KR101608248B1 - Apparatus and method for classifying photograph - Google Patents

Abstract

We propose an apparatus and method for classifying photographs by comparing shapes of figures in photographs. The photograph classification apparatus according to the embodiment of the present invention extracts a main character from photographs, extracts a shape of a main character, calculates a shape similarity of a main character between photographs including main characters for each main character, The clustering can be performed using the shape similarity of the images to classify the photographs.

At this time, the photo classification apparatus extracts a predetermined person as a main person, extracts the predetermined person as a main person by confirming the appearance frequency in the pictures, extracts the person as a main person by checking the frequency of appearing as a person group in the pictures, The main character can be extracted according to the focus position.

Photo, shape, classification, similarity

Description

[0001] APPARATUS AND METHOD FOR CLASSIFYING PHOTOGRAPH [0002]

The following describes an apparatus and method for sorting photographs, and relates to an apparatus and method for sorting photographs by comparing the shapes of figures in the photographs.

With the development of digital camera technology, terminals with camera functions have become popular. The user can directly view the photographed images using the digital camera through the liquid crystal of the terminal, or upload them on a PC to edit and process them according to their preferences. Then, photographed photographic images can be provided to other people through convenient means such as electronic mail, and the spread of the photographic images is rapidly increasing.

Also, as the storage capacity of the terminal increases, the number of pictures photographed and stored by the camera has increased to an incalculable number.

However, in spite of the increased storage capacity, there are methods for searching stored pictures by searching pictures by shooting time and date, by sorting them by file name of pictures, or by searching by face recognition .

Conventional search methods have limitations in classifying and searching a large number of pictures. This limitation becomes more apparent when the photographs taken by a plurality of terminals are categorized rather than by the photographs taken by one terminal. For example, if you do not change the time on your device when you visit another country, the local time is not reflected in the picture. Accordingly, when photographs taken by a plurality of terminals are classified, an error may occur that the pictures taken on the same day are classified into different days.

That is, it is difficult to classify photographs as incidents according to the conventional search method.

The photo classification apparatus according to the present invention includes a main character extraction unit for analyzing photographs and extracting a main character having a high importance, a shape extraction unit for extracting a shape of the main character included in each of the photographs, A similarity calculation unit for calculating a shape similarity of the main character among the photographs including the main character for each main character and a classification unit classifying the photographs having high similarity through the clustering considering the shape similarity of the main characters into the same event .

At this time, the important person extracting unit may extract a predetermined person from among the persons existing in the photographs as the main person.

At this time, the important person extracting unit can identify the appearance frequency in the photographs through face recognition, and extract it as the main character.

At this time, the key person extracting unit can identify the frequency of appearing as a person group in the photographs through face recognition, and extract the key person as the main character.

At this time, the key person extraction unit may extract the key person by applying a weight according to the focus position of the photographs.

At this time, the shape extracting unit may extract at least one of clothing of the main character, hairstyle of the main character, accessory of the main character, and shoes of the main character.

At this time, it is possible to compute at least one of the outline of the extracted shape, the hue of the shape and the pattern (pattern) of the shape, and to calculate the sum by considering the predetermined weight.

According to a preferred embodiment of the present invention, there is provided a picture classifying method comprising the steps of: extracting a main character having a high degree of importance by analyzing photographs; extracting a shape of the main character included in each of the photographs; Calculating the shape similarity of the main character among the photographs including the person, and classifying the photographs having high similarity into the same event through clustering considering the shape similarity of the main characters.

At this time, the step of extracting the important person may extract a predetermined person among the persons present in the photographs as the main person.

At this time, the step of extracting the important person can be performed by checking the appearance frequency in the photographs through face recognition and extracting the main person.

At this time, the step of extracting the important person can be performed by checking the frequency of appearing as a person group in the photographs through face recognition, and extracting the person as the main person.

At this time, the step of extracting the key person may extract the key person by applying a weight according to the focus position of the photographs.

At this time, the step of extracting the shape of the main character may extract at least one of the clothing of the main character, the hairstyle of the main character, the accessories of the main character, and the shoes of the main character.

In this case, the step of calculating the shape similarity of the main character may be calculated by comparing at least one of the outline of the shape extracted from the similarity, the color of the shape and the pattern (pattern) of the shape,

The photograph classification apparatus according to the embodiment extracts a main character from photographs, extracts a shape of a main character, calculates the shape similarity of a main character between photographs including major characters for each main character, Classify photos through clustering based on similarity. Therefore, the user can easily sort and search pictures of the same event.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

The proposed embodiment relates to an apparatus and a method for classifying photographs by events by comparing similarities between shapes of figures in photographs. Hereinafter, a photo classification apparatus according to an embodiment will be described with reference to FIG.

Fig. 1 is a diagram showing a configuration of an apparatus for sorting photographs according to the proposed embodiment. 1, the proposed apparatus includes a control unit 110, a main character extraction unit 112, a shape extraction unit 114, a similarity calculation unit 116, a classification unit 118, and a storage unit 120 .

The storage unit 120 stores pictures to be classified.

The main person extraction unit 112 extracts a main character by analyzing the photographs stored in the storage unit 120.

The main character extraction unit 112 can extract a main character in consideration of the appearance frequency of the characters in the photograph through face recognition. Then, the main character extraction unit 112 can extract a main character in consideration of the appearance frequency of the character group, which is a combination of characters included in the same photograph, through face recognition. The main person extraction unit 112 can extract a main character by applying a weight according to the focus position of the photograph. The main person extraction unit 112 can extract a predetermined person as a main character among the persons existing in the photographs. Also, the main character extraction unit 112 can extract a main character by combining the above-described main character extraction methods.

The shape extraction unit 114 extracts the entire shape of the main character extracted by the main character extraction unit 112 in each photograph to be classified. There are clothes, hairstyle, accessories, shoes and so on.

The similarity calculation unit 116 calculates the shape similarity of the main character among the photographs including the main character for each main character extracted from the main character extraction unit 112. At this time, the degree of similarity is calculated by summing the outline of the extracted shape, the color and the pattern (pattern), and taking into account the weighted weights respectively.

The classifying unit 118 classifies photographs having high similarity degree into the same event through clustering considering similarity between shapes.

The control unit 110 controls the overall operation of the apparatus for sorting photographs and includes a main character extraction unit 112, a shape extraction unit 114, a similarity calculation unit 116, and a classification unit 118 according to the proposed embodiment. . That is, the control unit 110 may perform the functions of the main character extraction unit 112, the shape extraction unit 114, the similarity calculation unit 116, and the classification unit 118. In the proposed embodiment, these are separately constructed and described in order to separately describe each function. Therefore, when the product is actually implemented, the control unit 110 may be configured to process all of them, or only a part of them may be processed in the control unit 110.

Hereinafter, a method of comparing photographs by comparing the shapes of the persons in the photographs constructed as described above will be described below with reference to the drawings.

FIG. 2 is a flowchart illustrating a process of sorting photographs according to an embodiment of the present invention.

Referring to FIG. 2, in step 210, the proposed photo classification apparatus extracts a main person in the photos to be classified in step 212 when it detects the occurrence of a photo classification event for sorting the stored photographs by events.

The extraction of the main person can extract the main person in consideration of the appearance frequency of the person in the photograph through face recognition. In addition, the extraction of a main character can extract a main character in consideration of the appearance frequency of a character group, which is a combination of characters included in the same photograph, through face recognition. In addition, a key person can be extracted by applying a weight according to the focus position of a photograph. The extraction of the main person can extract the predetermined person as the main person among the persons existing in the photographs. Also, the extraction of the main person can extract the main person by combining the main person extraction methods described above.

If the main character is extracted in step 212, the photo classification device extracts the overall shape of the main character extracted from the respective photographs to be classified in step 214. There are clothes, hairstyle, accessories, shoes and so on.

Then, the photo classification apparatus calculates the shape similarity of the main person between the photographs including the main person by the main person extracted in step 216. At this time, the degree of similarity is calculated by summing the outline of the extracted shape, the color and the pattern (pattern), and taking into account weight values predetermined for each.

Then, in step 218, the photo classification apparatus classifies photographs having high similarity degree into the same event through clustering considering similarity between shapes.

The above-described methods 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, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. The medium may be a transmission medium such as an optical or metal line, a wave guide, or the like, including a carrier wave for transmitting a signal designating a program command, a data structure, or the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

While the present invention has been described with reference to the exemplary embodiments and the drawings, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a diagram showing a configuration of an apparatus for sorting photographs according to a proposed embodiment,

FIG. 2 is a flowchart illustrating a process of sorting photographs according to an embodiment of the present invention.

Claims (14)

A main character extracting unit for analyzing photographs and extracting a main character having a high importance; A shape extracting unit for extracting a shape of the main character included in each of the photographs; A similarity calculation unit for calculating a shape similarity of the main character among the photographs including the main character for each main character; And And a classifying unit for classifying photographs having high similarity degree into the same event through clustering considering the shape similarity of the main characters, The main person extracting unit, The appearance frequency in the photographs through face recognition is extracted and extracted as the main character Photo classification device. The method according to claim 1, The main person extracting unit, Extracting a predetermined person from among the persons existing in the photographs as the main person Photo classification device. delete The method according to claim 1, The main person extracting unit, The frequency of appearing as a person group in the photographs through face recognition is extracted and extracted as the main character Photo classification device. The method according to claim 1, The main person extracting unit, And extracts the main character by applying a weight according to a focus position of the photographs Photo classification device. The method according to claim 1, The shape extracting unit, Extracting at least one of the clothing of the main character, the hairstyle of the main character, the accessory of the main character, and the shoes of the main character Photo classification device. The method according to claim 1, Wherein the similarity- At least one of the outline of the shape extracted from the similarity degree, the hue of the shape and the pattern (pattern) of the shape are compared and the sum is calculated by taking into consideration the weights predetermined for each Photo classification device. A photograph classification apparatus comprising: Analyzing the photographs to extract a main character having a high importance; Extracting a shape of the main character included in each of the photographs; Calculating shape similarity of the main character among photographs including the main character for each main character; And Classifying photographs having high similarity degree into the same event through clustering considering the shape similarity of the main characters, The step of extracting the shape of the main character comprises: The appearance frequency in the photographs through face recognition is extracted and extracted as the main character How to sort photos. 9. The method of claim 8, The step of extracting the main character comprises: Extracting a predetermined person from among the persons existing in the photographs as the main person How to sort photos. delete 9. The method of claim 8, The step of extracting the main character comprises: The frequency of appearing as a person group in the photographs through face recognition is extracted and extracted as the main character How to sort photos. 9. The method of claim 8, The step of extracting the main character comprises: And extracts the main character by applying a weight according to a focus position of the photographs How to sort photos. 9. The method of claim 8, The step of extracting the shape of the main character comprises: Extracting at least one of the clothing of the main character, the hairstyle of the main character, the accessory of the main character, and the shoes of the main character How to sort photos. 9. The method of claim 8, Wherein the step of calculating the shape similarity of the main character comprises: At least one of the outline of the shape extracted from the similarity degree, the hue of the shape and the pattern (pattern) of the shape are compared and the sum is calculated by taking into consideration the weights predetermined for each How to sort photos.

Images