JP2018045582A - Image extraction system, image extraction method, and program for the image extraction system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image extraction system for extracting an image from a plurality of images, so as to obtain the proportion of the image along a theme represented by the plurality of images and to provide an image extraction method and a program for the image extraction system.SOLUTION: A place where the image is stored is specified (step 21) and an object existing in the image is decided for each of the plurality of images stored in the specified storage space (step 22). The existence ratio of the object in the plurality of images is calculated for each kind of the object (step 23) and the image is extracted from the plurality of images, so that the object existing in the image after extraction is in the calculated existence ratio (step 24). An album is created by using the extracted image (step 25).SELECTED DRAWING: Figure 2

Description

  The present invention relates to an image extraction system, an image extraction method, and a program for controlling a computer of the image extraction system.

  Because digital cameras and smartphones have become popular, it has become easy to shoot, so users often own many images. Since it is very difficult for a user to extract an image from a large number of images and create an album, there are some that automatically extract images to create an album.

  When creating an album automatically, if the number of photographic images taken by a particular individual increases or decreases, an unfair feeling occurs, so the number of appearances of each person is counted, There is one that prevents the bias (Patent Document 1). In addition, there is a technique that determines the importance of an image according to the selection frequency of each subject and determines the layout of the album (Patent Document 2). Furthermore, based on the total number of albums, when creating an album so as to have an appropriate number and layout (Patent Document 3), when creating a photo book of an idol group, the frequency of appearance of members constituting the group is determined. There are also things to be decided (Patent Document 4).

JP 2010-57073 A JP 2009-223764 Japanese Unexamined Patent Publication No. 2006-74590 Patent 5838752

  However, when creating an album, it is not possible to create an album according to the user's intention unless images included in the album are extracted according to the theme of the album. For example, if the album is parenting theme, there will be more images of the child, if the album is travel theme, the images of the destination will be increased, and if the album is wedding theme, the wedding will be remembered. Increasing the number of images creates an album that meets the user's intention. However, Patent Document 1 adjusts the number of appearances of a specific individual, Patent Documents 2 and 3 determine the album layout, and Patent Document 4 only determines the appearance frequency of members. . In any case, it is not considered to extract images from a plurality of images so that the ratio of images along the theme represented by the plurality of images is appropriate.

  An object of the present invention is to extract images from a plurality of images so that the ratio of images along the theme represented by the plurality of images is appropriate.

  An image extraction system according to the present invention includes an object determination unit that determines an object existing in an image for each of a plurality of images, and the ratio of the presence of an object determined by the object determination unit in a plurality of images for each type of object. A ratio calculating means for calculating, and an image extracting means for extracting images from a plurality of images so that an object existing in the image after extraction has the ratio of existence calculated by the ratio calculating means. Features.

  The present invention also provides an image extraction method suitable for an image extraction system. That is, in this method, the object determining means determines an object existing in the image for each of the plurality of images, and the ratio calculating means determines the object determined by the object determining means in the plurality of images for each object type. The image extraction means extracts an image from a plurality of images so that the objects present in the extracted image have the existence ratio calculated by the ratio calculation means. .

  The present invention also provides a computer-readable program for controlling the image extraction system and a recording medium storing the program.

  The object determining unit may further include an important object determining unit that determines an important object among the plurality of objects when the image includes a plurality of objects. In this case, it is preferable that the object included in the image is determined for each of the plurality of images based on the important object determined by the important object determining means.

  You may further provide the theme determination means to determine the theme which utilizes the image extracted by the image extraction means, and the object kind memory | storage means to memorize | store the kind of object linked | related with the theme. In this case, for example, the object determining means determines at least an object of the type of object stored in the object type storage means in association with the theme determined by the theme determining means.

  The object determination means can determine one or a plurality of objects existing in one image, and calculates an object evaluation value for each of the plurality of images, which is an object evaluation value representing the object likeness of each of the determined one or more objects. And an object evaluation value adding means for adding the object evaluation value calculated by the object evaluation value calculating means for each object type. In this case, for example, the ratio calculation means sets the ratio of the addition value for each object type obtained by the object evaluation value addition calculation means as the ratio of the existence of the object determined by the object determination means.

  The object determination means can determine one or a plurality of objects existing in one image, and calculates an object evaluation value for each of the plurality of images, which is an object evaluation value representing the object likeness of each of the determined one or more objects. And an object evaluation value adding means for adding the maximum object evaluation value for each object type among the object evaluation values calculated by the means and the object evaluation value calculating means. In this case, for example, the ratio calculating means sets the ratio of the added value for each object type obtained by the object evaluation value adding means as the ratio of the existence of the object determined by the object determining means.

  You may further provide the similar image detection means which detects a similar image among several images. In this case, for example, the object evaluation value calculation means calculates the object evaluation value of the object existing in the similar image detected by the similar image detection means relatively high.

  In the case of further comprising similar image detection means for detecting a similar image among a plurality of images, the ratio calculation means, for example, for the similar images detected by the similar image detection means, only one image among the similar images. , The ratio of the existence of the object determined by the object determining means is calculated.

  For example, the similar image detecting unit detects a similar image for each image having the same object determined by the object determining unit.

  You may further provide the album production | generation means which produces | generates an album using the image extracted by the image extraction means.

  For example, the album generation unit generates an album by pasting the image extracted by the image extraction unit on the album so as to have a size corresponding to the ratio calculated by the ratio calculation unit.

  In the spread page of the album, the album generating means generates an album by pasting the image extracted by the image extracting means on the album so as to have a size corresponding to the ratio calculated by the ratio calculating means. Also good.

  In the entire album, the album generation means, for example, selects the image extracted by the image extraction means so that the area ratio is in accordance with the ratio of the existence of objects for each object type calculated by the ratio calculation means. Paste into an album to create an album.

  For example, the album generation means generates the album by pasting the image extracted by the image extraction means on the album according to the priority according to the ratio of the existence of the object for each object type calculated by the ratio calculation means. .

  The album generating means generates an album using the images extracted by the image extracting means, for example, according to the image capturing order.

  According to the present invention, an object existing in an image is determined for each of a plurality of images. For each determined object type, the ratio of the presence of objects in multiple images is calculated, and images are extracted from multiple images so that the objects present in the extracted images have the calculated existence ratio. Is done. By extracting the images in this way, the ratio of the images along the theme represented by the plurality of images becomes appropriate.

It is a block diagram which shows the electric constitution of an image extraction system. It is a flowchart which shows the process sequence of an image extraction system. Shows images stored on the memory card. It is an example of an image. It is an example of an image. It is an example of an image. It is an example of an image. It is an example of an object table. It is an example of the spread page which comprises an album. It is a flowchart which shows an object determination processing procedure. It is an example of an object table. It is a flowchart which shows an object determination processing procedure. It is an example of an object kind storage table. It is a flowchart which shows the calculation process procedure of the ratio of presence of an object. It is an example of an object table. It is a flowchart which shows the calculation processing procedure of an object evaluation value. It is a flowchart which shows the calculation process procedure of the ratio of presence of an object. It is an example of the spread page which comprises an album. It is an example of the spread page which comprises an album.

[Example 1]
FIG. 1 shows an embodiment of the present invention, and is a block diagram showing an electrical configuration of an image extraction system 1.

  The overall operation of the image extraction system 1 is controlled by a CPU (Central Processing Unit) 2.

  The image extraction system 1 includes a keyboard 3 and a mouse 4 as input devices for giving commands and the like to the image extraction system 1. Commands and the like from the keyboard 3 and mouse 4 are given to the CPU 2 via the input control device 5. The image extraction system 1 includes a memory 8 for temporarily storing data, a hard disk 7 and a hard disk drive 6 for accessing the hard disk 7, a display device 10 and a display control device for controlling the display device 10. 9 is included.

  The image extraction system 1 further includes a memory card reader / writer 11 for accessing the memory card 12, a CD-ROM drive 13 for accessing a compact disc read only memory (CD-ROM), and the Internet. A communication device 15 for connecting to the other network is included.

  A CD-ROM 14 (recording medium) in which a program for controlling operations to be described later is stored is read by the CD-ROM drive 13, and the read program is installed in the image extraction system 1. Such a program may be stored in a portable recording medium such as a memory card 12 other than the CD-ROM 14, or may be installed in the image extraction system 1 in advance. Alternatively, the program may be received by the communication device 15 via a network such as the Internet, and the received program may be installed in the image extraction system 1.

  FIG. 2 is a flowchart showing a processing procedure of the image extraction system 1.

  The processing procedure shown in FIG. 2 is to extract an image from a plurality of images stored in a specified storage location and generate an album using the extracted images.

  On the display screen of the display device 10, a storage location designation image for designating a storage location where a plurality of images are stored is displayed. Using the storage location designation image, the user can designate storage locations such as the hard disk 7, the CD-ROM 14, the memory card 12, and a storage area on the network. When a USB (Universal Serial Bus) memory (not shown) can be connected to the image extraction system 1, a USB memory can be designated as a storage location. The user uses the keyboard 3 or mouse 4 to specify the storage location of the image used for generating the album (step 21). In this embodiment, it is assumed that the memory card 12 is designated by the user as an image storage location.

  FIG. 3 shows an image stored in the memory card 12 designated as the image storage location.

  Assume that the memory card 12 stores 500 image files represented by image file names from IMG001 to IMG500. However, the image file names need not be consecutive. For example, the user goes to travel with the memory card 12, and 500 image files are stored in the memory card 12 by shooting at the travel destination.

  4 to 7 show images 31 to 34 represented by some of the 500 image files stored in the memory card 12.

  FIG. 4 is an example of the image 31 represented by the image file name IMG001. The image 31 is an example of an image including a mountain, a river, and the sun.

  FIG. 5 is an example of the image 32 represented by the image file name IMG002. The image 32 is an example of an image including flowers and clouds.

  FIG. 6 is an example of the image 33 represented by the image file name IMG003. The image 33 is an example of an image including a person, a mountain, a cloud, and a flower.

  FIG. 7 is an example of the image 34 represented by the image file name IMG004. The image 34 is an example of an image including food and flowers.

  Returning to FIG. 2, for each of a plurality of images stored in a storage location (memory card 12 in this embodiment) designated by the user, an object existing in the image is determined by the CPU 2 (object determination means). (Step 22). A shooting period may be designated, and among the images stored in the storage location, the object object existing in the image captured during the shooting period may be determined by the CPU 2. The type of object to be determined is determined in advance (it may not be determined in advance), and in this embodiment, scenery, flowers, people, and dishes are determined in advance as objects. When an object existing in the image is recorded in the header included in the image file, the recorded object may be determined as an object existing in the image.

  In this embodiment, an object (subject) included in the image is detected, and an object (target represented by the image) existing in the image is determined by the CPU 2 from the detected object. In the case of the image 31 shown in FIG. 4, objects of mountains, rivers, and sun are detected from the image 31, and the objects existing in the image 31 are “landscape” from the detected objects of mountains, rivers, and sun. It is determined that The detected object (mountain, river, sun) itself may be determined as an object existing in the image. Similarly, in the case of the image 32 shown in FIG. 5, objects of flowers and clouds are detected from the image 32, and the objects existing in the image 32 are “flowers” from the detected objects of flowers and clouds. Determined to be “landscape”. In the case of the image 33 shown in FIG. 6, objects of people, mountains, clouds, and flowers are detected from the image 33, and objects existing in the image 33 are detected from the detected objects of people, mountains, clouds, and flowers. Are determined to be “person”, “landscape” and “flower”. Further, in the case of the image 34 shown in FIG. 7, the object of food, drink and flower is detected from the image 34, and the object existing in the image 34 is “cooking” from the detected object of food, drink and flower. "And" flower ". For the images with the image file names IMG005 to IMG500 other than the image file names IMG001 to IMG004, the objects existing in the image are determined in the same manner.

  FIG. 8 is an example of an object table.

  The object table shown in FIG. 8 stores objects determined as objects existing in the image represented by the image file names corresponding to the image file names from IMG001 to IMG500.

  Returning to FIG. 2, when an object existing in the image is determined, the ratio of the presence of the object in the plurality of images is calculated by the CPU 2 (ratio calculation means) for each determined object type (step 2). twenty three). For example, among 500 images, there are 250 “landscape” type objects, 150 “flower” type objects, 50 “person” type objects, and “cooking” type objects. Assuming that there are 50 images, the ratio of the “landscape” type object, the “flower” type object, the “person” type object, and the “cooking” type object is 5: 3: 1: 1

  When the ratio of the presence of the object in the plurality of images is calculated, the CPU 2 (image extraction means) converts the image from the plurality of images so that the object existing in the extracted image has the calculated ratio of the presence. Extracted (step 24). The number of images to be extracted may be determined by the user or may be determined in advance. When the number of images to be extracted is 100, the determined object is 50 “landscape” images, the determined object is “flower” 30 images, and the determined object is “person” 10 images, 10 images with the determined object “cooking”, and 500 images represented by the image file stored in the memory card 12 are extracted. Even if the number of extracted images is other than 100, the CPU 2 uses the image file stored in the memory card 12 so that the objects existing in the extracted image have the calculated existence ratio. Extracted from the 500 images represented.

  If there are multiple types of objects in the image, such as the objects “flower” and “landscape” corresponding to the image file name IMG002 stored in the object table of FIG. 8, the extracted images are duplicated. If not, one of the plural types of objects may be determined as the object of the image. Further, an object evaluation value representing the object likeness of an object existing in the image may be calculated by the CPU 2 (object evaluation value calculating means), and an image may be extracted from those having a high object evaluation value. Images may be extracted so that the interval between image capturing dates and times is equal, or an image evaluation value indicating whether the image itself is good (whether the brightness of the image, the degree of focus, etc. are appropriate). The image may be extracted from the one with a high image evaluation value, or the image may be extracted at random.

  An album (electronic album) is generated by the CPU 2 (album generating means) using the extracted image (step 25).

  FIG. 9 is an example of a spread page constituting the generated album.

  Images 31-36 and the like are extracted, and the extracted images 31, 32, and 34 are pasted on the left side page 41 of the spread page that constitutes the album, and the extracted images 33, 35, and 36 are the spread pages that constitute the album. Affixed to the right page 42. The size and position of the image attached to the album may be determined in any way.

  According to the first embodiment, the ratio of existence of objects in a plurality of images is calculated, and images are extracted from the plurality of images so that the objects existing in the extracted image become the calculated ratio of existence of objects. Therefore, it is possible to extract an image reflecting the theme represented by the ratio of the presence of a plurality of types of objects in a plurality of images.

[Example 2]
10 and 11 show the second embodiment.

  FIG. 10 is a flowchart showing the object determination processing procedure (the processing procedure of step 22 in FIG. 2).

  An object (mountain, river, flower, person, dish, etc.) included in the image is detected by the CPU 2 (step 51), and it is determined by the CPU 2 whether a plurality of objects are included in the image (step 52). When the object included in the image is recorded in the header included in the image file, the object may be read out. If the image includes a plurality of objects (YES in step 52), an important object is determined by the CPU 2 (important object determining means) from the plurality of objects (step 53). Whether it is an important object is determined by the position in the image (for example, the closer to the center of the image, the more important the object is determined), the object evaluation value (the higher the object evaluation value, the more important the object is determined), the brightness, the degree of focus, etc. Image evaluation (objects with higher image evaluation are judged as important objects), objects that match a specific theme (eg travel, wedding, etc.) (eg travel, mountains, rivers, weddings, etc.) , Dresses, rings, bouquets, etc.). If the image does not include a plurality of objects (NO in step 52), the process in step 53 is skipped.

  If the image contains multiple objects, the objects present in the image are determined based on the important objects. If the image does not contain multiple objects, the objects are detected. It is determined that the object exists in the image (step 54).

  FIG. 11 is an example of an object table obtained by the processing procedure of FIG.

  In the object table shown in FIG. 8, a plurality of objects are defined corresponding to the image file names IMG002 to IMG004, but in the object table shown in FIG. 10, there is one object corresponding to the image file names IMG002 to IMG004. It is determined as one object (an object that is determined based on an important object). When one object is determined in the image, the CPU 2 determines the existence ratio of the object in the plurality of images for each object type as described above using the determined object (FIG. 2). Step 23).

  Since one object is determined for each image, the subsequent image extraction process is relatively easy.

  According to the second embodiment, since the important object is determined, it is possible to extract a more natural image reflecting the theme represented by the ratio of the presence of the important object when focusing on the important object in a plurality of images. .

[Example 3]
12 and 13 show a third embodiment.

  FIG. 12 is a flowchart showing an example of the object determination processing procedure (processing procedure of step 22 in FIG. 2).

  In this embodiment, the theme of the extracted image is determined by the CPU 2 (theme determination means) (step 61). By analyzing a plurality of images stored in the designated image storage location, the theme represented by the plurality of images may be determined, or several themes may be displayed on the display screen of the display device 10. The theme may be determined using the mouse 4 (theme determination means), or the theme may be determined by the user using the keyboard 3 (theme determination means).

  FIG. 13 is an example of an object type storage table.

  The object type storage table shown in FIG. 13 stores a plurality of object types corresponding to the theme. Themes include travel, weddings, and entrance ceremony. The travel theme stores the types of objects such as landscapes, flowers, and dishes. The wedding theme stores the types of objects such as dresses, rings, bouquets, and dishes. The entrance ceremony theme stores object types such as school bags and school gates. Similarly, for other themes, object types corresponding to the themes are stored. Such an object type storage table is stored in the memory 8 (object type storage means) in advance.

  Returning to FIG. 12, when the theme is determined, as shown in FIG. 13, the type of the object associated with the determined theme is at least determined as an object existing in the image (step 62). For example, when “travel” is determined as the theme, “landscape”, “flower”, “cooking”, and the like stored in correspondence with the travel are plural objects as existing in the image as described above. It will be determined in each of the images. The objects determined in this way are stored in the object table shown in FIG. 8 or FIG.

  According to the third embodiment, since the object is determined for the object of the type stored in association with the determined theme, the theme represented by the ratio of the presence of the object when focusing on the stored type of object is displayed. Reflected and more natural images can be extracted.

[Example 4]
14 and 15 show a fourth embodiment.

  FIG. 14 is a flowchart showing a processing procedure (processing procedure of step 23 in FIG. 2) for calculating the ratio of existence of an object in a plurality of images for each type of object. FIG. 15 corresponds to FIG. 8 and is an example of an object table.

  As shown in step 22 of FIG. 2, when an object existing in the image is determined, the object is stored in the object table corresponding to the image file name representing the image, as shown in FIG.

  In this embodiment, for each determined object, an object evaluation value representing the object likeness is calculated by the CPU 2 (object evaluation value calculating means) (step 71). A large amount of data representing the shape, color, etc. of a general material determined to be an object is stored in the memory 8, and the object evaluation value is calculated by calculating the degree of coincidence with the data. For example, in the case of an object of “landscape”, a lot of data representing mountains, rivers, plateaus, lakes, seas, clouds, sun, sky, and their arrangements are stored. The CPU 2 (object evaluation value calculation means) calculates an object evaluation value representing the object likeness of “landscape”. When the object evaluation value is calculated for an object existing in the image, the calculated object evaluation value is stored in the object table shown in FIG. The object evaluation value is set to a value of 100 or less, but such a value is not necessarily required.

  Referring to FIG. 15, the object existing in image 31 (see FIG. 4) represented by image file name IMG001 is determined as “landscape”, and its object evaluation value is calculated as “100”. Objects existing in the image 32 (see FIG. 5) represented by the image file name IMG002 are determined as “flower” and “landscape”, the object evaluation value of “flower” is calculated as “90”, and “landscape” The object evaluation value is calculated as “20”. Similarly, the objects existing in the image 33 (see FIG. 6) represented by the image file name IMG003 are determined as “person”, “landscape”, and “flower”, and the object evaluation value of “person” is “80”. The object evaluation value of “landscape” is calculated as “70”, and the object evaluation value of “flower” is calculated as “25”. Objects existing in the image 34 (see FIG. 7) represented by the image file name IMG004 are determined as “cooking” and “flower”, the object evaluation value of “cooking” is calculated as “100”, and “flower” The object evaluation value is calculated as “30”.

  Returning to FIG. 14, when the object evaluation value is calculated, the object evaluation value is added by the CPU 2 (object evaluation value adding means) for each object type (step 72). For example, in the case of an object of “landscape”, all the calculated object evaluation values of “landscape” are added to obtain an addition value of the object evaluation value of “landscape”. Similarly, the addition value of the object evaluation value of “flower”, the addition value of the object evaluation value of “person”, and the addition value of the object evaluation value of “cooking” are obtained.

  The ratio of the added value of the object evaluation value for each object type obtained in this way is set as the ratio of the existence of the object (step 73). For example, the addition value of the object evaluation value of “landscape” is N1, the addition value of the object evaluation value of “flower” is N2, the addition value of the object evaluation value of “person” is N3, and the addition of the object evaluation value of “cooking” When the value is N4, the ratio of the objects present in the plurality of images stored in the storage location is “landscape”: “flower”: “person”: “dish” = N1: N2: N3: N4 Become.

  When the proportion of objects present in a plurality of images is determined, as shown in step 24 of FIG. 2, the objects present in the extracted images are stored in the storage location so that the calculated proportion of existence exists. Images are extracted from the plurality of stored images.

  In the above embodiment, the object evaluation value is added for each object type to calculate the object evaluation value addition value. However, the maximum object evaluation value among the object evaluation values for the object determined to exist in the image is calculated. The value may be calculated by the CPU 2 (object evaluation value calculation means) for each type of object.

  For example, since the only object existing in the image 31 (see FIG. 4) represented by the image file name IMG001 is “landscape”, the maximum object evaluation value is “100” of the “landscape”. Objects existing in the image 32 (see FIG. 5) represented by the image file name IMG002 are “flower” and “landscape”, and the object evaluation value of “flower” is calculated as “90”. The object evaluation value is calculated as “20”. The maximum object evaluation value is “90” as the object evaluation value of “flower”. Similarly, the objects existing in the image 33 (see FIG. 6) represented by the image file name IMG003 are “person”, “landscape”, and “flower”, and the object evaluation value of “person” is “80”. The object evaluation value of “landscape” is calculated as “70”, and the object evaluation value of “flower” is calculated as “25”. The maximum object evaluation value is “80”, which is the object evaluation value of “person”. The objects existing in the image 34 represented by the image file name IMG004 (see FIG. 7) are “dishes” and “flowers”, and the object evaluation value of “dishes” is calculated as “100”. The object evaluation value is calculated as “30”. The maximum object evaluation value is “100” of “cooking”.

  In this way, the maximum object evaluation value among the object evaluation values calculated for each image is added by the CPU 2 for each type of object evaluation value (object evaluation value addition means), and the addition value for each type of object is can get. Object evaluation values other than the maximum object evaluation value are not subject to addition. For example, the objects existing in the image 32 (see FIG. 5) represented by the image file name IMG002 are “dishes” and “flowers”, and the object evaluation value of “dishes” is calculated as “100”. The object evaluation value of “30” is calculated as “30”. Among these object evaluation values, the object evaluation value “100” of “cooking”, which is the maximum object evaluation value, is the object evaluation value for each object type. The object evaluation value “30” for “flower” is not subject to addition.

  The ratio of the added value of the object evaluation value for each object type obtained in this way is set as the ratio of the existence of the object (step 73), as described above.

  According to the fourth embodiment, since the ratio of the added value for each object type is the ratio of the existence of the object, both the type of object having the maximum object evaluation value and the object evaluation value of that type in a plurality of images. This makes it possible to extract a more natural image that reflects the theme expressed by the ratio of the presence of the object when focusing on.

[Example 5]
FIG. 16 shows a fifth embodiment, and is a flowchart showing an object evaluation value calculation processing procedure (processing procedure of step 71 in FIG. 14).

  In this embodiment, a similar image is detected by the CPU 2 (similar image detecting means) among a plurality of images stored in the storage location designated by the user (step 81). The detection of similar images is performed based on the degree of coincidence by matching processing between images, but may be performed by other methods.

  When there is a similar image, the object existing in the similar image can be regarded as important by the user. For this reason, when a similar image is detected, the object evaluation value is calculated by the CPU 2 so that the object evaluation value of the object determined as an object existing in the similar image becomes relatively high (step 82). Since the object evaluation value is relatively high, the rate at which images including the object are extracted increases. The calculated object evaluation value may be multiplied by a coefficient larger than 1 to increase the object evaluation value.

  As described above, the calculation of the object evaluation value such that the object evaluation value becomes relatively high is not limited to the case where the maximum object evaluation value is added to obtain the addition value. May be carried out in any case where the addition value is obtained for each type of object. In addition, when an object evaluation value is calculated when an important object is determined in step 53 in FIG. 10, even if the object evaluation value of an object existing in a similar image is relatively high in the calculation of the object evaluation value, Good.

  According to the fifth embodiment, since the ratio of the added value for each object type is the ratio of the existence of the object, the ratio of the existence of the plurality of types of objects in the plurality of images and the intention of adopting more similar images. This makes it possible to extract more natural images reflecting the theme expressed by the ratio of the presence of objects when focusing on both.

[Example 6]
FIG. 17 shows the sixth embodiment and is a flowchart showing the calculation processing procedure of the ratio of the existence of objects (processing procedure of step 23 in FIG. 2).

  Also in this embodiment, a similar image is detected by the CPU 2 (similar image detecting means) among a plurality of images stored in the storage location designated by the user (step 91).

  If there is a similar image, as described with reference to FIG. 16, the object existing in the similar image can be regarded as important by the user, but conversely, it is considered important. Sometimes not. For this reason, with respect to similar images, the ratio of the presence of objects in a plurality of images is calculated by the CPU 2 for each type of object, out of similar images (step 92). For example, when it is determined that the image 31 shown in FIG. 4 and the image 33 shown in FIG. 6 are similar images, a plurality of images for each type of object for either one of the image 31 or the image 33 are displayed. The ratio of the existence of the object is calculated (step 92). Thereby, even when there are a plurality of similar images, an image can be extracted without being influenced by the plurality of similar images.

  According to the sixth embodiment, for similar images, the ratio of the existence of an object is calculated for only one image. Therefore, the presence of a plurality of types of objects and similar images in a plurality of images are obtained by using only one image. It is possible to extract more natural images reflecting the theme expressed by the ratio of the existence of objects when focusing on both of the intentions to be emphasized.

[Example 7]
FIG. 18 shows a seventh embodiment, which is an example of a spread page constituting an album, and corresponds to FIG.

  Images 31 and 32 are affixed to the left page 101, and images 33 to 36 are affixed to the right page 102. The album may be an electronic album displayed on the display screen of the display device 10 or a paper album.

  The image pasted on the spread page of the album shown in FIG. 18 has a size corresponding to the ratio calculated as described above on the spread page of the album. For example, assuming that the ratio of the presence of objects in a plurality of images is “landscape”: “flower”: “person”: “cooking” = 5: 3: 1: 1, “landscape” exists as an object. When the size of the image 31 and the image 35 is 5, the size of the images 32 and 36 in which “flower” exists as an object is 3, and the image 33 and “cooking” in which “person” exists as an object are objects. The size of the existing image 36 is 1, and images 31 to 36 are pasted on the spread page of the album.

  In the example shown in FIG. 18, the size of the image is adjusted on the spread page constituting the album, but the image size may be adjusted on the entire album instead of the spread page constituting the album. . When the size of the image is adjusted in the entire album, the extracted image is divided for each object existing in the image, and the ratio of the total area of all the images divided for each object is as described above. The size is set according to the calculated ratio. For example, assuming that the ratio of the presence of objects in a plurality of images is “landscape”: “flower”: “person”: “dish” = 5: 3: 1: 1, the total number of images extracted is Assuming 40 images and one album is generated from these 40 images, there will be 20 images with “landscape” as objects, 12 images with “flowers” as objects, There are four images in which "" exists as an object, and there are four images in which "food" exists as an object. Then, the total area of 20 images where “landscape” exists as an object, the total area of 12 images where “flower” exists as an object, and the total area of 4 images where “person” exists as an object. The ratio of the area and the total area of the four images in which “cooking” exists as an object is 5: 3: 1: 1, like the ratio of the existence of objects in a plurality of images. The extracted 40 images are pasted on the album.

  Such image size adjustment is performed by the CPU 2 (image size adjusting means) to generate an album. Needless to say, when a paper album is generated, the CPU 2 controls the printer.

  As described above, the album may be generated by pasting the extracted image on the album so as to have a size corresponding to the calculated ratio of the presence of the object.

  According to the seventh embodiment, since the image is pasted on the album so as to have a size corresponding to the calculated ratio, the existence of the object when focusing on the existence ratio of the plurality of types of objects in the plurality of images. By reflecting the ratio on the size of the album, the album can be generated more naturally.

[Example 8]
FIG. 19 shows Example 8, which is an example of a spread page constituting an album, and corresponds to FIG. 9 and FIG.

  In the spread page shown in FIG. 19, images 31 and 35 are pasted on the left page 111, and images 37 to 39 are pasted on the right page 112.

  In the example shown in FIG. 19, the extracted images are affixed to the album according to the priority order according to the ratio of the presence of the object calculated as described above. For example, assuming that the ratio of the presence of objects in a plurality of images is “landscape”: “flower”: “person”: “cooking” = 5: 3: 1: 1, the priority order according to these ratios In the extracted images, images with the “landscape” object are continuously pasted on the album, and then images with the “flower” object are continuously pasted on the album. Images with the “person” object are continuously pasted on the album, and then images with the “cooking” object are continuously pasted on the album. Also, five images with the “landscape” object are pasted on the album in succession, and then three images with the “flower” object are pasted on the album. An image in which an object exists may be pasted on a single album, and an image in which a “cooking” object exists may be pasted on a single album. Thereafter, the pasting of images may be repeated in the same manner.

  In this embodiment, when an image is pasted on a certain spread page, if the number of images pasted on the spread page is smaller than the sum of the numerical values indicating the existence ratio of the object, Affix images with a priority.

  For example, in the example shown in FIG. 19, as a result of adjusting the numerical value indicating the existence ratio so that it becomes 10 when all the existence ratios of the objects are added, 5 images including the object “landscape” are obtained. Attaching images to the album in succession, then pasting 3 images with the “flower” object on the album, then pasting an image with the “person” object onto the album, Further, a priority order is created in which an image having a “cooking” object is pasted on one album.

  Then, since the spread page of the example shown in FIG. 19 is attached with five images, five images with the “landscape” object are continuously pasted on the album, and the other pages As a result, the image in which the object exists is not pasted.

  Furthermore, the extracted images may be attached to the spread pages constituting the album according to the priority order corresponding to the calculated proportion of the objects. For example, assuming that the ratio of the presence of objects in a plurality of images is “landscape”: “flower”: “person”: “cooking” = 5: 3: 1: 1, In a two-page spread, the five images with the “landscape” object are pasted on the two-page spread, the three images with the “flower” object are pasted on the two-page spread, and the “person” A single image in which an object exists may be pasted on a spread page, and a single image in which a “cooking” object exists may be pasted on a spread page. An image may be similarly attached to other spread pages, and an extracted image may be attached to another spread page with a different configuration. Further, the pasting order in the spread pages may be in accordance with the ratio of the presence of objects in a plurality of images.

  Further, the extracted image may be attached to the album in accordance with the image shooting order. Images may be pasted according to the image capturing order in the entire album, or images may be pasted according to the image capturing order on the spread pages constituting the album.

  In the above-described embodiment, “person” is not distinguished as an object existing in the image, but “person” is included among a plurality of images stored in the designated storage location. 2 is divided into a non-human image that does not include “person”, and the processing from step 22 onward in FIG. 2 is performed on the non-human image to extract images from a plurality of non-human images. Good. Further, the plurality of images stored in the designated storage location are divided into a plurality of image groups according to the imaging order, and the plurality of images included in the divided image groups are described with reference to FIG. The subsequent processing may be performed.

  According to the eighth embodiment, the extracted image is pasted on the album according to the priority according to the calculated ratio. Therefore, even when the pasting area of the album is limited, the image including the object with the higher priority is prioritized. Is pasted into the album.

  1 image extraction system, 2 CPU, 3 keyboard, 4 mouse, 5 input control device, 6 hard disk drive, 7 hard disk, 8 memory, 9 display control device, 10 display device, 11 memory card reader / writer, 12 memory・ Card, 13 CD-ROM drive, 14 CD-ROM, 15 Communication device, 31-39 Image, 41 Left page, 42 Right page, 101 Left page, 102 Right page, 111 Left page, 112 Right page, IMG001-IMG005 Image file name

Claims (17)

An object determination means for determining an object existing in the image for each of the plurality of images;
For each type of object, a ratio calculation means for calculating the ratio of the existence of the object determined by the object determination means in the plurality of images, and an object existing in the extracted image are calculated by the ratio calculation means. Image extracting means for extracting an image from the plurality of images so as to obtain a proportion of the presence
Image extraction system with The object determination means is
An important object determining means for determining an important object among the plurality of objects when the image includes a plurality of objects;
Determining an object included in the image for each of the plurality of images based on the important object determined by the important object determining means;
The image extraction system according to claim 1. Theme determining means for determining a theme using an image extracted by the image extracting means;
Object type storage means for storing the type of object associated with the theme,
The object determination means is
Determining at least an object for the type of object stored in the object type storage means in association with the theme determined by the theme determination means;
The image extraction system according to claim 1. The object determination means is
One or more objects can be determined in one image,
An object evaluation value calculating means for calculating an object evaluation value representing the object-likeness of each of the determined one or more objects for each of the plurality of images, and an object evaluation value calculated by the object evaluation value calculating means, An object evaluation value adding means for adding for each object type;
The ratio calculation means is
The ratio of the added value for each object type obtained by the object evaluation value adding means is the ratio of the existence of the object determined by the object determining means.
The image extraction system according to claim 1. The object determination means is
One or more objects can be determined in one image,
Among the object evaluation values calculated by the object evaluation value calculation means, the object evaluation value calculation means for calculating the object evaluation value representing the object likeness of each of the determined one or more objects for each of the plurality of images. , Further comprising an object evaluation value addition means for adding the maximum object evaluation value for each object type,
The ratio calculation means is
The ratio of the added value for each object type obtained by the object evaluation value adding means is the ratio of the existence of the object determined by the object determining means.
The image extraction system according to claim 1. A similar image detecting means for detecting a similar image among the plurality of images;
The object evaluation value calculation means is:
Calculating a relatively high object evaluation value of an object present in the similar image detected by the similar image detecting means;
The image extraction system according to claim 4 or 5. A similar image detecting means for detecting a similar image among the plurality of images;
The ratio calculation means is
For the similar image detected by the similar image detecting means, the ratio of the existence of the object determined by the object determining means is calculated for only one image among the similar images.
The image extraction system according to claim 1. The similar image detection means
A similar image is detected for each image having the same object determined by the object determining means.
The image extraction system according to claim 7. Album generating means for generating an album using the image extracted by the image extracting means;
The image extraction system according to any one of claims 1 to 8, further comprising: The above album generation means
Creating an album by pasting the image extracted by the image extracting means on the album so as to have a size corresponding to the ratio calculated by the ratio calculating means;
The image extraction system according to claim 9. In the spread page of the album, the album generating means pastes the image extracted by the image extracting means to the album so as to have a size corresponding to the ratio calculated by the ratio calculating means. Generate,
The image extraction system according to claim 9. In the whole of the album, the album generating means is extracted by the image extracting means so as to have an area ratio corresponding to the ratio of the existence of each object type calculated by the ratio calculating means. Paste the images you created into the above album to create an album,
The image extraction system according to claim 9. The above album generation means
The album extracted by pasting the image extracted by the image extracting means on the album according to the priority order according to the ratio calculated by the ratio calculating means for each type of the object,
The image extraction system according to claim 9. The above album generation means
In accordance with the image capturing order, an album is generated using the images extracted by the image extracting means.
The image extraction system according to claim 9. The object determining means determines an object existing in the image for each of the plurality of images,
The ratio calculation means calculates the ratio of the existence of the object determined by the object determination means in the plurality of images for each type of the object,
The image extracting means extracts images from the plurality of images so that the objects existing in the extracted image have the presence ratio calculated by the ratio calculating means;
An image extraction method comprising: A computer readable program for an image extraction system,
For each of multiple images, determine the objects that exist in the image,
For each object type, calculate the proportion of the determined object in the multiple images,
A program of an image extraction system that extracts an image from the plurality of images so that objects existing in the image after extraction have a calculated existence ratio.   17. A recording medium storing the program according to claim 16.

Images