853A1 (en) * | 2019-07-09 | 2021-01-14 | Visuo Technology Pty Limited | A method and a system for processing an image, and for generating a contextually coherent video based on images processed thereby |
| CN112711998A (en) * | 2020-12-24 | 2021-04-27 | 珠海新天地科技有限公司 | 3D model annotation system and method |
| CN113610138A (en) * | 2021-08-02 | 2021-11-05 | 典基网络科技(上海)有限公司 | Image classification and identification method and device based on deep learning model and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103064921A (en) * | 2012-12-20 | 2013-04-24 | 北京工业大学 | Method of achieving intelligent digital tour guide in museum |
| CN107563379A (en) * | 2017-09-02 | 2018-01-09 | 西安电子科技大学 | For the localization method to natural scene image Chinese version |
| CN108287893A (en) * | 2018-01-18 | 2018-07-17 | 安徽大学 | A kind of self-help guide system and its method based on digital image understanding auxiliary positioning |
| CN108563702A (en) * | 2018-03-23 | 2018-09-21 | 美景听听（北京）科技有限公司 | Speech sound eeplaining data processing method and device based on showpiece image recognition |
-
2018
- 2018-11-16 CN CN201811369125.6A patent/CN109858318A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103064921A (en) * | 2012-12-20 | 2013-04-24 | 北京工业大学 | Method of achieving intelligent digital tour guide in museum |
| CN107563379A (en) * | 2017-09-02 | 2018-01-09 | 西安电子科技大学 | For the localization method to natural scene image Chinese version |
| CN108287893A (en) * | 2018-01-18 | 2018-07-17 | 安徽大学 | A kind of self-help guide system and its method based on digital image understanding auxiliary positioning |
| CN108563702A (en) * | 2018-03-23 | 2018-09-21 | 美景听听（北京）科技有限公司 | Speech sound eeplaining data processing method and device based on showpiece image recognition |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021003853A1 (en) * | 2019-07-09 | 2021-01-14 | Visuo Technology Pty Limited | A method and a system for processing an image, and for generating a contextually coherent video based on images processed thereby |
| CN110334234A (en) * | 2019-07-15 | 2019-10-15 | 深圳市祈锦通信技术有限公司 | A kind of scenery picture classification method and its device |
| CN110334234B (en) * | 2019-07-15 | 2022-03-18 | 深圳市祈锦通信技术有限公司 | Landscape picture classification method and device |
| CN112711998A (en) * | 2020-12-24 | 2021-04-27 | 珠海新天地科技有限公司 | 3D model annotation system and method |
| CN113610138A (en) * | 2021-08-02 | 2021-11-05 | 典基网络科技(上海)有限公司 | Image classification and identification method and device based on deep learning model and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105930841B (en) | The method, apparatus and computer equipment of automatic semantic tagger are carried out to image | |
| CN109858318A (en) | The classification recognition methods of landscape image and device | |
| CN104268603B (en) | Intelligent marking method and system for text objective questions | |
| CN109271401A (en) | Method, apparatus, electronic equipment and storage medium are corrected in a kind of search of topic | |
| CN108304835A (en) | character detecting method and device | |
| CN110457420B (en) | Point-of-interest point identification method, device, equipment and storage medium | |
| CN106528531A (en) | Artificial intelligence-based intention analysis method and apparatus | |
| Krukar et al. | The effect of orientation instructions on the recall and reuse of route and survey elements in wayfinding descriptions | |
| CN107679189A (en) | A kind of point of interest update method, device, server and medium | |
| CN109815955A (en) | Topic householder method and system | |
| Brata et al. | An effective approach to develop location-based augmented reality information support | |
| CN110276023A (en) | POI changes event discovery method, apparatus, calculates equipment and medium | |
| CN112528639B (en) | Object recognition method and device, storage medium and electronic equipment | |
| CN110020653A (en) | Image, semantic dividing method, device and computer readable storage medium | |
| CN110727816A (en) | Method and device for determining interest point category | |
| CN110414581A (en) | Picture detection method and device, storage medium and electronic device | |
| Vargas Munoz et al. | Deploying machine learning to assist digital humanitarians: making image annotation in OpenStreetMap more efficient | |
| CN107291774A (en) | Error sample recognition methods and device | |
| CN110457706B (en) | Point-of-interest name selection model training method, using method, device and storage medium | |
| CN109145956A (en) | Methods of marking, device, computer equipment and storage medium | |
| CN110414001A (en) | Sentence generation method and device, storage medium and electronic device | |
| CN105845020A (en) | Real-scene map making method and device | |
| CN113420692A (en) | Method, apparatus, device, medium, and program product for generating direction recognition model | |
| CN107491240A (en) | Travelling ordering method, system, equipment and storage medium based on dynamic icon | |
| CN113918802B (en) | Navigation method, device, equipment, medium and product |
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 |