KR102129345B1 - Picture search system and method for specific hierarchical - Google Patents

Abstract

The present invention relates to a photo retrieval system and method for a specific layer, the photo retrieval system for a specific layer, a photo retrieval system for a specific layer includes a plurality of photos extracted from an open DB and feature information for each of the plurality of photos A standard DB picture in a standard DB constructed by combining and a search engine module for searching a picture corresponding to the search keyword in the open DB through the Internet network with a search keyword input by the user and a standard DB picture in the standard DB corresponding to the search keyword In the open DB, including a matching module that extracts and compares the extracted standard DB picture and the picture retrieved from the search engine module and removes the picture that does not fit the search keyword from the pictures searched and extracted by the search engine module. The advantage of constructing a standard DB with a large number of extracted photos and comparing the standard DB photos of the standard DB with the photos obtained by searching in an open DB for a specific layer to remove unsuitable photos such as junk to get high quality photos. There is this.

Description

Picture search system and method for specific hierarchical}

The present invention relates to a photo retrieval system and method for a specific layer, and more specifically, to build a standard DB with a number of photos extracted from an open DB and to obtain a photo and a standard DB obtained by searching a specific layer in the open DB. It relates to a photo retrieval system and method for a specific class that can obtain high-quality photos by removing unsuitable photos such as junk by comparing standard DB photos.

Recently, with the spread of digital cameras, smartphones, etc., users are taking pictures using portable devices and viewing or sharing them. In particular, more and more people share their photos with others through wired/wireless interworking. have.

As a method of sharing one's picture with others through existing wired/wireless interworking, a method of sharing a picture by creating a blog or a method of transmitting a picture to another person using a multimedia messaging system (MMS) has been used.

In recent years, various types of services for sharing photos using a social network service (SNS) having an open DB of various types such as Instagram and Facebook are being provided.

On the other hand, certain classes such as web, document, and clothing designers require various photos opened in the open database of SNS for design work.

In Korean Patent Publication No. 2014-0119217 (Patent Document 1), a photo sharing system capable of searching for photos through keyword search comprises: a communication unit connected to a communication network; A photo database in which photo files are stored; When a photo file is uploaded from a user terminal through the communication network, a similarity between a photo file stored in the photo database and the uploaded photo file is compared through a content-based image search technique to extract at least one photo file having a predetermined similarity or higher. A CBIR module; The main control unit determines keywords of the uploaded photo files based on keywords registered for at least one photo file extracted by the CBIR module, and registers the uploaded photo files and the determined keywords in the photo database. Disclosed is a photo sharing system capable of searching for photos through keyword search, which includes a feature.

Patent Literature 1 uploads a photo file that a user wants to share freely online, and when a user who wants to utilize a photo downloads and uses it, keywords can be automatically assigned to easily search for a registered photo. Although there is an advantage in that a picture can be searched through keyword search, there is a disadvantage in that it is impossible to obtain a quality picture desired by a specific class.

Korean Patent Publication No. 2014-0119217

The present invention has been devised in view of the above points, and its purpose is to provide a photo retrieval system and method for a particular layer that can search for an open DB to obtain high-quality photos.

Other objects of the present invention will become more apparent through the preferred embodiments described below.

According to an aspect of the present invention, a standard DB constructed by combining a plurality of photos extracted from an open DB and feature information of each of the plurality of photos and a search keyword input by a user through the Internet network to perform the search in the open DB The search engine module searches for a picture corresponding to a keyword and extracts a standard DB picture in the standard DB corresponding to the search keyword, compares the extracted standard DB picture with the picture extracted by searching in the search engine module, and performs the search. It is a photo search system for a specific layer that includes a matching module that removes photos that are not suitable for a search keyword among photos extracted by searching in the engine module.

According to another aspect of the present invention, the open DB is characterized in that it is at least one of Instagram, Facebook, Twitter, the matching module is removed from the picture that is not suitable for the search keyword, the picture outside the search system Further comprising an output module for outputting, the picture output from the output module is characterized in that it is output to a display or output to a communication unit is transmitted to a personal portable terminal through the Internet network.

According to another aspect of the present invention, the matching module generates a feature point to generate a feature point from the standard DB photo extractor and the extracted standard DB photo to search the standard DB with the search keyword and extract the appropriate standard DB photo. Determination unit for determining whether the feature points generated by the wealth and the feature point generation unit do not exist in the photo searched and extracted from the open DB, and deletion to delete the pictures extracted by searching in the open DB without feature points in the determination unit It is characterized by including wealth.

According to another aspect of the present invention, the step of establishing the standard DB and the user by extracting a plurality of photos from the first open DB through the Internet network and recording a standard DB picture combining feature information with each of the extracted multiple photos The search engine module searches for the picture corresponding to the search keyword in the second open DB through the Internet network with the search keyword entered, and the matching module extracts the standard DB picture corresponding to the search keyword from the standard DB. Step and comparing the extracted standard DB picture and the picture extracted by searching in the search engine module, removing the picture that is not suitable for the search keyword among the pictures searched and extracted by the search engine module, and not suitable for the search keyword It is a photo search method for a specific layer, including the step of searching the extracted search engine module from which the photos have been removed and outputting the extracted photos to the output module.

According to another aspect of the invention, N1 is the number of the first open DB, N2 is the number of the second open DB, characterized in that satisfying the conditional expression of N1 <N2, the extracted standard The step of removing a picture that does not fit the search keyword among the pictures searched and extracted by the search engine module by comparing the DB picture and the picture extracted by searching in the search engine module generates a feature point in the standard DB picture extracted by the matching module And determining whether the generated feature point does not exist in the photo extracted by searching in the open DB and removing the photo extracted by searching in the open DB in which the feature point does not exist. do.

According to the present invention, a standard DB is constructed with a plurality of photos extracted from an open DB such as Instagram, and the standard DB photos of the standard DB are compared with the photos obtained by searching in the open DB for a specific layer and are not suitable as junk. There is an advantage that you can get a good quality picture by removing the unused picture.

1 is a block diagram of a photo retrieval system for a specific layer according to the present invention,
Figure 2 is a photo recorded on Instagram among the open DB applied in accordance with the present invention.
3 to 6 are pictures recorded in the standard DB according to the present invention.
FIG. 7 is a diagram for explaining extraction of a suitable picture by comparing the picture retrieved from the search engine module with the picture of the standard DB in the matching module according to the present invention.
8 is a detailed configuration diagram of a matching module according to the present invention.
9 is a view for explaining a range of feature points of a standard DB picture according to the present invention.
10 is a flowchart of a method for retrieving photos for a specific layer according to the first embodiment of the present invention.
11 is a flowchart of a method for retrieving photos for a specific layer according to a second embodiment of the present invention.

The present invention can be applied to various transformations and can have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all conversions, equivalents, and substitutes included in the spirit and scope of the present invention. In the description of the present invention, when it is determined that a detailed description of related known technologies may obscure the subject matter of the present invention, the detailed description will be omitted.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from other components.

The terms used in this application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, the terms "include" or "have" are intended to indicate the presence of features, numbers, steps, actions, components, parts or combinations thereof described herein, one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, operations, components, parts or combinations thereof are not excluded in advance. Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a configuration diagram of a photo search system for a specific layer according to the present invention. Referring to this, the photo search system 1000 for a specific layer according to the present invention includes a standard DB 100, an input module 200, and a search. It includes an engine module 300, a matching module 500, and a control module 600.

The standard DB 100 is constructed by combining a plurality of photos extracted from the open DB 10 and feature information of each of the plurality of photos. That is, the standard DB 100 contains a plurality of pictures combined with feature information.

Here, the open DB 10 is defined as a database containing pictures on social networks (SNS) such as Instagram, Facebook, and Twitter.

Instagram is a semi-open SNS (some information is open and the rest of the information is closed), and users and seniors in their 20s and 30s, 47% for men and 53% for women, have emotional and impact photos and videos. It consists of content.

Facebook is the world's leading SNS, a half-open SNS that is mostly used by teenagers in their twenties and 56% for men and 44% for women.

Twitter is an open SNS with 44% for men and 56% for women.

In addition, the Kakao Story and Naver bands are not included in the open DB 10 because they are closed SNS.

In addition, the feature information combined with the photo recorded in the standard DB 100 is generated from the features appearing in the photo and the hash tag attached to the photo.

Therefore, a standard DB 100 is constructed by combining a plurality of standard DB photos extracted from the open DB 10 and feature information of each of the plurality of standard DB photos.

The search engine module 300 searches for a picture corresponding to the search keyword in the open DB 10 through the Internet network with the search keyword (or search query) entered by the user.

That is, when the user inputs the search keyword'smiley face' through the input module 200, the search engine module 300 searches the open DB 10 through the Internet network to obtain the most suitable picture for the'smiley face'. Search.

At this time, the search engine module 300 searches for the photos contained in the open DB 10 while searching for a'smiling face' and hash tags in the form of'#smiling faces' combined with the photos.

As a result, the photos searched in the search engine module 300 searches for and extracts photos with smiley faces and photos combined with hashtags including search keywords.

Here, in the present invention, as a method for providing a search keyword to the search engine module 300, among the several methods presented by the user and the search engine module 300, the user inputs the search keyword directly into the input module 200. The category method provided by narrowing the range by selecting a desired item by the user may be applied.

And, it is preferable that at least two or more pictures extracted from the search engine module 300.

Also, the search engine module 300 may search the standard DB 100 with the search keyword input by the user to search for a standard DB picture corresponding to the search keyword.

The matching module 500 extracts the standard DB picture in the standard DB 100 corresponding to the search keyword input by the user, compares the extracted standard DB picture with the search engine module 300, and compares the extracted picture to the search engine Among the photos searched and extracted by the module 300, photos that are not suitable for a search keyword (which can be defined as a'junk') are removed.

As a result, the searched and extracted photos from the search engine module 300 become the photos desired by the user.

The control module 600 controls the standard DB 100, the search engine module 300, and the matching module 500 so that a user can extract a desired photo.

That is, when a user's search keyword is input through the input module 200, the control module 600 transmits the first control signal to the search engine module 300 to search and search the open DB 10 through the Internet network. Search for photos corresponding to keywords.

Thereafter, when the search engine module 300 searches for and extracts the pictures contained in the open DB 10 suitable for the search keyword, the second control signal is transmitted to the matching module 500 to search and extract the search engine module 300 The picture is compared with the standard DB picture, and the search engine module 300 removes the picture that is not suitable for the search keyword from the searched and extracted picture.

The photo search system 1000 is for a specific layer, which is a user who needs the searched photo, and the specific layer may be various, but in the present invention, it is preferable that the specific layer, that is, a designer who desires design information of the user. Do.

On the other hand, in the present invention, the matching module 500 may further include an output module that outputs a picture that is not suitable for the search keyword 300 to the outside of the search system.

That is, the picture in which the non-conforming picture is removed in comparison with the standard DB picture in the matching module 500 is output to the output module so that the user can use it for his/her desired work purpose.

Alternatively, in the present invention, the picture in which the non-conforming picture is removed from the matching module 500 compared to the standard DB picture is output from the output module to a display outside the search system or output to a communication unit to be transmitted to the personal portable terminal through the Internet network. Can.

In this case, the photo retrieval system 1000 for a specific layer according to the present invention is built in a server of a management company and a user using a personal portable terminal connected to the management company becomes a specific layer, and the user pays a predetermined fee. Photo search service.

Here, the search keyword input from the personal portable terminal is input to the input module 200 of the search system through the communication unit of the management company to operate the search system.

2 is a picture recorded on Instagram among the open DB applied according to the present invention, FIGS. 3 to 6 are pictures recorded on a standard DB according to the present invention, and FIG. 7 is a search engine module in a matching module according to the present invention Fig. 8 is a detailed configuration diagram of a matching module according to the present invention, and Fig. 9 is a range of feature points of a standard DB picture according to the present invention. It is a figure for illustration.

As described above, in the search system of the present invention, a picture is extracted from an open DB by a user's search keyword.

A hash tag is attached to the picture recorded in the open DB.

That is, as shown in FIG. 2, a picture recorded on Instagram, which is a kind of open DB, has a hash tag of'feel good in a sunny weather pole cactus'.

For example, if the search keyword is'cactus' or'sunny', the search engine module 300 searches and extracts the image of FIG. 2 as the image and hashtag of the picture, and the search keyword is'a woman wearing sunglasses', the The search engine module 300 searches and extracts the picture of FIG. 2 as a person and a hash tag.

In the present invention, the standard DB 100 is constructed by combining a plurality of photos extracted from an open DB and feature information of each of the plurality of photos.

The standard DB 100 may contain approximately 1000 photos, and may contain 500 to 5000 photos.

That is, in the standard DB 100 shown in FIG. 3, the characteristic information of'black' in race is combined.

And, in the standard DB 100, the feature information of'Sunset Heart' is included in a photo of men and women making hearts with arms at sunset on the beach as shown in FIG. 4, and the feature information of'Happy Couple' in the photo of FIG. Combined, the feature information of'pet dog and sympathy' is combined in the picture of FIG. 6.

Since this characteristic information is defined as an implicit word or phrase representing an image that can be highlighted in a standard DB picture, the standard DB picture contained in the standard DB 100 is not suitable for pictures extracted by searching in the search engine module 300. It can be used as a reference picture to remove unused pictures.

The matching module 500 extracts a suitable picture by comparing the picture retrieved from the search engine module 300 with the picture of the standard DB 100.

Referring to FIG. 7, the search keyword'smiling' is compared with the standard DB picture (D) in the standard DB (100) by comparing the extracted pictures (A, B, C) from the open DB by applying the search keyword'smiling face' Remove the unsuitable photos on the face and extract only the suitable photos.

Here, the matching module 500 compares the extracted pictures (A, B, and C) from the open DB with the standard DB picture (D) in the standard DB (100) to remove the picture unsuitable for the search keyword. The feature points related to the keyword are found in the standard DB picture (D), and the found feature points are searched in the open DB to determine whether the extracted pictures (A, B, C) are provided.

That is, the matching module 500 searches the standard DB 100 with the search keyword'smiley face', extracts the standard DB picture of'D', and'mouths open' from the smiley face of the extracted standard DB picture (D). Find the characteristic point of'the state where the circumscribed muscles are contracted and the circumscribed muscle surrounding the eye is contracted' and find the characteristic point in the open DB to compare and judge whether it exists in the extracted pictures (A, B, C) Is to do.

Therefore, the characteristic point of the'state where the mouth is wide open, the teeth are exposed and the circumferential muscles surrounding the eye are contracted and the half-eye is created' in the standard DB photo (D) is a photograph'A' photograph extracted from the open DB. And the'C' picture is removed because it exists in the'B' picture but not in the'C' picture. That is, the C'picture judged not to be included in the characteristic confirmation range of the standard DB picture (D) is deleted.

Thus, the matching module 500, as shown in Figure 8, a standard DB picture extraction unit 510 for extracting a suitable standard DB picture by searching in the standard DB 100 as a search keyword; A feature point generator 520 for generating a feature point from the extracted standard DB picture; A determination unit 530 for determining whether the feature point generated by the feature point generation unit 520 does not exist in a picture extracted by searching in an open DB; And a deletion unit 540 that deletes the picture extracted by searching in the open DB in which the feature point does not exist in the determination unit 530.

Here, the feature point generation unit 520 may generate a feature point using only a standard DB picture or extract feature points by relating feature information associated with a standard DB picture included in the standard DB to a standard DB picture image.

At this time, when the feature information associated with the standard DB picture recorded in the standard DB is extracted and generated by associating the feature point with the standard DB picture image, the feature point generation unit 520 works in conjunction with the subdivided definition DB defined by subdividing the feature information. You can create feature points.

For example, if the feature information is'Happy couple on the beach','Beach' among the feature information is '30~50% of the cloud occupying the sky except during the daytime when the sunlight is strong, and the wave of the sea is a calm beach. It is defined as'Segmentation and'Happy Couple' is defined as'A couple walking from right to left in the picture with young men and women holding hands and walking with a smile' and is included in the Subdivision Definition DB.

In addition, in the present invention, the feature points of the standard DB picture (D) may be multiple, and some of each of the multiple feature points may have a specified range (a range for determining a characteristic).

For example, in the standard DB picture (D) of FIG. 7, the feature point is'the mouth is opened, the teeth are exposed, and the circumscribed muscles surrounding the eye are contracted to form a half-moon', so (a)'the mouth is open The three characteristic points are composed of'state', (b)'exposed teeth', and (c)'state in which the circumferential muscles surrounding the eye contract and the half-moon is formed'.

Here, referring to FIG. 9, among the three feature points (a), (b), and (c), the feature point (a)'a state where the mouth is open' has a distance (d1) between the upper and lower lips of 1 to 3 cm, Feature point (b) The'tooth exposed state' may have a range of feature points that 10 or more teeth are exposed.

10 is a flowchart of a photo search method for a specific layer according to the first embodiment of the present invention, and FIG. 11 is a flowchart of a photo search method for a specific layer according to the second embodiment of the present invention.

Referring to FIG. 10, a method for retrieving photos for a specific layer according to the first embodiment of the present invention first extracts a plurality of photos from a first open DB through an Internet network and combines feature information with each of the extracted photos. Standard DB is built by recording standard DB photos (S100).

Thereafter, the search engine module searches for a picture corresponding to the search keyword in the second open DB through the Internet network with the search keyword input by the user (S110).

Here, the first open type DB may be the same as or different from the second open type DB.

In this case, when the first open DB is different from the second open DB, the number N1 of the first open DB may be smaller than the number N2 of the second open DB.

That is, it is designed to satisfy N1 &lt; N2, and the reason for doing so is advantageous in that a photo desired by a specific layer can be retrieved and extracted from more open DBs, thereby obtaining a high quality photo.

For example, the first open DB is an Instagram, and the second open DB can be at least two of Instagram, Facebook, and Twitter.

Subsequently, the matching module extracts a standard DB picture corresponding to the search keyword from the standard DB (S120).

Next, the extracted standard DB picture and the picture searched and extracted by the search engine module are compared, and a picture not suitable for the search keyword is removed from the picture searched and extracted by the search engine module (S130).

Subsequently, the search engine module in which the photos not suitable for the search keyword are removed is searched and the extracted photos are output to the output module (S140).

Here, the'step S140' may be replaced with a step of transmitting to the personal portable terminal through the Internet network.

Referring to FIG. 11, in the method of searching for photos for a specific layer according to the second embodiment of the present invention, steps S100 to S120 and S140 of the first embodiment described above are the same, and step S130 is implemented as a detailed step. Redundant explanation is omitted.

Therefore, step S130 of the first embodiment generates a feature point from the standard DB picture extracted by the matching module in step S120 as shown in FIG. 11 (S131), and the generated feature point is found in the picture extracted by searching in the open DB. It is determined whether or not (S132).

Thereafter, a picture extracted by searching in the open DB where the feature point does not exist is removed (S133).

The above-described preferred embodiments of the present invention are disclosed for the purpose of illustration, and those skilled in the art will appreciate various modifications, changes, and additions within the spirit and scope of the present invention. It should be considered that the addition is within the scope of the following claims.

100: standard DB
200: input module
300: search engine module
500: matching module
510: Standard DB picture extraction unit
520: feature point generation unit
530: judgment unit
540: deletion unit

Claims (8)

A standard DB constructed by combining a plurality of standard DB photos extracted from the open DB and feature information of each of the plurality of standard DB photos;
A search engine module that searches at least two pictures corresponding to the search keyword in an open DB through the Internet network as a search keyword input by a user through an input module;
It extracts the standard DB picture in the standard DB corresponding to the search keyword, compares the extracted standard DB picture with the picture extracted by searching in the search engine module, and retrieves the standard DB picture among the pictures searched and extracted in the search engine module. A matching module that removes photos not suitable for the search keyword by removing photos not included in the feature determination range; And
And an output module that outputs a picture from which the picture not suitable for the search keyword is removed from the matching module to the outside of the search system.
The picture output from the output module is output to a display or output to a communication unit and transmitted to the personal portable terminal through the Internet network.
The user is a designer who performs design work,
The open DB is a database that contains pictures on at least one social network (SNS) among Instagram, Facebook, and Twitter,
If the search keyword is'smiley face',
The search engine module searches for at least two pictures corresponding to the'smiley face' search keyword in the open DB,
The matching module extracts a standard DB picture having a hash tag corresponding to a'smiling face' search keyword from a number of standard DB photos in the standard DB, and extracts a standard DB picture extracted from the search engine module. Compare,
In the standard DB picture that has a hash tag corresponding to the'smiling face' search keyword, feature points (a)'open mouth', feature points (b)'exposed teeth' and feature points (c)'half-eye' The three feature points of the losing state are combined,
The feature point (a)'open mouth' is between 1 and 3 cm between the upper and lower lips, and the feature point (b)'the state where the teeth are exposed' includes a feature confirmation range that 10 or more teeth are exposed. And
A photo search system characterized by removing and providing a photo that is not suitable for a search keyword by removing a photo that is not included in the standard DB photo feature determination range from the photo searched and extracted by the search engine module. delete delete delete delete delete delete delete

Images