KR102114369B1 - Online shopping mall banner design generation method, apparatus and system - Google Patents

Abstract

One aspect of the present invention discloses a method for generating a banner design associated with shopping mall promotion. The method includes: receiving at least one image and text associated with the promotion of the shopping mall; selecting a first image for use in the promotion of the shopping mall through image analysis of the inputted at least one image; and Analyzing the selected first image and setting the attribute value of the input text according to the analysis result, and generating a banner design associated with shopping mall promotion by merging the first image and text having the set attribute value Includes.

Description

Shopping mall banner design creation method, device and system {ONLINE SHOPPING MALL BANNER DESIGN GENERATION METHOD, APPARATUS AND SYSTEM}

The present invention relates to a method for generating a banner design in an online shopping mall, and more particularly, to a method for generating a banner phrase for enhancing a promotional effect of a shopping mall.

In the course of operating online e-commerce (which can be called a shopping mall), promoting the shopping mall is an indispensable activity to increase the awareness of the shopping mall and increase sales. The publicity activities of these shopping malls are often referred to as "promotions", which can be understood as "a variety of publicity activities conducted on a single subject."

Shopping mall operators use a variety of promotional means to increase the effectiveness of the promotion. Among them, a banner showing concise and impact of the promotion concept and related information can be said to be the core of the promotion means.

Therefore, the performance of the promotion has become an important factor to be influenced by the quality of the banner, but creating a banner with a high degree of completion is an area of experts who specialize in design, and ordinary people who have insufficient experience in this field pay a lot of money. The reality is that we are going to ask experts for production.

Accordingly, an application for efficiently creating a banner design is required without relying on experts.

An object according to an aspect of the present invention for solving the above-described problem is to process an image and text in a form optimized for publicity using the image and text processed when the user inputs only the image and text to create a banner design. It is to provide a method, apparatus, and system for generating a banner design by merging properly.

In order to achieve the above object, a method of generating a banner design associated with shopping mall promotion according to an aspect of the present invention includes receiving at least one image and text associated with shopping mall promotion, and inputting the at least one input Selecting a first image to be used for promotion of the shopping mall through image analysis on an image, analyzing the selected first image and setting an attribute value of the input text according to an analysis result and the agent And generating a banner design associated with shopping mall promotion by merging 1 image and text having the set attribute value.

The step of selecting a first image for use for promotion of the shopping mall through image analysis of the input at least one image is used for promotion based on information related to a face of a model in the at least one input image And determining whether it is possible.

The step of determining whether it can be used for promotion based on the model face in the at least one input image is used based on whether there is at least one preset face component in the face of the model in the at least one input image. Determining a possibility, determining availability based on a model face size in the at least one input image, and determining availability based on a direction of a model face in the at least one input image can do.

The step of selecting a first image to be used for promotion of the shopping mall through image analysis of the input at least one image may be determined based on the feature points of the model in the at least one input image to determine whether it can be used for promotion. It may include steps.

The step of selecting a first image to be used for promotion of the shopping mall through image analysis of the input at least one image is based on a result of semantic analysis of at least some of the plurality of words included in the input text. It may include the step of screening the image including the content to be matched with the first image.

Analyzing the selected first image and setting an attribute value of the input text according to an analysis result includes: extracting a size of the first image, extracting an eye position of a model in the first image And determining the position of the input text by determining whether the extracted eye position is on the left side or the right side in comparison to the central reference plane of the first image.

The banner design generation method associated with the shopping mall promotion further includes determining the number of grids associated with image segmentation by comparing the extracted image size with a reference value, wherein the central reference plane is set based on the determined number of grids. Can be.

The step of determining the number of grids associated with image segmentation by comparing the extracted image size with a reference value includes: a first mode and a second value row for dividing the grid into rows or columns of a first value compared to the reference value Alternatively, the method may include determining one of the second modes of dividing into a grid having columns.

Determining the position of the input text according to the determination whether the extracted eye position is on the left side or the right side in comparison with the center reference plane of the first image is the left side of the extracted eye location compared to the center reference plane Place the input text on the right side of the first image in response to being in, or place the input text in the center of the first image in correspondence with the extracted eye position being in the center of the central reference plane Alternatively, the method may include disposing the input text on the left side of the first image in response to the extracted eye location being on the right side of the central reference plane.

Analyzing the selected first image and setting an attribute value of the input text according to an analysis result includes: extracting emotion information associated with the first image and text style attributes corresponding to the extracted emotion information And matching the values.

The method of claim 1, wherein the setting of the attribute value of the input text according to an analysis result by analyzing the selected first image comprises: extracting complexity information associated with the first image and extracting complexity information. And matching the text thickness attribute value corresponding to.

Analyzing the selected first image and setting an attribute value of the input text according to an analysis result includes: extracting main color information associated with the first image and text corresponding to the extracted main color information And matching font attribute values.

Analyzing the selected first image and setting an attribute value of the input text according to an analysis result includes: extracting brightness information associated with the first image and text color attributes corresponding to the extracted brightness information And matching the values.

The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result includes: extracting complementary color information associated with the first image and text emphasis color corresponding to the extracted complementary color information And matching attribute values.

The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result may include: dividing the input text into a plurality of text elements, and setting the attribute values of each distinguished text element. And individually matching information extracted in association with the first image.

The banner design generation method associated with the promotion of the shopping mall may further include selecting a highlight phrase by extracting a key word based on a result of semantic analysis of at least one word included in the text.

The processing of the first image based on the result of semantic analysis of at least one word included in the text may include extracting emotional words representing emotion from the input text and based on the extracted emotional words. And converting the first image into a color corresponding to the emotion of the emotional word.

In order to achieve the above object, a banner design generating device associated with shopping mall promotion according to an aspect of the present invention includes: an input unit receiving at least one image and text associated with a shopping mall promotion; An image analysis unit that selects a first image to be used for promotion of the shopping mall through image analysis on an image, and performs image analysis on the selected first image to obtain an analysis result, and the selected first It includes a text attribute setting unit for setting the attribute value of the input text according to the analysis result for the image and a merging unit for merging the first image and the text having the set attribute value to generate a banner design associated with shopping mall promotion. Can be.

In order to achieve the above object, a method of generating a banner design associated with shopping mall promotion according to another aspect of the present invention includes receiving at least one image and text associated with a shopping mall promotion, and selecting one of the input images. Extracting information related to an object in an image by applying an object recognition algorithm to one image, obtaining an image color spectrum for the first image using a color analysis algorithm, information related to the object, and the image color The method may include setting a property value of the input text using a spectrum and generating a banner design associated with shopping mall promotion by merging the first image and text having the set property value.

In order to achieve the above object, a banner design generation device associated with shopping mall promotion according to another aspect of the present invention includes: an input unit receiving at least one image and text associated with a shopping mall promotion; An image analysis unit that extracts information related to an object in an image by applying an object recognition algorithm to one image, and obtains an image color spectrum for the first image using a color analysis algorithm, information related to the object, and the image It may include a text attribute value setting unit for setting the attribute value of the input text using a color spectrum and a merge unit for merging the first image and text having the set attribute value to generate a banner design associated with shopping mall promotion. have.

In order to achieve the above object, a banner design generation system associated with shopping mall promotion according to another aspect of the present invention provides a banner design request for banner design generation including at least one image and text associated with shopping mall promotion. A first device for receiving the banner design generation request from the client device and the client device transmitting to the generation device, and for use in the shopping mall promotion through image analysis of the at least one image included in the banner design generation request. Selecting an image, analyzing the selected first image, setting the attribute value of the input text according to the analysis result, merging the first image and the text having the set attribute value, and banner design associated with shopping mall promotion It may include a banner design generating device for generating a.

According to the banner design generation method, apparatus, and system associated with the promotion of the shopping mall of the present invention, by using artificial intelligence to automate the banner production process that relied only on human intuition and experience, a general shopping mall operator who has not studied design professionally They can also create banners with a high degree of completion, which will increase the competitiveness of small shopping mall operators.

1 is a block diagram showing a system to which a banner design generation method according to an embodiment of the present invention is applied,
Figure 2 is a block diagram showing the configuration of a banner design generating device according to an embodiment of the present invention,
3 is a flowchart illustrating a banner design generation method according to an embodiment of the present invention,
Figure 4 is a detailed flow diagram specifically showing the process of selecting a promotional image of the banner design generation method according to an embodiment of the present invention,
5 is a diagram showing input text and image information,
6A to 6C are exemplary views for explaining a process of analyzing the usability of the input image for promotional purposes,
Figure 7 is a detailed flow diagram showing the text positioning process of the banner design generation method according to an embodiment of the present invention,
8A to 8B are exemplary views showing an example of a grid-related mode determined according to a text size,
8C to 8E are exemplary views for exemplarily showing a process of determining a text position based on the user's eye position information;
9 is a detailed flow chart specifically showing a process of matching attribute values of text in a banner design generation method according to an embodiment of the present invention;
10A to 10B are exemplary views for explaining a process of matching attribute values of text and information obtained from a promotional image,
11 is a matching diagram for matching emotional vocabulary and image color in a banner design generation method according to an embodiment of the present invention;
12 is a table exemplarily showing a result of matching attribute values of text based on emotion information and complexity information extracted from an image in a banner design generation method according to an embodiment of the present invention,
13 is a conceptual diagram conceptually showing the entire process of deriving the final result through the image/text merging process from the first input image through the banner design generation method according to an embodiment of the present invention;
14 is a flowchart illustrating a process of determining a banner design layout using a grid system and/or Fibonacci sequence based ratio in a method of generating a banner design according to an embodiment of the present invention, and generating a banner design according to the determined layout;
15 is a view showing a breakpoint of a grid system used in a method for generating a banner design according to an embodiment of the present invention;
16 is a conceptual diagram for explaining a single grid and a single control point of a grid system used in a banner design generation method according to an embodiment of the present invention;
FIG. 17 shows the control points on the canvas in 1200 grid mode,
18 is a view showing a control center (center plane) and a component area in a 1200 x 740 grid mode;
19A to 19C are conceptual views illustrating features having different grid sizes according to device breakpoints in the horizontal direction;
20 is a view showing a device breakpoint in the vertical direction,
21 is a diagram showing a layout configuration in which a ratio based on Fibonacci sequence is combined with a grid system;
22 is an exemplary view showing a two-part layout in a horizontal direction and a three-part layout in a horizontal direction;
FIG. 23 is an exemplary view showing an exemplary shape of a banner design completed by inserting a promotional image and text into the two-part layout and the three-part layout of FIG. 22;
24 is a detailed flow chart specifically showing a process of inserting a banner design according to the size of a divided layout section of the method for generating a banner design according to an embodiment of the present invention.

The present invention can be applied to various changes and can have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail.

However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

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. For example, the first component may be referred to as a second component without departing from the scope of the present invention, and similarly, the second component may be referred to as a first component. The term and/or includes a combination of a plurality of related described items or any one of a plurality of related described items.

When an element is said to be "connected" or "connected" to another component, it is understood that other components may be directly connected to or connected to the other component, but there may be other components in between. It should be. On the other hand, when a component is said to be "directly connected" or "directly connected" to another component, it should be understood that no other component exists in the middle.

The terms used in the present 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, terms such as “include” or “have” are intended to indicate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, 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.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person skilled in the art to which the present invention pertains. Terms such as those defined in a commonly used dictionary should be interpreted as having meanings consistent with meanings in the context of related technologies, and should not be interpreted as ideal or excessively formal meanings unless explicitly defined in the present application. Does not.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In order to facilitate the overall understanding in describing the present invention, the same reference numerals are used for the same components in the drawings, and duplicate descriptions for the same components are omitted.

Creation of banner design

1 is a block diagram showing a system to which a method for generating a banner design according to an embodiment of the present invention is applied. 1, the banner design generation system according to an embodiment of the present invention may include a shopping mall 110-1 to 110-N, a communication network 120, and a banner design generation device 130.

Referring to FIG. 1, the shopping malls 110-1 to 110 to N may refer to a shopping mall selling products and/or services through the Internet, and include a server handling information related to the shopping mall or a terminal of an operator of the shopping mall. do. The terminal may include a computing device such as a smart phone or a PC.

According to an embodiment of the present invention, a plurality of shopping malls 110-1 to 110-N may interwork with the banner design generating device 130 through the communication network 120. Shopping mall (110-1 ~ 110-N) may include a device for requesting the creation of a banner design for the promotion of their own shopping mall, for this purpose, to create a banner design to request image design and/or text information to create a banner design It can be provided to the device 130. The image information and/or text information may be referred to as candidate images and/or text to be applied to banner design. Here, the text includes letters (including English, Korean, Korean, ..., etc.), numbers, symbols, formulas, etc., and may include all languages in a form that can be processed by a computing device. The image may include images in various formats such as JPG, PNG, BMP, GIF, JPEG, PDF, JPE, and TIFF. However, since the GIF file is a moving image file including multiple frames, it may be desirable to extract each frame and convert it to a JPG file.

According to an embodiment of the present invention, the shopping malls 110-1 to 110-N include the promotional purpose and/or promotional subject information in the banner design creation request including the image and text to the banner design generating device 130. Can provide. This promotion subject information can be used to perform various functions such as extracting key words from the banner design generating device 130 and selecting a promotional image. Promotional topics may be largely classified into at least two of discount/sale, new product/season, brand/category, free gift/point, SNS/member, anniversary/official and free delivery.

The communication network 120 is a network connecting the shopping malls 110-1 to 110-N and the banner design generating device 130, and may include a wired network and/or a wireless network (internet).

The banner design generation device 130 receives a banner design generation request including image and/or text information from at least one of the shopping malls 110-1 to 110-N through the communication network 120, and the received image and/or Or, it is a device that creates a banner design based on text. The banner design generation device may be implemented as a computing device. The computer device referred to herein can be implemented as a server-class computer terminal. The computer device may include an input device, a display device, a networking device, a hard disk, a memory for storing a program, and a processor for executing a program stored in the memory. However, it is not necessarily implemented as a server-class computer terminal.

After acquiring the received image and/or text, the banner design generating device 130 may determine whether the received image is suitable for publicity through image analysis. Judgment on whether it is suitable for publicity may be made by acquiring information related to the existence of the model in the image and the face of the model. Then, the device 130 sets a location where the text will exist in the image based on the image determined to be suitable for promotion. At this time, the attribute value of the text is determined by matching the attribute value of the text by referring to the object (objetc) information included in the image. Then, the input text may be converted to have the determined attribute value, and a final banner design may be generated by merging the converted text with an image determined to be suitable for promotion. Then, the generated banner design can be returned to the shopping malls 110-1 to 110-N that requested it.

The shopping malls 110-1 to 110-N may receive the banner design generated by the banner design generating device 130 and determine whether to use the shopping malls 110-1 to 110-N for publicity according to the operator's judgment. . The banner design generated according to the determination may be utilized in web pages, applications, and/or SNS associated with shopping malls 110-1 to 110-N.

According to another embodiment of the present invention, the terminal associated with the shopping malls 110-1 to 110-N downloads an application provided by the banner design generating device 130, and inputs an image performed by the banner design generating device 130. And/or generating a banner design from text to be performed in the terminal. That is, without a separate server device, a processor (not shown) of the terminal may execute the downloaded application to generate a banner design through image and/or text analysis.

2 is a block diagram showing the configuration of a banner design generating device according to an embodiment of the present invention. As shown in FIG. 2, the banner design generating device according to an embodiment of the present invention includes an input unit 210, an image analysis unit 220, a text attribute setting unit 230, a text processing unit 240, and image processing The unit 250 and the image/text merge unit 260 may be included. Each of these components may be integrally implemented as a single microprocessor, and/or a plurality of processors exist for each functional block, and may be implemented as a combination of a plurality of processors.

Referring to FIG. 2, the input unit 210 receives text information and image information for generating a banner design. The input text and images may be those obtained from individual shopping malls. The text may include a plurality of sentences. There may be multiple images. The input unit 210 may provide the input image to the image analysis unit 220 and the input text to the text attribute setting unit 230.

The image analysis unit 220 analyzes the input image. The input images are candidates that are likely to be selected as promotional images for promotional purposes (promotional images are some different concepts from banner design, and can be selected as suitable for promotion before final design creation). Therefore, it can be called a candidate image. Alternatively, it can be called an input image. In some cases, only text may be input and images may not be input from the shopping mall. In this case, the image input to the input unit 210 may be images selected by the device to match the text information among the images related to the shopping mall based on the input text information. These may be images including content (or object) close to the meaning of the key word through the meaning analysis of the individual words included in the text. In this case, for candidate image selection, image analysis and text analysis may be performed together.

The image analysis unit 220 analyzes the input image, selects a promotional image, and analyzes the promotional image. As an image analysis tool, a face recognition/tracking algorithm, an object recognition algorithm, a color extraction algorithm, and the like can be used.

The image analysis unit 220 receives at least one candidate image input to the input unit 210. Then, in order to determine the usability for promotional purposes, information of a model included in the corresponding image may be extracted and analyzed. The extracted model information may include (i) whether a face of the model exists in the image, (ii) face size information of the model, and/or (iii) face orientation information of the model. Then, information on whether the extracted model is present, face size information of the model, and face direction information of the model are compared with predetermined reference values, and only images having information included in the reference value are selected as a promotional image. In this embodiment, image analysis is performed mainly on a model (that is, a person who appears as a model) as a promotional image. In some cases, when a product to be promoted is less relevant to a person (for example, a pet product, etc.), the product and Relevant specific objects, animals, or specific buildings can be considered by replacing them with models. To this end, the device displays the user interface for inputting the content related to the promotion purpose (or subject) on the client terminal (terminal of the shopping mall operator), so that the image analysis target may be changed according to the promotion purpose. For example, in the case of buildings, objects such as windows and/or doors may be subject to image analysis.

After the determination of the usability as a promotional image is completed, the image analysis unit 220 may perform a task of selecting some of the promotional images that are usable through a separate process. For example, if it is determined that there are too many publicity images, an effort can be made to reduce this. At this time, among the multiple promotional images, the more clearly the model exists (for example, the size of the face of the model is relatively larger or vice versa, and/or the model's face orientation is more suitable for the promotion, etc.) ). Alternatively, an image more suitable for the purpose of publicity or the subject of publicity (or the target product for publicity) may be selected. For this, an image that clearly displays content related to a publicity purpose or a publicity topic may be selected.

When selection of a promotional image to be used for banner design is completed, the image analysis unit 220 extracts size information of the image, determines a grid system to be applied to the image, and extracts location information of the model's eyes. Then, it is determined to which side of the central reference surface of the grid system the extracted eye position. Then, the text location information may be determined to place the text in an area corresponding to the position of the eye in the image. The determined location information is provided to the text attribute setting unit 230.

In addition to determining the text position, the image analysis unit 220 may extract emotion information, complexity information, main color information, brightness information, and complementary color information associated with the image. Then, the related information is provided to the text attribute setting unit 230.

According to an embodiment of the present invention, the image analysis unit 220 may perform image preprocessing. Noise filtering on the image may be performed.

In addition, the image analysis unit 220 may include a layout determination unit (not shown) that performs processing related to setting of a grid system and layout. At this time, the layout determining unit determines the layout associated with the arrangement of the section in which the promotional image is inserted. According to an embodiment of the present invention, the layout determining unit may control the golden ratio based on the Fibonacci sequence to be applied to the layout. This will be described in more detail through FIGS. 14 to 23.

The text attribute setting unit 230 receives text information input from the input unit 210. At this time, if the text information is in a plurality of distinguishable sentences (or in a plurality of phrases), through morpheme and semantic analysis, the text is divided into sentences and/or classification units, and the text attribute for each separated text Perform setup tasks. That is, the batch attribute values are not necessarily applied to the text entered at one time.

The text attribute setting unit 230 may determine a text position value, text style, text thickness, text font, text color, and/or text highlight color for at least some of the input text. The text position value may be determined by directly receiving the position value from the image analysis unit 220. Alternatively, in some cases, the image analysis unit 220 receives only information on which side the model's eyes are compared to the central reference surface of the canvas, and the text attribute setting unit 230 directly positions the text based on the direction information. The value can be determined. The text attribute setting unit 230 receives at least one of image-related emotion information, complexity information, main color information, brightness information, and complementary color information from the image analysis unit 220, and the text style value and text thickness corresponding to the corresponding values Set values, text font values, text color values, and/or text highlight color values. The correspondence relationship between the result analyzed by the image analysis unit 220 and the text attribute value in the text attribute setting unit 230 may be a value previously stored in a database (not shown) or a memory (not shown). In order to make this correspondence more accurate, an image analysis result value according to input image and text information and a text attribute value may be generated as learning data, so that they can be learned through a deep learning model.

According to an embodiment of the present invention, the text attribute setting unit 230 may analyze the syntax and words of the text through the behavioral analysis and extract key words. Highlighting phrases can be selected through key words, so that the highlight color can be applied independently to the selected highlighting phrases. At this time, the key word is a key phrase of a message to be delivered by the user according to the promotional theme, and can be classified into a minimum unit form. For example, in the case of a discount, "50%" and "SALE" are key words in "50% SALE EVERYTHING", and in the case of new products/seasons, "2018" and "WINTER" are key words in "2018 WINTER NEW ARRIVALS". Can be

The text processing unit 240 converts the text according to the attribute value set by the text attribute setting unit 230.

The image processing unit 250 processes the selected promotional image. Processing includes processing such as image quality processing and area separation. This may be performed based on image analysis results and/or text information. For example, when the emotion contained in the text exists through the semantic analysis of the text (including the emotion vocabulary), the image may be converted into the corresponding color. Alternatively, processing to increase the quality of the image can be performed. For example, processing to increase the sharpness of the image may be performed. Alternatively, a process of generating a multiple image by dividing a single image into a plurality of images may be performed. Additionally, processing such as deleting all margins except the pure image area may be performed. The operation of the image processing unit 250 may be omitted in some cases.

The image/text merging unit 260 receives the image processed by the image processing unit 250 and the text converted into attribute values set by the text processing unit 240 and merges them appropriately to generate a final banner design.

3 is a flowchart illustrating a banner design generation method according to an embodiment of the present invention.

Referring to Figure 3, the device receives the input information (S310). It contains at least one image and text information. The device determines the usability of the input image as a promotion (S320). It is preferable that the judgment on this is performed based on a model existing in the image. When image analysis of the model is performed, the device may select a promotional image based on the analysis result (S320). With regard to the selection of promotional images, various methods such as a method of selecting an image with excellent readability and a method of selecting an image to be used for promotion based on the presence or absence of text may be applied in addition to the basic method.

After selecting the promotional image, the device sets the position of the text based on the image analysis result, in particular, the eye position of the model (S340). Then, among the image analysis results, the attribute values of the text are set in consideration of emotion analysis data, color information, and the like (S350). Then, a banner design is generated by merging the text having the set attribute value and the selected promotional image (S360).

FIG. 4 is a detailed flow chart specifically showing a process of selecting a promotional image of a banner design generation method according to an embodiment of the present invention.

Referring to FIG. 4, an image is first input for selection of a promotional image (S410). At this time, the input text may be in a form of dividing and arranging plural phrases, plural sentences, and sometimes a plurality of words. One image may be sufficient, or multiple images may be sufficient.

5 is a diagram showing input text and image information.

Referring to FIG. 5, the text has a theme related to winter collections of 17 and 18 years (17 18 PRE WINTER COLLECTION), and each word of "WINTER", "SNOW", and "IN SAPPORO" is different. It is placed in a row. In this case, words arranged in different columns may be interpreted as different phrases, and each distinguished phrase may have different text attribute values (which may include position values) as described above.

Multiple images can also be input. The input image may include various images, such as including text, not, including a model, not, displaying a model's face clearly, and not.

4, when text and image input is completed, the device scans the face of the model (S420 ). Then, it is determined whether it can be used for publicity (S430). The determination of usability at this time is performed based on whether all eyes, nose, and mouth of the model are visible. Whether or not pre-stored components (eye, nose and mouth) in the model's face are present in the image can be achieved through a face tracking technique, and used as a promotional image based on the coordinate values of the component. Judge the possibilities. If there is no one of the eyes, nose and mouth, the image can be discarded (S435).

Next, information related to the face size of the model is extracted (S440). This is done by identifying the contours of the face and calculating the area formed by the identified contours. The apparatus compares the extracted face size information with a preset face size reference value, and determines that an image having a reference value or more is usable (S450), and discards the other image (S435).

Then, information related to the face direction of the model is extracted (S460). The extracted data is data related to the face angle, and may include Roll degree, Pan degree, and Tilt information. That is, it includes left and right directionality of the face, head tilt information, and neck turning information. Then, it is determined whether or not the extracted left and right directional information of the face, the upper and lower tilt and neck turning information are included in the set reference values, respectively, and whether it is usable (S470). For example, if one side of the face does not come out, or if the head is turned too much or the head of the head is turned too much, it may be determined that there is no usability. The image with no usability is discarded (S435).

After this process, an image that has passed all of the usability-related processes may be selected as a promotional image (S480). However, if more images than the preset reference number are selected as promotional images, or if the user limits the number of images included in the banner image to a specific number, the number of promotional images may need to be reduced. In this case, the device may select a relatively larger face area and a more suitable face orientation. In addition, a method of converting a combination of features of an image for a plurality of criteria into a value (selection score), sorting the feature values in descending order, and selecting a top N (N is a natural number) image may be considered. Additionally, among the selected images, a method of selecting an image that matches the promotional topic, that is, an image including an object related to the promotional topic, may be considered.

According to another embodiment of the present invention, the device may consider using a feature point. A feature point is a set of all points that represent a specific property distinguished from the rest of the points, and is a point used when tracking and recognizing an object. This may include the outline of a particular object. In addition, it may include a point for distinguishing the background from the object. For example, a point that can be identified even if the shape, size, or position of an object changes, the point of view of the camera, and the point where the point can be easily found even when lighting changes may be considered as a feature point. It is extracted through the feature point extraction algorithm. In an embodiment of the present invention, it is also possible to consider a method in which an image including feature points above a reference value is selected as a promotional image because the feature points are clear.

6A to 6C are exemplary views for explaining a process of analyzing the usability of the input image for promotional purposes.

FIG. 6A is associated with step S450 of FIG. 4, and after calculating the area of the face based on the hot-recognition coordinates (face contour coordinates), the image having a large calculated area is selected.

Referring to FIG. 6B, the device may extract roll information, tilt information, and pan information in relation to step S470 of FIG. 4, and analyze the degree of face left and right direction, head tilt, and neck rotation to be included in the reference value. Select only images.

Referring to FIG. 6C, the device uses only the images 610-1, 610-3, and 610-4 that have passed the above selection-related criteria among the input images 610-1 to 610-4 as promotional images. do. The image 610-2 does not pass because all components in the model's face did not exist in the image. The image 610-2 that does not pass is discarded.

7 is a detailed flow chart showing in detail the text positioning process of the banner design generation method according to an embodiment of the present invention.

Referring to FIG. 7, the device acquires the selected promotional image through the process of FIG. 4 (S710 ). Then, the size of the obtained promotional image is extracted (S720). Then, it is determined whether the size of the image (eg, the width of the image) is greater than 600px (S730). When it is larger than 600px, a grid system of 24 grids is applied (S740). Otherwise, a grid system of 12 grids may be applied (S745). The grid system is a system defined by a control grid having a certain area as a set of a baseline grid and a baseline grid for defining a grid unit. In this regard, FIGS. 14 to 23 will be described below. This will be described in more detail.

8A to 8B are exemplary views illustrating an example of a grid-related mode determined according to text size.

Referring to FIG. 8A, the apparatus applies a grid system of a first mode having 12 grids (one grid may be 10px X 10px) horizontally when the image size, that is, the width of the image is less than 600px. can do.

Referring to FIG. 8B, when the width of the image exceeds 600px, the apparatus may apply a second mode grid system having 24 grids horizontally. That is, it may be desirable to determine the size of the width (or height) of the image in units of pixels to compare with a reference value, and to determine a corresponding grid system according to the comparison result.

Returning to FIG. 7 again, when the grid system to be applied to the image is determined, the device extracts eye position information according to the grid system (S750). The apparatus is performed using an algorithm for tracking the eye, and as a result, the coordinates of the left eye (LEFT_EYE) and the coordinates of the right eye (RIGHT_EYE) are obtained.

Then, the device compares the obtained coordinate values with the central reference plane of the grid system (S750). Determine in which direction it is positioned in comparison to the central reference plane (S760), when positioned on the left side, determines the insertion position of the text to the right (S770), and when located in the center, determines the insertion position of the text on the center, (S772), when located on the right side, the insertion position of the text may be determined to the left (S774).

8C to 8E are exemplary views for exemplarily showing a process of determining a text position based on the user's eye position information.

Referring to FIG. 8C, the apparatus sets the grids included in the 5 rows located in the center of the grid system of the 24 grids as the central reference plane of the image. Then, it is determined where the coordinates associated with the position of the eye are positioned relative to the central reference plane. In the embodiment of Fig. 8c, it may be determined that the central portion is disposed on top. Accordingly, the device may decide to place the text in the center portion, but the face portion may not be present, that is, the text may be placed in the lower portion of the center portion. At this time, it is preferable that the text is arranged by aligning the center of the text to the position of the eyes, that is, the center of both eyes.

Referring to FIG. 8D, in a grid system of 24 grids, the model's eyes are on the left side from the central reference plane. Thus, the device determines that the text is placed on the right side, but it is desirable to properly consider the horizontal and vertical arrangements in consideration of the relationship with the object in the image (in this embodiment, the arm portion of the model). At this time, it is good to place a margin on the outermost line of the right side, excluding the central reference plane, so that no text is placed there. This is because the text is not too readable in one direction of the image.

Referring to FIG. 8E, when the eyes of the model exist on the right side with respect to the central reference plane, the device determines that the text is disposed on the left side, as described above, in consideration of the relationship with the objects in the image, among the left regions of the image , It is preferable to properly consider the horizontal and vertical arrangement.

According to another embodiment of the present invention, when the model's eyes exist partially over the central reference plane, it is possible to determine the placement surface of the eye based on how much the central reference plane spans. For example, if 60% of the width between both eyes is on the central reference plane, it can be determined to exist in the central portion, and if 40% is crossed, it is determined that the eye is on the right or left side rather than the central reference plane. Can handle it. In some cases, if the placement of the eye is ambiguous relative to the central reference plane (i.e., almost similarly across both regions), the model is shifted to the entire, left or right side, one side (e.g., the left or right side of the image, Alternatively, it may be considered to determine the insertion position of the text after making it clear at the center).

9 is a detailed flowchart illustrating a process of matching attribute values of text in a banner design generation method according to an embodiment of the present invention.

Referring to FIG. 9, the device acquires the selected promotional image according to the process of FIG. 4 (S910), and analyzes the obtained promotional image. At this time, emotion information of the promotional image is extracted (S920). According to an embodiment of the present invention, emotion information obtainable from an image may be obtained from a facial expression, a behavior of a model, and/or a color of all or part of an image. Basically, emotion can be extracted based on the face recognition data obtained in the process of FIG. 7 above. It can be largely divided into emotion factors such as anger, contempt, disgust, fear, happiness, neutrality, sadness, and surprise. have. Each image may have scores for at least some of these eight emotion factors. Scores can be given from 0 to 1.0. The emotion information extraction algorithm for extracting the emotion factor can be evolved by matching the object or color in the image with the corresponding emotion score value to learn the machine. For example, it can be learned by generating (object-related information, emotion scores for each emotion factor) as learning data and inputting it into a deep learning model. The device may previously store text attribute values corresponding to the obtained emotion scores. For example, the primary classification of the font and the emotion data can be matched. To this end, an emotional matrix in the form of a regular polygon, which is generated by placing at least one emotion factor in one of up, down, left, and right directions, may be used.

According to an embodiment of the present invention, the first mode for defining the emotion factor that is the most exposed among the eight emotion factors as the emotion of the corresponding image, and the second mode for calculating the overall emotional value by combining the eight emotion factors Can be used. As a result of calculating the emotion value, when the emotion value is 0.5 or less, it can be regarded as not including emotion. Here, when the emotion values for the plurality of emotion factors are the same, emotions may be classified by smile values.

Then, a text style corresponding to the extracted emotion information is determined (S925). The style can be basically divided into the style of Serifs and Sans Serif. Sans serif is a style in which the thickness does not change from the start to the end of the horizontal/vertical stroke. Serif is a style with a decorative serif at the end of the stroke. That is, the emotion value and the text style according to a specific emotion factor may be matched in advance. For example, when the emotion factor of happiness and surprise is determined as the emotion of the corresponding image by 0.5 or more, the style may be determined as "serif", and when sorrow and expressionlessness dominate, "sans serif".

Additionally, in one embodiment of the present invention, the device secondaryly matches the adjective image scale data to the matching result of the text's basic style (serif, sans serif) and emotion data (e.g., happiness, sadness, no expression, surprise, etc.) By matching the emotion and color of the image model, the text font can be matched. That is, a plurality of fonts may exist in the serif, and at this time, the plurality of fonts are assigned to the adjective group of the image scale, and included in the adjectives (eg, cute, neat, stylish, ..., etc.) corresponding to the image. The style and font of the text can be determined by the fonts.

Next, when the text style is determined, the complexity information of the image is extracted (S930), and the thickness attribute value of the corresponding text is matched (S935). The higher the complexity of the image, the more highly readable text should be used, so if the complexity is high, bold text can be applied and if the complexity is low, thin text can be applied. That is, the complexity is divided into upper/middle/lower, and if "upper", "Bold", "mid", "Regular font", and "lower", "thin font" is applied. .

According to an embodiment of the present invention, the font of the text may be determined using complexity. For example, if the complexity of the image is high, a font having high readability can be controlled to use a font with high readability when the complexity of the image is low. Readability refers to the degree to which text is easily read, and readability refers to the degree to which the text is prominent in identifying and recognizing the shape of the font.

With regard to the method for calculating complexity, the device extracts tags (eg, related words) associated with objects included in the corresponding image, and calculates complexity based on the number of extracted tags. For example, if there are tags (including colors and adjectives) related to many related objects such as airplanes, clouds, buildings, people, mobile phones, heads, necks, arms, and legs, complexity increases. This can be done through a complexity extraction algorithm. In addition, a value related to the presence or absence of text included in an image, the number of objects included in the image, and the like may also be considered as a complexity calculation factor.

When the complexity is 26-30, it can be classified as "upper", 15-25 as "middle", and 14 or less as "lower". The criteria for classifying scores may vary depending on user settings. Then, among the plurality of fonts included in the previously determined text style, a font corresponding to each divided complexity may be determined as a font to be applied to the corresponding text.

After calculating the complexity, the main color information of the image is extracted (S940), and the corresponding text font attribute value is matched (S945). The apparatus extracts the primary colors arranged in the image through color ratio calculation of color spectrum data extraction color distribution data of the image analysis unit (not shown). Then, the extracted primary color can be used to determine the color and type of the text by matching the image sensitivity. At this time, it may be desirable to use the main color of the image as the main color of the banner design. That is, although color may directly match font attribute values, it can be used to extract image emotion based on color theory, and can be used to indirectly select fonts by melting in sensitivity and tag complexity.

Next, the device extracts the brightness information of the image (S950), and matches the color attribute value of the text (S955). Brightness represents the light and dark degree of color. This brightness can affect the color of the text. Accordingly, the device may determine the color of the text based on the text color attribute value table corresponding to the pre-stored image brightness value. If the brightness is high, the image is in a bright state, so the color of the text may be a dark color corresponding to it. When the brightness is low, since the image is dark, it is preferable to use a noticeable color (eg, red color) for the text.

Then, the device may extract complementary color information of the image (S960), and match the text highlight color attribute value (S965). The device is capable of obtaining complementary color information based on the data of the primary color code values (information including hex codes, RGB values and percentages) in the image analysis unit, and using the complementary color information, elements or elements to be emphasized on the banner screen. It can be used for elements that interact with it. Complementary color is calculated through │(H value of main color + 180°)-360°│ based on the main color data. Using the complementary color information, the color of the highlighted area of the text can be determined. In particular, if the background color of the corresponding image is the main color, the highlighting part of the text is made white or black so that it can be emphasized. The highlighted part of the text can be extracted through morphological analysis and semantic analysis.

In addition, in a method using a color spectrum, similar content may be used to represent related content or information centering on an area using a primary color. For example, it can be used as a main/sub color of banner design. In addition, by using a complementary color (split complementary color) it can be a combination with the complementary color, it can be a color match to the portion where the complementary color is utilized. With respect to the shade color tone, in order to secure readability between the background and the text, the shade color of the background and the text may be determined based on a reference contrast value between the colors. Shading color transformation is a color spectrum that adds white/black/gray to the main color, and can be used to create various levels of brightness without changing the color based on the main color. In particular, when it is necessary to increase the color contrast with the text color when used as a background color, a set value may be used. An alternative highlight color can be considered a combination of dark shades and light shades.

The shaded gradient is used to secure text readability by applying a gradient effect to the text area to protect the text on the image. For example, by using the reference value of the shaded color as an example, a reference point of a contrast value between the background and the text is determined to give a text area a gradient effect based on the primary color to ensure readability of the text. In this case, the dark end of the gradient may be defined to have a opacity value according to the situation, the center point of the gradient may be defined as a position 3/10 closer to the dark end, and the end of the gradient may be defined to have zero opacity.

10A to 10B are exemplary views for explaining a process of matching attribute values of text and information obtained from a promotional image.

Referring to FIG. 10A, the device may match a text style corresponding to the information extracted from the image. In the extracted information, since the neutral is highest at 0.619, and represents an emotion value higher than 0.5, the emotion of the image can be defined as expressionless, and in this case, the style of the text can be defined as the sans serif corresponding to the expressionless expression. Can be.

Referring to FIG. 10B, the device may extract various tag information from an image. According to the embodiment of FIG. 10B, the device can extract about 30 tag information such as “outdoor”, “persoon”, and “wearing”, and since it is more than 30, it can be determined that the complexity is “up”. have. In this case, the Bold body corresponding to the complexity “upper” may be determined as the thickness attribute value of the text.

11 is a matching diagram for matching emotional words and image colors in a method for generating a banner design according to an embodiment of the present invention.

Referring to FIG. 11, as described above, the device performs matching on two styles of sans serif and serif in one step and matching on an adjective image scale in two steps according to the emotion information of the image. Can be. At this time, the color distribution of the image may be used for the second-level emotion information.

The matching system of the color spectrum and the image scale of FIG. 11 may be used for matching for the two-step adjective image scale. Image scale is high by classifying adjectives into multiple depths. 1depth can be categorized as "soft", "static/cold", "dynamic/warm", "hard", in which 2depth is "cute", "neat", "stable", "natural" Etc. can be classified. 1depth and 2depth are related. For example, the 1depth of "soft" may include "cute", "neat", and "natural", and the 1depth of "hard" may include "heavy", "chic", "luxury", and the like. In this way, if even deeper emotion information is extracted according to the depth relationship of emotion, the device determines the font of the text as one of the fonts corresponding to the emotion information of the corresponding depth.

12 is a table exemplarily showing a result of matching attribute values of text based on emotion information and complexity information extracted from an image in a method for generating a banner design according to an embodiment of the present invention.

Referring to FIG. 12, the device extracts emotion data from an image using a face recognition data extraction algorithm. In this embodiment, since "happiness" is the highest at 0.992, it can be determined that "happiness" is the emotion of the image. The complexity is extracted through an image analysis tag data extraction algorithm, and the image can be determined to have 17 tags and a complexity of "medium".

Thus, in the embodiment of FIG. 12, the device determines the serif corresponding to happiness as the style of the text, and the complexity "medium" corresponds to the font with high readability, so the font with high readability is determined as the font of the text. Can be. In addition, the thickness of the text can be determined as "Regular" so as to correspond to the complexity "medium".

13 is a conceptual diagram conceptually showing the entire process of deriving the final result through an image/text merging process from an initial input image through a banner design generation method according to an embodiment of the present invention.

Referring to FIG. 13, the banner design generation device may receive an original image 1310 and text information for work. Then, based on the information related to the model of the image, it is determined the availability for publicity, and if it is determined that it is available, the eye position information of the image is extracted. Then, the location of the text is determined by comparing the extracted eye location information with the central reference plane. Then, it is possible to primarily extract emotion information from the image based on the facial expression of the model. Then, the style of the text is determined by matching the text style information corresponding to the emotion (eg, sans serif). Then, the complexity of the image is extracted, and the thickness of the text is determined using the corresponding text thickness value.

In addition, the device may acquire secondary emotion information through color information (eg, primary color information) of the image. At this time, secondary emotion information (which corresponds to adjective emotion information using an adjective image scale) can be obtained, and the font of the text can be determined by combining it with complexity information. Additionally, the basic color value of the text may be determined based on the brightness information of the image, and the complementary color corresponding to the main color of the image may be extracted from the color spectrum to determine the highlight color of the text.

Then, the banner design 1320 can be completed by converting text based on the attribute value of each determined text and inserting the converted text into the input promotional image 1310.

Determination of grid system and layout

14 is a flowchart illustrating a process of determining a banner design layout using a grid system and/or Fibonacci sequence based ratio in a method of generating a banner design according to an embodiment of the present invention, and generating a banner design according to the determined layout .

Referring to FIG. 14, the device may receive images and text (S1410). Then, an image usable for publicity is selected (S1420), and the layout can be determined in consideration of the grid system and the golden ratio based on the Fibonacci sequence. The layout considering the Fibonacci sequence-based ratio is a layout that combines the concept of the Fibonacci sequence, which represents an array of numbers where the sum of the preceding two numbers becomes the number immediately following. It takes into account the golden ratio of 1:1.618 and the golden ratio of 1:1.618 in at least one of the horizontal and vertical directions. At this time, it is not necessarily that the ratio must be exactly 1:1.618, and it may be desirable in that the ratio is adjusted to about 1:1.6 to 1:1.7. Then, a banner design is generated by inserting a promotional image and text according to the determined layout (S1440).

According to an embodiment of the present invention, in relation to a grid system used in a device, the device designes a banner based on a breakpoint that adjusts space on different screens and a control point based on a baseline grid. Produces The control point is a reference point for controlling the grid system, and may exist in the same number in the grid systems of different modes. The web grids that the device can use can create visual consistency between layouts in the design, where the number of grids can vary depending on the breakpoint scheme.

15 is a view showing a breakpoint of a grid system used in a method for generating a banner design according to an embodiment of the present invention.

Referring to FIG. 15, the device may use a grid of 6 grids for a display screen of 600px or less (can be called a canvas), and a grid of 12 grids for a display screen of 600px or more. For example, a 6-grid system may be applied to 480px and 600px, and a 12-grid system may be applied to 840px, 960px, 1280px, 1440px, and 1920px. The 6 grid system represents a system that divides the screen into 6 columns to form a grid, and the 12 grid system represents a system that divides the screen into 12 columns to form a grid.

In addition, according to one embodiment of the present invention, all components are adjusted to a square baseline grid for mobile, tablet, and desktop. Here, the baseline grid may be changed according to the production environment. In the case of a 1200px production environment, a unit of 1 grid may be defined as 10px X 10px. In addition, the control point is a set of baseline grids, which can be defined as areas of 5 grids horizontally and 3 grids vertically. It is desirable to have a margin of 2 grids between control points.

16 is a conceptual diagram illustrating a single grid and a single control point of a grid system used in a banner design generation method according to an embodiment of the present invention.

Referring to FIG. 16, the module grid is a sub-concept of the web grid, and based on 1200px, 1G may have a size of 10px X 10px. It can be variable depending on the horizontal and vertical size. 10X10 is possible, but 8X12 is also possible on devices with large vertical dimensions.

On one screen, the number of control points can be defined constant. One control point may have a width of 5G and a height of 3G (where G represents a grid). Basically, if a control point of a fixed number is placed on a grid array that can have a variable size, a basic grid system is created.

According to an embodiment of the present invention, as the most basic grid system, the entire screen is composed of 120G*74G, and a system in which 10px X 10px constitutes one grid can be used.

17 is a view showing a control point on a canvas in 1200 grid mode.

Referring to FIG. 17, the device may generate a grid system such that there is a preset number of control points on the canvas. Here, the canvas can arrange the control points so that there are 224 control points. One control point is 5G x 3G, and there is a 2G margin in the horizontal and vertical directions between each control point.

18 is a view showing a control center (center reference plane) and a component area in a 120 x 74 grid mode.

Referring to FIG. 18, the component area may mean an area in which an image and/or text is disposed. According to an embodiment of the present invention, a grid mode of 120G X 74G may be applied to 1200px X 740px, and at this time, 224 of 16 horizontal and 14 vertical control points may be arranged as described above. At this time, it is preferable that the central reference plane is an area in which four rows of control points in the center are present, and the component area in which the image and/or text is placed excludes the outermost control point as a margin and is an area excluding the central reference surface. . At this time, there may be one component area on the center surface and two (three in total) one on the left and right, and at this time, the left and/or right component area, except for the outermost control point, has a horizontal 5 located therein. It can be an area defined by 12 control points.

19A to 19C are conceptual views illustrating features having different grid sizes according to device breakpoints in the horizontal direction.

Referring to FIG. 19A, in a grid system according to an embodiment of the present invention, a horizontal size of one grid may have a size of “control breakpoint/120”. For example, a device with a width of 1200px is preferably a system based on a grid having a horizontal size of "1200/120=10px". Therefore, the size of the horizontal direction changes at 1080px, 960px, 840px, ... 240px, 120px.

The end point of the device may be slightly different from the control end point. In other words, the end point of the device is a value associated with the size of the display screen of the device, divided into units of about 100 to 110px, 1 section is 1200 to 1100px, 2 section has a range of 1099px to 990px, and 3 section is 989 It can have a value of ~880px.

The table is as follows.

Control breakpoint Horizontal (1 GRID) device Control breakpoint Horizontal (1 GRID) device width 1200px 10px ~1100px width 600px 5px 659px~550px width 1080px 9px 1099px~990px width 480px 4px 549px~440px width 960px 8px 989px~880px width 360px 3px 439px~330px width 840px 7px 879px~770px width 240px 2px 329px~220px width 720px 6px 769px~660px width 120px 1px Minimum unit

The pixel size can be converted according to the breakpoint of the device.

According to the embodiment of FIG. 19A, it is stipulated that a grid system based on a grid of 10 px horizontally is introduced with 100 px as a variable value in a section of 1100 to 1200 px based on 1100 (see FIG. 19B). However, in the case of 1099px, it may be desirable to introduce a grid system based on 9px (see FIG. 19C).

As such, in addition to the basic grid system defined by the control breakpoint, the grid system applied based on the device breakpoint may be different.

At this time, based on the existence of 120 grids based on 1200px horizontally, the number of grids before and after the breakpoint is variable. However, even at this time, as described above, the number of control points does not change. That is, the number of control points is fixed, and only the number of grids is variable. In addition, the grid system may be operated in a manner of maintaining a grid of 1200px when it is 1201px or more, and extending a margin area in a 1200px-based grid.

Moreover, after belonging to the corresponding breakpoint, it is preferable to align the entire control point area to the center and to process the remaining pixel values with a right margin if they fall oddly.

20 is a view showing a device breakpoint in the vertical direction.

The grid size in the vertical direction is calculated through "1G = control break point/74", unlike the grid size calculation in the horizontal direction in FIG. 19. Control breakpoints can be assigned to 1184px, 1110px, 1036px, 962px, 888px, ... 148px, 74px. The following table shows the control breakpoint and 1 grid pixel value per device.

Control breakpoint Vertical
(1 GRID) device Control breakpoint Vertical
(1 GRID) device height 1184px 16px ~1088px height 592px 8px 611px~544px height 1100px 15px 1087px~1020px height 518px 7px 543px~476px height 1036px 14px 1087px~952px height 444px 6px 475px~408px height 962px 13px 951px~884px height 370x 5px 407px~340px height 888px 12px 951px~816px height 296px 4px 339px~272px height 814px 11px 815px~748px height 222px 3px 271px~204px height 740px 10px 747px~680px height 148px 2px 203px~136px height 666px 9px 679px~612px height 74px 1px Minimum unit

Referring to FIG. 20, since 680px is a device breakpoint, a device having a height of 680 to 740px has a height of one grid in 74G mode of 10px. However, in a device having a height of 679px, it is preferable to apply a grid system having a grid size smaller in the form of a height of 9px in one grid in 74G mode.

In other words, according to the variable height value of the control point, the number of grids is varied before and after the breakpoint. In addition, the height of the control point may also be changed according to the size of the variable grid. In addition, after belonging to the corresponding breakpoint, if the entire control point area is center-aligned and falls odd, it is preferable to process the remaining pixel values as a bottom margin so that the image is not affected.

FIG. 21 is a diagram showing a layout configuration in which a Fibonacci sequence-based ratio is combined with a grid system.

Referring to FIG. 21, the apparatus utilizes a layout in which the concept of a Fibonacci sequence in which the sum of the preceding two numbers is the number immediately following the 120G x 74G system defined as a basic grid system is combined. In other words, the principle of balance and ratio is applied to the layout.

After constructing the canvas (physical sympathy for placing a banner design: the display screen of the device), the grid system is applied to the canvas. At this time, it is desirable to determine the size of the grid system and grids corresponding to the device endpoints.

Then, each section is divided by considering the golden ratio based on the Fibonacci sequence in the applied grid system. At this time, the division and division form of the section may be determined in consideration of the number of selected promotional images. For example, if there are two selected promotional images, a two-part layout, and three, a three-part layout can be used. The layout defines an arrangement of sections to which individual images are applied, and the defined sections may be applied with a Fibonacci sequence-based golden ratio for at least one of the horizontal and vertical directions.

In addition, after the section is divided, one section may be divided into a plurality of subsections, and individual images may be independently inserted into the divided subsections. For Fibonacci sequence based ratio, ie 1:1.618 can be applied. In some cases, a 1:1 ratio may also be applied.

22 is an exemplary view showing a two-part layout in the horizontal direction and a three-part layout in the horizontal direction.

Referring to FIG. 22, the device determines a two-part layout and a three-part layout according to the Fibonacci sequence based golden ratio. The two-part layout may consider the form of dividing the A section and the B section in a ratio of 1:1.618 in the horizontal direction. In addition, the form of dividing the A section and the B section in a ratio of 1:1.618 in the vertical direction may be considered, but in the normal case, since the device is a long rectangular shape in the horizontal direction, the number of considerations for dividing in the vertical direction is less You can. It is adaptively determined according to the environment of the device.

In the case of the three-part layout, the section A having the largest area has a horizontal size of 1:1.618 in the horizontal direction and a ratio corresponding to 1.618, but may have a square shape having horizontal and vertical lengths. Then, it is preferable to divide the area corresponding to the remaining 1 into two sections (the section having the larger area is the C section, and the section having the smaller area is the D section) in consideration of the 1:1.618 ratio in the vertical direction.

According to an embodiment of the present invention, a section A of a square may be divided into 1:1 in a horizontal direction to generate an E section, and may be arranged side by side with the B section. At this time, it is possible to consider placing two E sections together on the left side of the B section, and considering placing two E sections together on the right side of the B section, centering the B section, and having two E on each side. You may also consider arranging sections.

The apparatus may generate a multi-partition layout that is 4 to more severe by applying a golden ratio to the above-described two-part layout and three-part layout. The determination of the division layout may be set by the user how many division layouts to be used, or may be selected in consideration of the number of division sections automatically according to the number of promotional images to be selected. Or, by collecting data on the promotional image and the resulting banner design layout and inputting and learning it into a deep learning model, a more advanced layout can be derived.

With regard to the area of the divided layout, the device can be arranged alphabetically from the area having the largest width (width x height) (A section, B section, ..., Z section). The ratio of the number of grids in the horizontal and vertical directions of each section may represent the golden ratio. For example, in the case of sections A and B of the two-part layout on the upper left of FIG. 22, the horizontal size of the A section represents 74G, and the horizontal ratio of the B section represents 46G. This represents a ratio of 1.618:1. In addition, in the layout consisting of the A, C, and D sections at the upper left of the three-part layout, the ratio of the horizontal size of the A and C or A and D sections is also 74G to 46G, representing the golden ratio, and the C and D sections Also, the ratio of the vertical size of C is 46G for the C section and 28G for the D section, maintaining a ratio of 1.618:1.

FIG. 23 is an exemplary view showing an exemplary shape of a banner design completed by inserting a promotional image and text in the two-part layout and the three-part layout of FIG. 22.

As shown in FIG. 23, individual images are inserted into each section or subsection in the determined divided layout, and text can also be inserted.

At this time, the insertion of the text may be inserted together with the image after being inserted on the individual image, or after the insertion of the image is completed, the text may be inserted independently, and the text may be inserted in a form that spans a plurality of images. . This may be considered in various ways according to user settings, and in this case, text may be inserted beyond the image in consideration of the central reference plane of the grid system applied to the canvas and eye position information of multiple models in the multiple images. Control.

24 is a detailed flow chart specifically showing a process of inserting a banner design according to the size of a divided layout section of the method for generating a banner design according to an embodiment of the present invention.

24, the device generates a canvas (S2410). Then, the size information of the canvas (proportional to the screen size of the device) is extracted. Then, the device determines a division layout by determining a section suitable for the number and shape of divisions (S2420). In this case, there may be a preset layout, otherwise, the number of sections and the arrangement of sections may be determined adaptively according to the number of selected images.

According to another embodiment of the present invention, when a plurality of results are listed, a divided layout corresponding to a predetermined ratio may be applied according to a subject of publicity. For example, a probability that a two-part layout is applied to a promotional topic related to a discount/sale may be set to 30%. The applicability of the two-part layout is considered 60% for promotional topics related to new products/seasons, 70% for brands/categories, 5% for free gifts/points, 5% for SNS/members, anniversaries/ In relation to official events, it may be set to apply 20%, and 5% for free shipping. In addition, for the first type (A+C+D) of the three-part layout, 50% for discount/sale, 25% for new products/season, 13% for brand/category promotions, etc. And public relations themes have a connection, and corresponding applicability may be pre-set. Another applicability of the second form (B+E+E) may be considered. The degree of applicability of the split layout for each promotional topic can be arbitrarily changed through user settings. Fibonacci sequence-based ratios may be considered in the arrangement of sections belonging to the divided layout determined in this way.

After determining the split layout, the device calls the grid system by the device breakpoint.

Then, the size information (horizontal and vertical size) of the canvas extracted in the previous process is compared with the size information of the called grid system grid (S2430). If the difference between the horizontal size of the grid system and the horizontal size of the canvas is divided by the number of divisions based on the horizontal section of the corresponding section, the number related to the horizontal zoom is displayed. When the remaining value is generated, it is preferable to enlarge/reduce the quotient value from the right value to the left by 1px from the remaining value after expanding/reducing the quotient by the divided area. For example, in the case of size comparison for section A, if the horizontal size (SGw) of the grid system is 960, the horizontal size (Cw) of the canvas is 968, and the number of divisions is 2, (SGw-Cw)/fraction = ( 960-968)/2 results in "-4". At this time, since the negative number means enlargement, it is preferable to enlarge the transverse direction of the A section by 4 pixels.

Through the above process, when the size comparison in the horizontal and vertical directions for the grid system and the canvas size is completed, the device calculates each section size information based on the grid system to be applied to the canvas size. For example, when the size of the grid is 8px in the horizontal and vertical directions, the A section results in (Gw*74)(Gh*74)=592px*592px. Then, based on the previous comparison result, substituting the zoom value for each section, in the case of section A, the horizontal size of 596px is 592px + 4px. If you obtain the vertical size in the same way, 600px may appear. When the size of each section is determined, an individual image corresponding to each section is inserted (S2450).

With regard to the insertion of images per section, the device considers priorities in the image insertion order to put more important images in a large area section. In the foregoing, when selecting a promotional image, the relative scores of each selection factor may be comprehensively considered, or an image suitable for a specific promotional topic and/or purpose may be preferentially selected. For example, in relation to a new product/season, an image containing the product may be preferentially selected and inserted into the A section. Then, the device extracts and recognizes objects in the image through the object recognition algorithm, and if a product related to publicity among the recognized objects is included in the image, the image may be placed in priority. At this time, if there are multiple images containing the product related to the promotion, the image that discloses the product more clearly and more clearly may have a relative priority. If the product is not included, the overall score according to the factors related to the promotion image selection may be considered secondary.

With regard to the insertion of text, the text may be implemented in the layout in the form of being first inserted into the image, or may be implemented independently in the layout separately from the embedded image. At this time, when the key word of the input text and the result of object recognition in a specific image match, the corresponding text can be arranged on the matched image. At this time, if the semantic analysis results of the key words of the plurality of images and the input text are the same, the text may be displayed over the plurality of images, and the text may be controlled to be displayed only in the image having a higher priority.

Prior to the insertion of the image, in the relationship between the image and the canvas, if the image size is smaller than the canvas, zoom in, and conversely, if the image size is larger than the canvas, zoom out. At this time, an image sharpening effect is applied in order to prevent quality degradation due to image enlargement. In addition, it can be controlled so that the margin can be solved by inverting left and right. In addition, extracting the cross-section color of the image can also be considered. In addition, the primary color among image analysis data may be used to designate the background color of the entire canvas.

The system or device described above may be implemented with hardware components, software components, and/or combinations of hardware components and software components. For example, the systems, devices, and components described in embodiments include, for example, processors, controllers, arithmetic logic units (ALUs), digital signal processors (micro signal processors), microcomputers, and field programmable arrays (FPAs). ), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose computers or special purpose computers. The processing device may run an operating system (OS) and one or more software applications running on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For convenience of understanding, a processing device may be described as one being used, but a person having ordinary skill in the art, the processing device may include a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that may include. For example, the processing device may include a plurality of processors or a processor and a controller. In addition, other processing configurations, such as parallel processors, are possible.

The software may include a computer program, code, instruction, or a combination of one or more of these, and configure the processing device to operate as desired, or process independently or collectively You can command the device. Software and/or data may be interpreted by a processing device, or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodied in the transmitted signal wave. The software may be distributed over networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiments may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer readable medium. The computer-readable medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded in the medium may be specially designed and configured for the embodiments or may be known and usable by those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic media such as floptical disks. -Hardware devices specifically configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter, etc., as well as machine language codes produced by a compiler. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described by a limited embodiment and drawings, those skilled in the art can make various modifications and variations from the above description. For example, the described techniques are performed in a different order than the described method, and/or the components of the described system, structure, device, circuit, etc. are combined or combined in a different form from the described method, or other components Alternatively, even if replaced or substituted by equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (21)

In the banner design generation device in the banner design (banner design) generation method associated with shopping mall promotion,
Receiving at least one image and text associated with the promotion of the shopping mall;
Selecting a first image to be used for promotion of the shopping mall through image analysis of the input at least one image;
Analyzing the selected first image and setting an attribute value of the input text according to an analysis result; And
Generating a banner design associated with shopping mall promotion by merging the first image and text having the set attribute value,
The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting emotion information associated with the first image; And
And matching a text style attribute value corresponding to the extracted emotion information. The method of claim 1, wherein the step of selecting a first image for use for promotion of the shopping mall through image analysis of the input at least one image,
And determining whether it can be used for promotion based on information related to the face of the model in the at least one input image. According to claim 2, The step of determining whether it can be used for promotional purposes based on the model face in the at least one input image,
Determining usability based on whether there is at least one predetermined face component in the face of the model in the at least one input image;
Determining usability based on a model face size in the at least one input image; And
And determining availability based on the orientation of the model face in the at least one input image. The method of claim 1, wherein the step of selecting a first image for use for promotion of the shopping mall through image analysis of the input at least one image,
And determining whether it can be used for promotion based on the feature points of the model in the at least one input image. The method of claim 1, wherein the step of selecting a first image for use for promotion of the shopping mall through image analysis of the input at least one image,
A method of generating a banner design associated with shopping mall promotion, comprising the step of selecting an image containing the matched content as the first image based on a result of semantic analysis of at least some of the plurality of words included in the input text. According to claim 1, The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting the size of the first image;
Extracting an eye position of the model in the first image; And
And determining the location of the input text according to determining whether the extracted eye position is on the left side or the right side in comparison to the central reference plane of the first image. The method of claim 6,
Further comprising the step of determining the number of grids associated with the image segmentation by comparing the extracted image size with a reference value,
The central reference plane is a banner design generation method associated with shopping mall promotion that is set based on the determined number of grids. The method of claim 7, wherein comparing the extracted image size with a reference value to determine the number of grids associated with image segmentation comprises:
Shopping mall promotion comprising the step of determining one of a first mode of dividing into a grid having a row or column of a first value and a second mode of dividing into a grid having a row or column of a second value compared to the reference value; How to create an associated banner design. The method of claim 6, wherein the step of determining the position of the input text is determined by determining whether the extracted eye position is on the left side or the right side in contrast to the central reference plane of the first image.
Placing the input text on the right side of the first image in response to the extracted eye position being on the left side of the central reference plane;
Placing the input text in the center of the first image in correspondence with the extracted eye position being in the center of the central reference plane; or
And placing the input text on the left side of the first image in response to the extracted eye location being on the right side of the central reference plane, creating a banner design related to shopping mall promotion. delete In the banner design generation device in the banner design (banner design) generation method associated with shopping mall promotion,
Receiving at least one image and text associated with the promotion of the shopping mall;
Selecting a first image to be used for promotion of the shopping mall through image analysis of the input at least one image;
Analyzing the selected first image and setting an attribute value of the input text according to an analysis result; And
Generating a banner design associated with shopping mall promotion by merging the first image and text having the set attribute value,
The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting complexity information associated with the first image; And
And matching a text thickness attribute value corresponding to the extracted complexity information. According to claim 1, The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting main color information associated with the first image; And
And matching a text font attribute value corresponding to the extracted main color information. According to claim 1, The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting brightness information associated with the first image; And
And matching text color attribute values corresponding to the extracted brightness information. According to claim 1, The step of analyzing the selected first image and setting the attribute value of the input text according to the analysis result,
Extracting complementary color information associated with the first image; And
And matching a text highlight color attribute value corresponding to the extracted complementary color information. 15. The method of any one of claims 1, 11 to 14, wherein the step of setting the attribute value of the input text according to the analysis result by analyzing the selected first image,
Dividing the input text into a plurality of text elements; And
And individually matching attribute values of the separated text elements in association with the extracted information in association with the first image. In the banner design generation device in the banner design (banner design) generation method associated with shopping mall promotion,
Receiving at least one image and text associated with the promotion of the shopping mall;
Selecting a first image to be used for promotion of the shopping mall through image analysis of the input at least one image;
Analyzing the selected first image and setting an attribute value of the input text according to an analysis result; And
Generating a banner design associated with shopping mall promotion by merging the first image and text having the set attribute value; And
A method of generating a banner design related to shopping mall promotion, comprising extracting a key word based on the result of semantic analysis of at least one word included in the text and selecting a highlight phrase. The method of claim 16, wherein the processing of the first image based on the result of semantic analysis of at least one word included in the text comprises:
Extracting emotional words representing emotion from the input text; And
And converting the first image into a color corresponding to the emotion of the emotional word based on the extracted emotional word. In the banner design (banner design) generation device associated with shopping mall promotion,
An input unit that receives at least one image and text associated with the promotion of the shopping mall;
An image analysis unit that selects a first image to be used for promotion of the shopping mall through image analysis on the input at least one image, and performs image analysis on the selected first image to obtain an analysis result ;
A text attribute setting unit configured to set attribute values of the input text according to an analysis result of the selected first image; And
And a merging unit for merging the first image and text having the set attribute value to generate a banner design associated with shopping mall promotion,
The text attribute setting unit extracts emotion information associated with the first image, and a banner design generation device associated with shopping mall promotion matching a text style attribute value corresponding to the extracted emotion information. In the banner design generation device in the banner design (banner design) generation method associated with shopping mall promotion,
Receiving at least one image and text associated with the promotion of the shopping mall;
Extracting information related to an object in the image by applying an object recognition algorithm to the first image among the input images;
Obtaining an image color spectrum for the first image using a color analysis algorithm;
Setting an attribute value of the input text using the information related to the object and the image color spectrum; And
Generating a banner design associated with shopping mall promotion by merging the first image and text having the set attribute value,
The step of setting the attribute value of the text,
Extracting emotion information associated with the first image; And
And matching a text style attribute value corresponding to the extracted emotion information. In the banner design (banner design) generation device associated with shopping mall promotion,
An input unit that receives at least one image and text associated with the promotion of the shopping mall;
An image analysis unit for extracting information related to an object in the image by applying an object recognition algorithm to the first image among the input images, and obtaining an image color spectrum for the first image using a color analysis algorithm;
A text attribute value setting unit configured to set attribute values of the input text using the information related to the object and the image color spectrum; And
And a merging unit for merging the first image and text having the set attribute value to generate a banner design associated with shopping mall promotion,
A banner design generation device associated with shopping mall promotion that extracts emotion information associated with the first image and matches text style attribute values corresponding to the extracted emotion information by extracting emotion information associated with the first image. In the banner design (banner design) generation system associated with shopping mall promotion,
A client device that sends a banner design creation request including at least one image and text associated with the promotion of the shopping mall to the banner design generation device; And
Receive a request for generating the banner design from the client device, select a first image for use for promotion of the shopping mall through image analysis of the at least one image included in the banner design creation request, and select the selected image Includes a banner design generation device that analyzes the first image to set the attribute value of the input text according to the analysis result, and merges the first image and the text having the set attribute value to generate a banner design associated with shopping mall promotion. Ha,
The banner design generation device extracts emotion information associated with the first image, and a banner design generation system associated with shopping mall promotion matching a text style attribute value corresponding to the extracted emotion information.

Images