Summary of the invention
In view of the above problems, the embodiment of the present application is proposed to provide a kind of overcoming the problems referred to above or a kind of image search method solved the problem at least in part and corresponding a kind of image search apparatus.
According to an aspect of the application, disclose a kind of image search method, described method comprises:
What reception was uploaded by mobile terminal comprises the first image of first object object and the solicited message of image category;
Extract the characteristic information of the first object object in described first image;
In image data base, search meets second image being matched with the characteristic information of the first object object in described first image of described request information;
The first object object of the first object object of described first image and described second image and the second destination object is adopted to generate the 3rd image and show.
Alternatively, described method also comprises:
Obtain pending image in internet in advance, and determine the image category of described pending image according to the characteristic information of the second destination object of described pending image;
Described pending image is classified according to different image category and is kept at described image data base.
Alternatively, the described step being matched with the second image of the characteristic information of the first object object in described first image meeting the solicited message of described image category of searching in image data base comprises:
In described image data base, search meets the pending image of certain classification of described request information;
Extract the characteristic information of the first object object in the pending image of certain classification described;
Calculate the matching degree of the characteristic information of the first object object in the characteristic information of the first object object in described first image and the pending image of certain classification described;
When described matching degree is greater than predetermined threshold value, using the pending image of certain classification described as described second image.
Alternatively, the first object object of described first image of described employing and the first object object of described second image and the second destination object generate the 3rd image and the step shown comprises:
Extract the pixel coordinate information of described first image and described second image respectively;
The pixel coordinate information of described first image and described second image is adopted to determine the first object object of described first image and the first object object of described second image and the target viewing area of the second destination object respectively;
By described 3rd image of described target viewing area synthesis.
Alternatively, describedly obtain pending image in internet in advance, and determine that according to the characteristic information of the second destination object of described pending image the step of the image category of described pending image comprises:
The characteristic information of the second destination object of described pending image is extracted by convolutional neural networks system;
Mate the characteristic information of the second destination object of described pending image and the standard feature information of different each image category corresponding respectively;
The image category of image category as described pending image of the standard feature information of the characteristic information of the second destination object of described pending image will be matched with.
Alternatively, the position of the first object object of described second image is within the second destination object of described second image, and the first object object of described first image of described employing and the first object object of described second image and the second destination object generate the 3rd image and the step shown comprises:
The first object object of described first image and second destination object containing first object object of described second image generate described 3rd image with layout type side by side and show.
According to another aspect of the application, disclose a kind of image search apparatus, described device comprises:
Image receiver module, for receive uploaded by mobile terminal comprise the first image of first object object and the solicited message of image category;
Characteristic information extracting module, for extracting the characteristic information of the first object object in described first image;
Search module, meets second image being matched with the characteristic information of the first object object in described first image of the solicited message of described image category for search in image data base;
Computer image genration module, generates the 3rd image for the first object object and the second destination object adopting the first object object of described first image and described second image and shows.
Alternatively, described device also comprises:
Pretreatment module, for obtaining pending image in internet in advance, and determines the image category of described pending image according to the characteristic information of the second destination object of described pending image;
Preserve module, for being classified according to different image category by described pending image and being kept at described image data base.
Alternatively, described search module comprises:
Pending picture search submodule, meets the pending image of certain classification of the solicited message of described image category for search in described image data base;
Pending image zooming-out submodule, for extracting the characteristic information of the first object object in the pending image of certain classification described;
Pending image calculating sub module, for the matching degree of the characteristic information of the first object object in the pending image of the characteristic information and certain classification described that calculate the first object object in described first image;
Second image determination submodule, for being greater than predetermined threshold value, using the pending image of certain classification described as described second image when described matching degree.
Alternatively, described Computer image genration module comprises:
Pixel coordinate information extracts submodule, for extracting the pixel coordinate information of described first image and described second image respectively;
Submodule is determined in target viewing area, determines the first object object of described first image and the first object object of described second image and the target viewing area of the second destination object respectively for adopting the pixel coordinate information of described first image and described second image;
3rd Images uniting submodule, for synthesizing described 3rd image by described target viewing area.
Alternatively, described pretreatment module comprises:
Convolutional neural networks system sub-modules, for extracting the characteristic information of the second destination object of described pending image by convolutional neural networks system;
Standard feature information matches submodule, for the standard feature information of the characteristic information from different each image category corresponding that mate the second destination object of described pending image respectively;
Image category determination submodule, for the image category of the standard feature information of the characteristic information of the second destination object using being matched with described pending image as the image category of described pending image.
Alternatively, the position of the first object object of described second image is within the second destination object of described second image, and described Computer image genration module comprises:
3rd image layout's submodule, generates described 3rd image for the first object object of described first image and second destination object containing first object object of described second image with layout type side by side and shows.
The embodiment of the present application comprises following advantage:
1, what the embodiment of the present application was uploaded by mobile terminal by reception comprises the first image of first object object and the solicited message of image category, extract the characteristic information of the first object object in the first image, in image data base, search meets second image being matched with first object characteristics of objects information in the first image of solicited message, the first object object of the first object object of the first image and described second image and the second destination object is adopted to generate the 3rd image and show, the 3rd image that user gets meets the demand of user to image category, improve the relevance of the image of image and the user's actual needs searched.
2, by matching characteristic information in the multiple pending image of the image category in the solicited message uploaded corresponding to user, user can according to the image of some image category, even can be with search target image category do not have related image, search image that is multiple or specific image classification, thus improve extensibility and the dirigibility of picture search.
Embodiment
For enabling above-mentioned purpose, the feature and advantage of the application more become apparent, below in conjunction with the drawings and specific embodiments, the application is described in further detail.
With reference to Fig. 1, show the flow chart of steps of a kind of image search method embodiment 1 of the application, specifically can comprise the steps:
Step 101, receive uploaded by mobile terminal comprise the first image of first object object and the solicited message of image category.
In the application scenarios of reality, user may to several article or object in conjunction with interested, utilize fashion designer to want to utilize zoomorphism to design a fashionable dress, want to search for some images with the clothes of animal pattern as design reference; Or furniture design Shi Xiang designs a sofa for certain natural landscape, other furniture being decorated with this natural landscape that meeting removal search is correlated with.When searching for, user can by one or more article or object such as with taking pictures, the mode of 3D scanning be made into image, or utilize the existing image containing these article or object, by the mobile terminal such as computer, mobile phone by image uploading to server.User can also limit by the solicited message containing image category the image containing the content of certain particular category according to actual needs, user is by containing the image of certain article or object and containing the solicited message of image category, so that search package contains the multiple images belonging to certain specific image classification of article in institute's upload images or object, such as user needs the image of certain sharpei to be attached on women's dress shirt, this sharpei can be taken pictures and upload images onto server, and the image of request search women's dress shirt classification, to search the image of the women's dress shirt with sharpei pattern.
Step 102, extracts the characteristic information of the first object object in described first image.
In concrete realization, server can identify one or more first object object respectively by existing image recognition technology, and extracts the characteristic information of this first object object from the first object object identified.Characteristic information can be the computer-readable data of the various appearances characteristic of reflection first object object color in the picture, size, shape, decorative pattern etc., such as, color can be the data of first object object RGB color system (RedGreenBlueColorSystem) in the picture, size can be first object object each pixel coordinate data etc. in the picture, those skilled in the art can adopt the data of the different various appearance characteristics of reflection destination object as characteristic information according to actual conditions, and the embodiment of the present application is not restricted this.
Step 103, in image data base, search meets second image being matched with the characteristic information of the first object object in described first image of the solicited message of described image category.
It should be noted that, the second destination object of above-mentioned pending image is the destination object that can reflect image category.In specific implementation, in advance the multiple pending image be kept in image data base can be classified according to the characteristic information of pending image, the mode of classification can be by extracting and identifying the characteristic information of the second destination object of pending image and realize, such as, extract the characteristic information of pending image, from characteristic information, identify in image the resemblance with women's dress shirt, the image of the image category of women's dress shirt can be classified as.
Image category in the solicited message that server is uploaded according to user meets the multiple pending image of the image category in this solicited message at image database search, and carry out feature information extraction process for the multiple pending image first object object of this image category further, such as, for the pending image of women's dress shirt image category, at the image section of the second destination object women's dress shirt, extract the profile within the scope of women's dress shirt, color, the characteristic information of the resemblances such as decorative pattern, and carry out matching degree calculating with the characteristic information of the first object object of the first image, when matching degree is greater than default matching degree threshold value, can using pending image as the second image.Described matching degree can be set by those skilled in the art according to actual conditions, and the embodiment of the present application does not limit this.
Step 104, adopts the first object object of the first object object of described first image and described second image and the second destination object generate the 3rd image and show.
In the application of reality, the first image that user uploads and the second image searched, user's unwanted picture information may be contained, therefore can by the first object object of extraction first image and the first object object of the second image and the second destination object, and the part of extraction is cut out again merge and become a new image and be shown to user as the 3rd image.3rd image only remains with the image information that user needs, and decreases the garbage in image and disturbs the vision of user, improve the efficiency that user obtains image information.
What the embodiment of the present application was uploaded by mobile terminal by reception comprises the first image of first object object and the solicited message of image category, extract the characteristic information of the first object object in the first image, in image data base, search meets second image being matched with the characteristic information of the first object object in the first image of image category in solicited message, the first object object of the first object object of the first image and described second image and the second destination object is adopted to generate the 3rd image and show, the 3rd image that user gets meets the demand of user to image category, improve the relevance of the image of image and the user's actual needs searched.
Further, by matching characteristic information in the multiple pending image of the image category in the solicited message uploaded corresponding to user, user can according to the image of some image category, even can be with search target image category do not have related image, search image that is multiple or specific image classification, thus improve extensibility and the dirigibility of picture search.
With reference to Fig. 2, show the flow chart of steps of a kind of image search method embodiment 2 of the application, specifically can comprise the steps:
Step 201, obtains pending image in internet in advance, and determines the image category of described pending image according to the characteristic information of the second destination object of described pending image.
Step 202, classifies described pending image according to different image category and is kept at described image data base.
In specific implementation, server can get a large amount of images as pending image by internet, and the characteristic information of the second destination object of pending image can be extracted by existing image recognition technology, by the standard feature information matches that this characteristic information and certain classification are preset, can using the classification of this certain classification as pending image after the match is successful.Such as, the standard feature information of men's clothing pants can be preset with, for pending image, can the part of trousers shape be met in recognition image and carry out feature information extraction, when being extracted the characteristic information of trousers shape part in pending image, carry out matching degree calculating with the standard feature information of the men's clothing pants preset, after matching degree meets certain threshold value, this pending image can be defined as the image category of men's clothing pants.
After each pending image determines image category, can correspondingly add image category label and be kept in image data base.
As the preferred exemplary of the embodiment of the present application, described step 201 can comprise following sub-step:
Sub-step S11, extracts the characteristic information of the second destination object of described pending image by convolutional neural networks system.
Sub-step S12, mates the characteristic information of the second destination object of described pending image and the standard feature information of different each image category corresponding respectively.
Sub-step S13, will be matched with the image category of image category as described pending image of the standard feature information of the characteristic information of the second destination object of described pending image.
Convolutional neural networks system (ConvolutionalNeuralNetworks, CNN) be one can extract image feature information and realize the recognition system of degree of depth study by utilizing image dissimilar in a large number to carry out training, so-called degree of depth study is the cutting edge technology of artificial intelligence field, it can complete the artificial intelligence task needing high abstraction feature, as speech recognition, image recognition and retrieval, natural language understanding etc.Degree of depth study have employed deep-neural-network model (DeepNeuralNetworks, DNN), Deep model is the artificial neural network comprising multiple hidden layer, and multilayered nonlinear structure makes it possess powerful feature representation ability and to complex task modeling ability.Training Deep model obtains successfully in multiple application with stratification, the serial of methods that is successively initialized as representative.From the convolutional neural networks system of deep-neural-network model development, the statistical nature information of the image local inherent characteristic consistent with the characteristic information of other local is utilized to carry out graphical analysis and identification, be applicable to the large-size images that pixel is more, convolutional neural networks system is utilized to carry out feature information extraction and the matching treatment of image, can recognition image more accurately, thus improve the accuracy of picture search.
In addition, convolutional neural networks systemic presupposition has the standard feature information of each image category.The characteristic information of the second destination object of pending image is extracted by controlling convolutional neural networks system, and by this characteristic information and the standard feature information matches that correspond to each different images classification, when the standard feature information matches of the characteristic information of the second destination object of pending image and certain image category, can using the image category of this image category as pending image.
Step 203, receive uploaded by mobile terminal comprise the first image of first object object and the solicited message of image category.
Step 204, extracts the characteristic information of the first object object in described first image.
Step 205, in described image data base, search meets the pending image of certain classification of described request information.
Step 206, extracts the characteristic information of the first object object in the pending image of certain classification described.
Step 207, calculates the matching degree of the characteristic information of the first object object in the characteristic information of the first object object in described first image and the pending image of certain classification described.
Step 208, when described matching degree is greater than predetermined threshold value, using the pending image of certain classification described as described second image.
In specific implementation, user can pass through computer, the mobile terminals such as mobile phone send to server to include the image of certain article or object and the solicited message of image category, the characteristic information of the main object in the image that server extraction user uploads, and the multiple pending image of respective image classification can be searched in image data base according to the image category in solicited message, feature information extraction process is carried out to the plurality of pending image, the characteristic information of the first object object in the first image that the characteristic information extracted and user upload carries out matching degree calculating, when matching degree is greater than default matching degree threshold value, can using pending image as the second image.
Step 209, adopts the first object object of the first object object of described first image and described second image and the second destination object generate the 3rd image and show.
As the preferred exemplary 1 of the embodiment of the present application, described step 209 can comprise following sub-step:
Sub-step S21, extracts the pixel coordinate information of described first image and described second image respectively.
Sub-step S22, adopts the pixel coordinate information of described first image and described second image to determine the first object object of described first image and the first object object of described second image and the target viewing area of the second destination object respectively.
Sub-step S23, by described 3rd image of described target viewing area synthesis.
In specific implementation, the pixel coordinate information of existing image recognition technology recognition image can be utilized, pixel coordinate information can be each pixel coordinate data in the picture of composition destination object, utilize pixel coordinate information, the differentiation border of destination object and non-targeted object in image can be sketched the contours of, the region that destination object needs in the picture show can be determined according to distinguishing border, the target viewing area of the first object object of the first image and the first object object of the second image and the second destination object can be determined by the way respectively.Those skilled in the art also can adopt the target viewing area in other mode determination images, and the embodiment of the present application is not restricted this.
After determining each target viewing area, the 3rd image can be synthesized.The mode of target viewing area synthesis the 3rd image can be had multiple, such as can by the target viewing area of the target viewing area of the first image and the second image respectively from original image cutting out, be reassembled into the 3rd new image by left and right layout side by side.
As the preferred exemplary 2 of the embodiment of the present application, the position of the first object object of described second image is within the second destination object of described second image, and described step 209 can comprise following sub-step:
Sub-step S31, the first object object of described first image and second destination object containing first object object of described second image generate described 3rd image with layout type side by side and show.
It should be noted that, in practical application, the position of first object object is within the second destination object usually, as the animal pattern in clothes, animal pattern can not exceed the scope of clothes usually, therefore when generation the 3rd image, only need to determine the first object object of the first image and the second destination object of the second image, namely this 3rd image can be generated, the 3rd image generated contains the first object object of the first image and the first object object of the second image and the second destination object, thus process resource is saved while ensure that the image information that the 3rd image comprises.
The first object object of the first image comprised in the 3rd image and second destination object containing first object object of the second image combine with layout type side by side and show, make user contrast article in the image uploaded or object and the article in the image searched more easily or object whether similar.
According to the embodiment of the present application, by obtaining a large amount of images in advance in internet as pending image, feature information extraction process is carried out to pending image, characteristic information according to the second destination object of pending image is classified and is kept at image data base, search for the multiple pending image of the image category in the solicited message uploaded corresponding to user in a database, the characteristic information of the further first object object of coupling first image and the first object object of the second image, make user can get a large amount of images larger with the image relevance of user's actual needs, extend the search coverage rate of picture search.
It should be noted that, for embodiment of the method, in order to simple description, therefore it is all expressed as a series of combination of actions, but those skilled in the art should know, the embodiment of the present application is not by the restriction of described sequence of movement, because according to the embodiment of the present application, some step can adopt other orders or carry out simultaneously.Secondly, those skilled in the art also should know, the embodiment described in instructions all belongs to preferred embodiment, and involved action might not be that the embodiment of the present application is necessary.
With reference to Fig. 3, show the structured flowchart of a kind of image search apparatus embodiment 1 of the application, specifically can comprise as lower module:
Image receiver module 301, for receive uploaded by mobile terminal comprise the first image of first object object and the solicited message of image category.
Characteristic information extracting module 302, for extracting the characteristic information of the first object object in described first image.
Search module 303, meets second image being matched with the characteristic information of the first object object in described first image of described request information for search in image data base.
Computer image genration module 304, generates the 3rd image for the first object object and the second destination object adopting the first object object of described first image and described second image and shows.
3rd image only remains with the image information that user needs, and decreases the garbage in image and disturbs the vision of user, improve the efficiency that user obtains image information.
What the embodiment of the present application was uploaded by mobile terminal by reception comprises the first image of first object object and the solicited message of image category, extract the characteristic information of the first object object in the first image, in image data base, search meets second image being matched with the characteristic information of the first object object in the first image of solicited message, the first object object of the first object object of the first image and described second image and the second destination object is adopted to generate the 3rd image and show, the 3rd image that user gets meets the demand of user to image category, improve the relevance of the image of image and the user's actual needs searched.
Further, by matching characteristic information in the multiple pending image of the image category in the solicited message uploaded corresponding to user, user can according to the image of some image category, even can be with search target image category do not have related image, search image that is multiple or specific image classification, thus improve extensibility and the dirigibility of picture search.
With reference to Fig. 4, show the structured flowchart of a kind of image search apparatus embodiment 2 of the application, specifically can comprise as lower module:
Pretreatment module 401, for obtaining pending image in internet in advance, and determines the image category of described pending image according to the characteristic information of the second destination object of described pending image.
Preserve module 402, for being classified according to different image category by described pending image and being kept at described image data base.
As the preferred exemplary of the embodiment of the present application, described pretreatment module 401 can comprise following submodule:
Convolutional neural networks system sub-modules, for extracting the characteristic information of the second destination object of described pending image by convolutional neural networks system.
Standard feature information matches submodule, for the standard feature information of the characteristic information from different each image category corresponding that mate the second destination object of described pending image respectively.
Image category determination submodule, for the image category of the standard feature information of the characteristic information of the second destination object using being matched with described pending image as the image category of described pending image.
Image receiver module 403, for receive uploaded by mobile terminal comprise the first image of first object object and the solicited message of image category.
Characteristic information extracting module 404, for extracting the characteristic information of the first object object in described first image.
Search module 405, meets second image being matched with the characteristic information of the first object object in described first image of described request information for search in image data base.
As the preferred exemplary of the embodiment of the present application, described search module 405 can comprise following submodule:
Pending picture search submodule, meets the pending image of certain classification of described request information for search in described image data base.
Pending image zooming-out submodule, for extracting the characteristic information of the first object object in the pending image of certain classification described.
Pending image calculating sub module, for the matching degree of the characteristic information of the first object object in the pending image of the characteristic information and certain classification described that calculate the first object object in described first image.
Second image determination submodule, for being greater than predetermined threshold value, using the pending image of certain classification described as described second image when described matching degree.
Computer image genration module 406, generates the 3rd image for the first object object and the second destination object adopting the first object object of described first image and described second image and shows.
As the preferred exemplary one of the embodiment of the present application, described Computer image genration module 406 can comprise following submodule:
Pixel coordinate information extracts submodule, for extracting the pixel coordinate information of described first image and described second image respectively.
Submodule is determined in target viewing area, determines the first object object of described first image and the first object object of described second image and the target viewing area of the second destination object respectively for adopting the pixel coordinate information of described first image and described second image.
3rd Images uniting submodule, for synthesizing described 3rd image by described target viewing area.
As the preferred exemplary two of the embodiment of the present application, described Computer image genration module 406 can comprise following submodule:
3rd image layout's submodule, generates described 3rd image for the first object object of described first image and second destination object containing first object object of described second image with layout type side by side and shows.
According to the embodiment of the present application, by obtaining a large amount of images in advance in internet as pending image, feature information extraction process is carried out to pending image, characteristic information according to the second destination object of pending image is classified and is kept at image data base, search for the multiple pending image of the image category in the solicited message uploaded corresponding to user in a database, the characteristic information of the further first object object of coupling first image and the first object object of the second image, make user can get a large amount of images larger with the image relevance of user's actual needs, extend the search coverage rate of picture search.
For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part illustrates see the part of embodiment of the method.
Each embodiment in this instructions all adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar part mutually see.
Those skilled in the art should understand, the embodiment of the embodiment of the present application can be provided as method, device or computer program.Therefore, the embodiment of the present application can adopt the form of complete hardware embodiment, completely software implementation or the embodiment in conjunction with software and hardware aspect.And the embodiment of the present application can adopt in one or more form wherein including the upper computer program implemented of computer-usable storage medium (including but not limited to magnetic disk memory, CD-ROM, optical memory etc.) of computer usable program code.
The embodiment of the present application describes with reference to according to the process flow diagram of the method for the embodiment of the present application, terminal device (system) and computer program and/or block scheme.Should understand can by the combination of the flow process in each flow process in computer program instructions realization flow figure and/or block scheme and/or square frame and process flow diagram and/or block scheme and/or square frame.These computer program instructions can being provided to the processor of multi-purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminal equipment to produce a machine, making the instruction performed by the processor of computing machine or other programmable data processing terminal equipment produce device for realizing the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
These computer program instructions also can be stored in can in the computer-readable memory that works in a specific way of vectoring computer or other programmable data processing terminal equipment, the instruction making to be stored in this computer-readable memory produces the manufacture comprising command device, and this command device realizes the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
These computer program instructions also can be loaded on computing machine or other programmable data processing terminal equipment, make to perform sequence of operations step to produce computer implemented process on computing machine or other programmable terminal equipment, thus the instruction performed on computing machine or other programmable terminal equipment is provided for the step realizing the function of specifying in process flow diagram flow process or multiple flow process and/or block scheme square frame or multiple square frame.
Although described the preferred embodiment of the embodiment of the present application, those skilled in the art once obtain the basic creative concept of cicada, then can make other change and amendment to these embodiments.So claims are intended to be interpreted as comprising preferred embodiment and falling into all changes and the amendment of the embodiment of the present application scope.
Finally, also it should be noted that, in this article, the such as relational terms of first and second grades and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or terminal device and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or terminal device.When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the terminal device comprising described key element and also there is other identical element.
A kind of image search method above the application provided and a kind of image search apparatus, be described in detail, apply specific case herein to set forth the principle of the application and embodiment, the explanation of above embodiment is just for helping method and the core concept thereof of understanding the application; Meanwhile, for one of ordinary skill in the art, according to the thought of the application, all will change in specific embodiments and applications, in sum, this description should not be construed as the restriction to the application.