KR101468457B1 - method for smart searching of goods - Google Patents

Abstract

The present invention relates to a smart product search providing method for allowing a user to search for an optimal product in consideration of a user's preference and a user's body shape information. A method of searching for a desired product from the product information DB to provide information, the method comprising the steps of: (A) selecting an item classification of a search target product from a user; (B) inputting a requirement for searching for a product including at least one of a user's preference or a user's body shape from the user; (C) setting a search range in the goods information DB according to the item classification; (D) if the preference of the user is included in the search requirement, performing a search by preference to calculate a search result; (E) if the user's body type is included in the search requirement, performing a body type search to calculate a search result; And (F) ranking the search results according to the degree of matching and the number of matching and providing the ranking to the user. According to the present invention, since the user's preference is calculated from the past purchase history and the product is searched and provided based on the past purchase history, the product can be searched by accurately reflecting the user's taste.

Description

METHOD FOR SMART SEARCHING OF GOODS

The present invention provides a smart product search providing method that allows a user to search for an optimal product in consideration of the user's preference and the user's body shape information.

In recent years, technologies have been developed and applied to extract various parts of an image and use it in various industrial fields. For example, techniques for automatically extracting a specific target using an image captured through a camera and recognizing information about the target have been actively studied.

Further, a product search technique for selecting and searching products having an image most similar to a product image inputted by a user using these image recognition technologies is disclosed in prior arts such as Korean Patent Laid-Open No. 10-2012-0001285.

However, as shown in FIG. 1, the prior art is merely to compare an image input by a user with an image stored in a database, and to search for and provide a target product according to the similarity, and thus the user's preference, A product search system that considers the information of the product information collectively can not be provided.

Korean Patent Publication No. 10-2012-0001285

SUMMARY OF THE INVENTION The present invention has been made in order to solve the conventional problems as described above, and it is an object of the present invention to provide a method and apparatus for forming a preferred product image group according to a purchase pattern of a user, The present invention also provides a method for providing a smart product search.

In addition, the present invention provides a method for providing a smart product search providing a product suitable for a user's body using a user's body shape information in the case of a fashion product.

According to an aspect of the present invention, there is provided a method of providing a user information DB and a product information DB to search for a desired product from the product information DB to provide information (A) selecting an item classification of a search target product from a user; (B) inputting a requirement for searching for a product including at least one of a user's preference or a user's body shape from the user; (C) setting a search range in the goods information DB according to the item classification; (D) if the preference of the user is included in the search requirement, performing a search by preference to calculate a search result; (E) if the user's body type is included in the search requirement, performing a body type search to calculate a search result; And (F) ranking the search results according to the degree of matching and the number of matching and providing the ranking to the user.

Here, the user information DB includes a preference goods DB and a user body information DB for each user: the preference goods DB stores purchase goods image data, which is an image of a product that the user has previously purchased, And average image data, which is a representative image of images belonging to the same category, is stored. In the user's body information DB, the user's body type may be stored according to a normalized classification criterion.

The average image data may be purchased product image data having the highest degree of similarity with the average form image of the purchased product images belonging to the same category.

Further, the purchased merchandise image data may be added by the user's input.

The classification criteria of the user body type information DB may include at least one of a rasband type classification or an ITYB lower body type classification.

The user's body type information DB may be constructed by detecting a type from an image of an input user and comparing the detected type with a representative image according to a classification criterion to discriminate the classification of the input image.

The merchandise information DB includes a merchandise classification information DB and a purchaser body information DB for each merchandise; Wherein the product classification information DB stores a product image and a product classification; The buyer's body type information DB may store body information of buyers who purchase the goods.

The preference-based search may be performed by comparing the average image data of the user information DB with the product images stored in the product information DB, calculating the similarity, and generating the search results in descending order of the degree of similarity.

In addition, the body type search may be performed by comparing the body shape information of the user stored in the user body type information DB with the purchaser body type information of the product information DB to generate the search results in the order of high degree of match.

In the smart product search providing method according to the present invention as described above, the following effects can be expected.

That is, in the present invention, since the user's preference tastes are calculated from the past purchase history and the product is searched and provided on the basis thereof, there is an advantage that the product can be searched accurately by reflecting the taste of the user.

In the present invention, since the body type of the purchaser purchasing the commodity is compared with the body type of the user to search for the commodity, the purchaser of the same type as the user searches for the commodity which is generally preferred, Thus, there is an advantage that a product search can be provided to a user.

Brief Description of the Drawings Fig. 1 is a flow chart showing a product search method according to the prior art; Fig.
2 is an exemplary diagram showing an example of a morphology detection technique according to the present invention;
3 is a block diagram showing a specific embodiment of a user information DB according to the present invention;
4 is a diagram illustrating an example of a method for generating average image data of a user information DB according to the present invention.
FIG. 5 is an exemplary diagram showing a Rasband body type classification applied to an embodiment of the present invention; FIG.
6 is a block diagram showing a specific embodiment of a goods information DB according to the present invention;
FIG. 7 is a flowchart showing a specific embodiment of a smart product search providing method according to the present invention. FIG.
FIG. 8 is a flow chart illustrating an embodiment of a smart product search method according to the present invention; FIG.
9 is a diagram illustrating an embodiment of an image similarity determination method among smart goods search providing methods according to the present invention.

In the following, a description will be given of a technical principle for detecting a target article from an image and determining similarity in the practice of the present invention, and a specific embodiment of the present invention will be described.

1. Target detection (Haar-like feature + AdaBoost)

The Haar-like feature is a technique applied by Viola to face detection, and is a technique for expressing feature values using a simple sum image. It can have many shapes depending on the position and shape size. It uses the AdaBoost learning algorithm to combine the weak identifiers into a group of robust identifiers. In this case, the generated feature value has an advantage of embracing the characteristics of the face well. Since the calculation method is simple, it is very easy to apply to moving images and it becomes possible to extract facial image data within one second.

2. Edge detection (Active contour model)

As shown in FIG. 2, an edge of an object is detected to refer to an image processing technique for extracting edges from an image.

- Provides information on the location of objects, the shape and size of objects, textures, and design patterns of items

- The brightness of the image is suddenly changed.

- Numerous edge orientations are possible depending on area and shape

3. Template matching

Step 1: Similarity ( Similarity ) find

Template Matching is used to generate a and b correlation maps. a image and a B image are overlapped to obtain how much similarity exists.

The similarity can be determined by using the Euclidean method or the like

(R (x, y) = sum x ', y' [T (x ', y') - I (x + x ', y + y')] 2)

Step 2: Maximum value ( Maximum value ) find

And calculates a maximum value having the maximum similarity in the correlation coefficient map obtained in the first step.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a method for providing a smart product search according to the present invention will be described in detail with reference to the accompanying drawings.

In order to implement the smart product search providing method according to the present invention, a user information DB and a product information DB are constructed.

The term " user " used herein means a user who searches for goods through the service.

As shown in FIG. 3, the user information DB includes a DB for each user who uses the service according to the present invention. Each user DB includes a favorite goods DB 100 and a user body information DB 200 .

As shown in FIG. 3, the preference merchandise DB 100 is a part for storing an image of a product that the user has previously purchased, and image (image) data of various goods are classified and stored according to categories.

That is, in the preferred goods DB 100, the category according to the type of the goods is classified and stored from the upper category to the lower category, and the category of the final category is stored with the image data 110, 120 of the purchased product belonging to the lower category, Are stored.

Meanwhile, the image data 110 and 120 of the purchased product may further include image data as well as article information on the article of the image data. The article information may be, for example, the place where the article is purchased, the blend information, the product number, the price information, and the like.

In the category of the final detailed classification, the average image data 130 obtained by calculating an average of the purchased product images is stored. As shown in FIG. 4, the average image data 130 may be generated by extracting the types of product images previously purchased by users belonging to the same sub-category and calculating an average shape image of the products.

Furthermore, it is also possible to replace the average image data by selecting the purchased product image having the highest degree of similarity with the average image.

Basically, the purchased merchandise image data is basically collected and stored according to the purchasing history of the user, but it is also possible to additionally store the image data of the merchandise not purchased according to the input of the user.

This is because the preference merchandise DB reflects the user's preference of the merchandise in the search for the merchandise so that the image of the merchandise preferred by the user can be additionally reflected even if the purchase is not actually made.

Meanwhile, the user's body type information DB 200 is a part for normalizing and storing the user's body type, and stores the body type information of the user classified by various classification methods.

In this specification, the body type of the user is classified by the Rasband classification method proposed by Rasband in 1994 and the ITYB lower body type classification method as an example.

That is, the user's body type information DB includes Rasband body type classification information 210 for storing the body shape of the user classified by the Rasband body type classification, ITYB lower body classification information 220 for storing the body shape of the user classified by the ITYB lower body type classification, .

As shown in FIG. 5, the Rasband body type classification is a body type classification proposed by Rasband in 1994. The body type of a human body is classified into 1) ideal body shape, 2) triangular body shape, 3) inverted triangular body shape, , 5) hourglass type, 6) rhombic body type, 7) tubular body type, and 8) round body type.

In addition, the ITYB lower body type classification is based on the shape of the lower body of a person based on the shape of the pelvis, thigh and calf. 1) I-shaped body with similar pelvis, thigh and calf circumference, 2) It is a lower body type classification in which a calf is a thin body type T shape, 3) a pelvis and a thigh are large, a calf body is a Y shape, and 4) a pelvis, a thigh, and a calf are large.

The user's body type information DB 200 may further include various pieces of body information such as skin color, face type, neck girth, shoulder width, chest girth, waist girth, hip girth, thigh girth, upper body length, .

If the user inputs his or her own image, the processing system of the present invention detects the type of the input image, and if the detected type is the Rasband body type classification standard Image type and the ITYB lower body classification reference image, and the user's body type information DB is constructed by selecting the type of the body type having the highest similarity.

Of course, in this case, it is also possible to construct a user's body type information DB by designating a specific body type classification according to a user's selection without inputting an image of the user.

6, a DB is constructed for each product provided to the service according to the present invention. Each product DB includes a goods classification information DB 300 and a purchaser body information DB 400 ).

The commodity classification information DB 300 includes a commodity image 310, a commodity classification 320, and conditional tag information 330.

The goods classification means the categories from the major category to the detailed category of the goods classified by the same classification system as the goods classification as shown in FIG.

The conditional tag information 330 stores various information of the corresponding product. When the user searches for a product to be described later, if the user wants to search for a product by condition, To search for and provide matching products.

Therefore, the conditional tag information 330 may be seasonal information, age information, and information on wearing situation (funeral, wedding, etc.) when the product is a garment.

On the other hand, the buyer's body type information DB 400 is a part for storing the body types of the purchasers who have purchased the goods, and stores the body type information for the previous buyers classified by the same criterion as the body type classification standard of the user- Information is stored.

Therefore, in the present embodiment, the Rasband figure information 410 and the ITYB lower body information 420 of the purchasers who purchased the product are separately stored.

Hereinafter, a method for providing product search using the user information DB and product information DB constructed as described above will be described in detail.

FIG. 7 is a flowchart illustrating a method of providing a product search according to a specific embodiment of the present invention.

As shown in the figure, the smart product search providing method according to the present invention starts with a user selecting a search target product (S100). At this time, the selection of the search target product means selecting the kind of goods to be searched, and inputting one of the major classification and the detailed classification of the goods desired to be searched.

In operation S300, the user searches for a product to be searched based on the category of the search target product inputted in operation 100, and sets the search range to the same classified product in operation S300.

Here, the retrieval requirement means inputting whether to search for a product through 1) user's preference, 2) user's body type, and 3) other criteria. At this time, as the other search conditions, various search conditions such as the use of the product, the use time, and the usage method can be inputted.

Thereafter, when the user's preference is included in the search requirement inputted by the user, the search result is calculated by performing the search by preference (S410). 2) When the user's body is included, The search result is calculated (S420), and 3) when the search condition is included, the search result by condition is performed to calculate the search result (S430).

Then, the search result is ranked according to the degree of matching and the matching number (S500).

Next, search results sorted by rank are provided to the user (S600).

Hereinafter, specific retrieval methods of the retrieval method by preference, the retrieval method by body type, and the retrieval method by condition will be respectively described.

1) How to perform search by preference

As shown in FIG. 8, the product search by preference is a search for a product of the style by reflecting the user's preference style of the product. Data is extracted (S412).

In this case, as described above, the average image data is an average image of an average image of an image of a product previously purchased by the user or an image of products previously purchased by the user and images directly input by the user, It is a video that represents the style of the product.

Next, only the product information of the same classification as the same sub-classification as the product the user wants to search among the goods information stored in the goods information DB is limited to the search object (S414).

As shown in FIG. 9, the average image data extracted in step 412 and the product images of the product information defined in step 414 are compared to determine their similarity (step S416).

Thereafter, search results are sorted and stored as search results of the products having high similarity (S418).

2) Search by body type

The body type search method is for performing a product search based on the user's body type and the degree of fit of the product. The body type information of the user stored in the user's body type information DB is compared with the buyer's body type information of the product information DB, And the result is calculated.

That is, a comparison is made between the body type of the purchasers who have previously purchased the product and the body type of the user, and the search result is provided by searching for a product that is popular among the users similar to the user.

3) Search by condition

And searches for a matching product by comparing the condition inputted by the user with the tag information by the condition of the product classification information DB.

In this way, it is possible to search for preferred products considering the weather, the place of visit, the occupation, the age of the product, etc., by using the weather information and the season information of the product such as outside temperature and precipitation probability, A job or the like, and the product predicted by the above-mentioned product search for each body type, thereby providing search results.

It is to be understood that the invention is not limited to the disclosed embodiment, but is capable of many modifications and variations within the scope of the appended claims. It is self-evident.

The present invention relates to a smart product search providing method for allowing a user to search for an optimal product in consideration of a user's preference and a user's body shape information. According to the present invention, And the product is searched and provided on the basis thereof, so that it is possible to search for a product accurately reflecting the taste of the user.

100: Preferred item DB 200: User figure information DB
300: goods classification information DB 400: buyer body information DB

Claims (9)

A method of providing information by constructing a user information DB and a product information DB, searching for a desired product from the product information DB,
(A) selecting an item classification of a search target product from a user;
(B) inputting a requirement for searching for a product including at least one of a user's preference or a user's body shape from the user;
(C) setting a search range in the goods information DB according to the item classification;
(D) if the preference of the user is included in the search requirement, performing a search by preference to calculate a search result;
(E) if the user's body type is included in the search requirement, performing a body type search to calculate a search result; And
(F) ranking the search results according to the degree of matching and the number of matching, and providing the ranking to the user;
The user information DB,
A preferred product DB and a user's body type information DB for each user,
In the preferred merchandise DB,
Image data of purchased product which is an image of a product that a user has already purchased is divided and stored for each category and average image data which is a representative image of images belonging to the same category is stored:
In the user's body type information DB,
And the user's body type is stored according to a normalized classification criterion.
delete The method according to claim 1,
The average image data may include:
And the purchase order product image data having the highest degree of similarity with the average shape image of purchase product images belonging to the same category.
The method of claim 3,
The purchased merchandise image data includes:
Wherein the smart product search information is added by a user's input.
The method according to claim 1,
The classification standard of the user body type information DB is,
A Rasband body type classification, or an ITYB lower body body type classification.
The method according to claim 1,
The user's body type information DB,
Detecting a type from an image of an input user, comparing the detected type with a representative image according to a classification criterion, and discriminating the classification of the input image.
The method according to claim 1,
The product information DB includes:
A goods classification information DB and a buyer body type information DB for each product;
Wherein the product classification information DB stores a product image and a product classification;
And the body shape information of the purchaser who purchased the product is stored in the purchaser's body information DB.
8. The method according to any one of claims 1 to 7,
The preference-based search may include:
Comparing the average image data of the user information DB with the product images stored in the product information DB, calculating the similarity of the product images, and generating the search results in descending order of the degree of similarity.
8. The method according to any one of claims 1 to 7,
The body-
And comparing the body shape information of the user stored in the user body type information DB with the purchaser body type information of the product information DB to generate a search result in the order of higher degree of match.

Images