JP4534571B2 - Face detection device - Google Patents

Description

The present invention relates to an apparatus for detecting a face area related to a face image of a person from an input image and registering a face image with an appropriate angle of view from the face area in an electronic file.

Conventional information search devices, information search systems, and servers create a face image phone book by registering a face image attached to an e-mail or an image input from an external connection device in an electronic file ( For example, Patent Document 1).

JP2002-92013

JP-A-9-309137

JP-A-11-110529

JP 2000-322588 A

Conventional information retrieval apparatuses, information retrieval systems, and servers have a problem that when creating a face phone book, a single photograph image taken by only one person must be obtained from an e-mail or an external connection apparatus. It is also possible to use the group photo as it is in the phone book. In this case, however, there is a problem that it is difficult to determine who is calling from the group photo. Furthermore, even if it is a single photo image, if the subject is photographed small or taken at the edge of the image, the background becomes redundant, and if the image is registered as it is, the image registered in the electronic file becomes difficult to see. There was a problem.

The present invention has been made to solve the above-described problems. A face image is automatically extracted from a group photo image or a single photo image, and an electronic file in which individual face images are registered even from the group photo image is obtained. An object of the present invention is to make it easy to see a registered face image by cutting out a face area with an appropriate angle of view.

An electronic file generation device according to the present invention includes an image input device that inputs an image, a face image database in which face image data is registered, and an image inside a search block that moves on the image input by the image input device . A face area determination device that compares a feature amount with a feature amount of a face image registered in the face image database, and determines whether the image inside the search block is a part or all of the face area, and the face area When a plurality of images determined to be part or all of the face area by the determination device are determined to be the same person's face area, the plurality of images determined to be the same person's face area are integrated, and an input image And an overlapping area integrating device for generating a first face area that is an area in which a human face image captured in the input image exists , and the face image registered in the feature amount of the input image and the face image database Features By comparing a face region detecting device of the first face region detection number of people, based on the properties of such a face image on the first face area detected by the face region detecting device, and registers the electronic file A face for extracting a face image relating to the second face area obtained by obtaining an appropriate angle of view and extending the first face area vertically and horizontally with a predetermined extension rate so that the entire face is included in consideration of the face characteristics An area extraction apparatus and an electronic file registration apparatus that registers a face image extracted by the face area extraction apparatus in an electronic file are configured.

An electronic file generation device according to the present invention includes an image input device that inputs an image, a face image database in which face image data is registered, and an image inside a search block that moves on the image input by the image input device . A face area determination device that compares a feature amount with a feature amount of a face image registered in the face image database, and determines whether the image inside the search block is a part or all of the face area, and the face area When a plurality of images determined to be part or all of the face area by the determination device are determined to be the same person's face area, the plurality of images determined to be the same person's face area are integrated, and an input image And an overlapping area integrating device for generating a first face area that is an area in which a human face image captured in the input image exists , and the face image registered in the feature amount of the input image and the face image database Features By comparing a face region detecting device of the first face region detection number of people, based on the properties of such a face image on the first face area detected by the face region detecting device, and registers the electronic file An appropriate angle of view is obtained, and a facial image of the second face area is extracted by extending the first face area vertically and horizontally at a predetermined extension rate so that the entire face is included in consideration of the face characteristics. A face area extracting device and an electronic file registering device for registering a face image extracted by the face area extracting device in an electronic file are configured to automatically extract a face image from a group photo image or a single photo image. Extracting and making it possible to create an electronic file in which individual face images are registered even from a group photo image, making it easy to see the registered face image by cutting out the face area with an appropriate angle of view including the hair and chin effect A.

Embodiment 1 FIG.
The configuration of the electronic file generation device 100 in the embodiment of the present invention will be described with reference to FIG. In FIG. 1, an electronic file generation apparatus 100 includes an image input apparatus 1 for inputting an image, a face image database (hereinafter referred to as a face image DB) 7 in which face image data is registered, and an image input by the image input apparatus 1. Is compared with the data registered in the face image DB 7 to detect the first face area, which is the area where the face image of the person copied from the image is present, for the number of persons present in the image. Based on the characteristics of the face area detecting device 2 and the face image of the first face area detected by the face area detecting device 2, an appropriate angle of view for registration in the electronic file is obtained, and the first face area A face area extracting device 3 for extracting a face image relating to the second face area, and an electronic file registering device 4 for registering the face image extracted by the face area extracting device 3 in an electronic file. Yes.

The face area detection device 2 generates a multi-scale image generation device 5 that generates an image having a plurality of resolutions obtained by enlarging or reducing the input image, and an image in which the face is illuminated according to the illumination conditions of the input image. An image normalization device 6 that normalizes an image, a face image DB 7 in which a face image feature quantity in which the features of the face image are quantified is stored, and a rectangular (24 (pixels) × 24 rectangle, for example, ) And the feature amount of the image inside the search block 14 that moves on the input image and the face image feature amount registered in the face image DB 7 are compared, and the image inside the search block 14 is a part of the face region. Alternatively, the face area determination device 8 that determines whether or not the face area is determined, and the search block 14 are sequentially moved to determine face area determination, and a plurality of fragment areas determined to be part or all of the face area of the same person For the number of people in the image The overlap region integrating unit 9 generates a first face region, in being configured.

The face area extraction device 3 calculates an appropriate angle of view for calculating an appropriate angle of view when the first face area detected by the face area detection device 2 is registered in the electronic file based on the characteristics of the face image. The apparatus 10 includes a face area extracting apparatus 11 that extracts a face image related to the second face area detected based on the calculated appropriate angle of view from the input image.

When a plurality of face images are extracted, the electronic file registration device 4 displays a list display device 40 that displays a list of face images for several persons or all of the plurality of persons, and the displayed face images. And a face image selection device 41 that allows a user to select a face image to be registered in an electronic file from the list.

Next, the operation of the electronic file will be described with reference to FIGS. FIG. 2 is a flowchart for explaining the operation of the electronic file generation apparatus 100 of the present invention. FIG. 3 is a flowchart for explaining the operation of the face area detection apparatus 2. FIG. 4 is a flowchart for explaining the operation of the face area extracting apparatus 3. FIG. 5 is a flowchart for explaining the operation of the electronic file registration apparatus 4. FIG. 6 is a diagram for explaining the operation of the multiscale image generation apparatus 5. FIG. 7 is a diagram for explaining the operation of the face area determination device 8. FIG. 8 is a diagram for explaining the operation of the overlapping area integrating device 9. FIG. 9 is a diagram for explaining the operation of the appropriate angle-of-view calculation apparatus 10. FIG. 10 is a diagram for explaining the electronic file registration device 4.

First, the operation of the electronic file generation apparatus 100 will be described with reference to FIG. An image is input in the image input device 1 (step S1-1). For example, in the case of a mobile phone, the input image covers all images that can be input to the mobile phone, such as an image attached to an e-mail, an image taken by a camera, and an image input from an external storage device. Next, the face area detection device 2 detects the first face area from the input image (step S1-2), the face area extraction apparatus 3 obtains an appropriate angle of view when registering in the electronic file, and the first face A face image relating to the second face area expanded from the area is extracted (step S1-3), and the extracted face image is registered in the electronic file in the electronic file registration device 4 (step S1-4).

Next, the operation of the face area detection device 2 will be described with reference to FIGS. 3, 6, 7, 8 and 20 . Specifically, consider a case where an image indicated by 12 in FIG. 6 is input.

When the image is input, the face area detection device 2 uses the multi-scale image generation device 5 to generate a plurality of images in which the input image is enlarged or reduced as shown by 13 in FIG. 6 (step S2-1). ). Then, the following processing is repeated for each obtained scale image (step S2-2). The feature amount of the image inside the search block 14 having a fixed size shown in 14 of FIG. 7 is compared with the feature amount of the face image (face image learned by machine learning or the like) in the face image DB 7 in advance. In step S2-4, it is determined whether or not the face is a part or all of the face (a part or the whole of the face area).

When the search block size is fixed, the multi-scale image generation device 5 can be used to expand / reduce the image, that is, to enlarge or reduce the face image to match the search block size.

At this time, the pixel value inside the search block 14 is normalized in order to solve the problem that the feature amount of the face changes according to the change of the illumination condition (step S2-3). As a normalization method, for example, there are a method using dispersion of pixel values inside the search block 14, a method using a pixel value average value, a method using a histogram, and any method may be used. Here, it demonstrates as using the method of normalizing using dispersion | distribution as following Formula.

ここでV(x、y)は座標値(x、y)における正規化後の画素値である。v(x、y)は座標値（x、y）におけるオリジナルの画素値である。varは探索ブロック１４内側の画素値の分散で

Here, V (x, y) is a pixel value after normalization in the coordinate value (x, y). v (x, y) is the original pixel value at the coordinate value (x, y). var is the variance of the pixel values inside the search block 14

is there. After normalizing the pixel values inside the search block 14 using the above formula, the face image feature quantity stored in the face image DB 7 is matched with the feature quantity inside the search block 14 (step S2-4). As a result of matching, it is determined whether or not the image on the inside of the search block 14 is a face area (step S2-5). If it is a face area, the coordinate values of four points of the face area rectangle in the current image are obtained ( Step S2-6), the area is stored (step S2-7). The matching process between the face area search block and the learning data is performed as follows, for example. First, the face image data registered in the face image database 7, for example, learned the facial features, and are quantified based on RectangleFeature shown in FIG. 20. RectangleFeature is a value obtained by subtracting the total pixel value of the hatched rectangular area from the total pixel value of the white rectangular area inside the search block.

ｐｖ，ｎｖは重み

マッチング処理が完了すると、顔領域検出装置２は探索ブロック１４を例えば、一画素ずつ移動させ（ステップＳ２−８）、スケール画像全体を操作し、画像にある人数分上記処理をした後、次のスケール画像を処理する（ステップＳ２−９）。以上の処理を全てのスケール画像に対して行った後（ステップＳ２−１０）、重複領域統合装置９において、検出領域のうち重なり合った領域を統合する（ステップＳ２−１１）。ここで、重複領域統合装置９をさらに詳しく説明する。顔領域検出装置２は、前述の通り探索ブロック１４を例えば、１画素ずつ移動させながら顔領域判定を行うため、図８の１５のように、顔の付近では複数の探索ブロック１４が顔領域と判定することにより顔領域判定装置８で記憶した顔領域矩形が重なり合う場合がある。これらは本来１つであるべきであるので、矩形同士が重なり合っている場合、その重なり合う割合に応じて矩形同士を統合し、その結果、１６に示す検出結果を得るようにする。重なり合う割合は、例えば矩形１、矩形２が重なり合っている場合、次式で求めることができる。
ｉｆ 矩形１の面積 ＞ 矩形２の面積
重なり率 ＝ 重なり部分の面積／矩形１の面積
ｅｌｓｅ
重なり率 ＝ 重なり部分の面積／矩形２ の面積
そして、重なり率がある閾値よりも大きい場合に２つの矩形を統合し、１つの矩形にする。２つの矩形を統合する場合、各４点の座標の平均値を求める、あるいは座標値の大小関係から求めることができる。この統合は、画像の人数分それぞれ行われる。 There are a plurality of rectangle features in which the size, position, and rotation angle of the basic pattern shown in FIG. 20 are changed. The feature of the face is expressed using a face area discriminant function called Classifier combining this RectangleFeature. Specifically, a weight (pv or nv) is given according to whether or not each RectangleFeature value (Rf (i)) exceeds a threshold (th), and then the sum of all RectangleFeature weights (Cls) When a certain threshold value (th2) is exceeded, it is determined whether or not the face is used.

pv and nv are weights

When the matching process is completed, the face area detecting unit 2 the search block 14 for example, by moving One not a single pixel (Step S2-8), operates the entire scale image, after the number of people the process in the image, the following Are processed (step S2-9). After the above processing is performed on all scale images (step S2-10), the overlapping area integration device 9 integrates the overlapping areas of the detection areas (step S2-11). Here, the overlapping area integrating device 9 will be described in more detail. Since the face area detection apparatus 2 performs the face area determination while moving the search block 14 pixel by pixel as described above, for example, as shown in FIG. By determining, the face area rectangles stored in the face area determining device 8 may overlap. Since these should originally be one, when the rectangles overlap, the rectangles are integrated according to the overlapping ratio, and as a result, the detection result shown in 16 is obtained. For example, when the rectangle 1 and the rectangle 2 are overlapped, the overlapping ratio can be obtained by the following equation.
if Area of rectangle 1> Area of rectangle 2 Overlap ratio = Area of overlap / Rectangle 1 area
else
Overlap ratio = area of overlapped portion / area of rectangle 2 Then, when the overlap ratio is greater than a certain threshold, two rectangles are integrated into one rectangle. When two rectangles are integrated, an average value of coordinates of four points can be obtained or can be obtained from a magnitude relationship of coordinate values. This integration is performed for each person in the image.

Next, the operation of the face area extraction device 3 will be described with reference to FIGS. 4 and 9. In many cases, the first face area detected by the face area detection device 2 does not include the entire face including the hair as indicated by 17 shown in FIG.

Therefore, the appropriate field angle calculation device 10 of the face region extraction device 3 expands the length and breadth of the 17 regions in consideration of the face characteristics, and obtains a second face region 18 (step S3-1). Here, the appropriate angle of view is a range suitable for registration in an electronic file, and refers to a range in which the entire face image can be appropriately viewed when the user later opens the electronic file and views the image. Also, the method that considers the characteristics of the face is, for example, the method of enlarging as follows when the height of the detected rectangle is h, the width is w, and the center coordinates of the rectangle are (cx, cy) is there.

Upper left coordinates: (left, top) = (cx−w / 2 * p1, cy + h / 2 * p2), upper right coordinates: (right, top) = (cx + w / 2 *, cy + h / 2 * p2) Lower left coordinates: (left, bottom) = (cx−w / 2 * p1, cy − h / 2 * p3), lower right coordinates: ( right , bottom,) = (cx + w / 2 * p1, cy − h / 2 * p 3 ), p 1, p 2, and p 3 are extension rates of the respective sides, for example, p 1 = 1.5, p 2 = 2.0, and p 3 = 1.5. As a result, as shown in FIG. 9, the detection rectangle does not include even the hair and chin, but is included in the face image based on the appropriate angle of view.

The area extraction device 11 extracts a face image relating to the obtained rectangular face area from the input image (step S 3-2), and enlarges / reduces the extracted face image in accordance with the size registered in the electronic file (step S 3-2). S3-3). Examples of the method for enlarging and reducing the face image include bilinear interpolation and thinning. As described above, when there are a plurality of images from a photographic image, face images for the number of people can be extracted.

Next, the operation of the electronic file registration apparatus 4 will be described with reference to FIGS. As shown in FIG. 10, among the face images for a plurality of persons extracted by the face area extraction device 3 by the list display device 40, a list for several persons or all of the persons is displayed (step S4-1), and face image selection is performed. The device 41 selects a face to be registered in the electronic file from the face images displayed as a list by the user. Next, one of the menus 20 for new registration or registration editing for adding a face image to an already created electronic file is selected from the menu (step S4-3). When the user selects the new registration 21, the new registration screen 22 is displayed, and necessary information is input there and registered in the electronic file together with the face image (steps S4-7, S4-8). When the user selects the registration edit 23, the existing electronic file list 24 is displayed (step S4-4), and the user selects an item for which a facial image is to be registered from the displayed electronic file list (step S4-5). ). Then, a face image is registered in the data of the selected item, and a screen for confirming the registered content is displayed in the same manner as 22 (step S4-8).

As described above, by extracting a plurality of face images from an input image and registering the images in an electronic file, it is not necessary to take or obtain a single photo for electronic file registration, and the face image is converted into an electronic file. It becomes possible to simplify the registration.

In this embodiment, the medium for registering the face image has been described as an electronic file. An electronic file is a concept that includes a telephone directory that can store electronic information. In this embodiment, the image to be registered has been described as a face image. However, the present invention is not limited to a face image as a registration target, and widely includes object images. In this case, the image database corresponds to the object database, the face area detection device corresponds to the object area detection device, and the face area extraction device corresponds to the object area extraction device. For example, the case where a flower is a target registered in an electronic file is also included. In this embodiment, it has been described that the image value of the image is normalized after the multiscale image is generated. However, the present invention is not limited to the case where the image value of the image is normalized after the multiscale image is generated. For example, a case where a multiscale image is generated after the image value of the image is normalized is also included.

Embodiment 2
The configuration of electronic file generation apparatus 200 in the present embodiment of the present invention will be described using FIG. Note that description of portions common to the first embodiment is omitted. Reference numeral 201 denotes a face area extraction apparatus that extracts a face image based on the first face area detected by the face area detection apparatus 2, and reference numeral 25 denotes a part of the face area extraction apparatus 201 that is both eyes and mouth. The part detection device 26 is a two-eye detection device that detects both eyes of the part detection device 25, 27 is a mouth detection device that detects the mouth of the part detection device 25, and 28 is matching information for detecting both eyes and mouth. Is an appropriate angle-of-view calculation device that performs an appropriate angle-of-view adjustment in consideration of the positions of both eyes and mouth.

Next, the operation of each apparatus according to the present invention will be described with reference to FIGS. FIG. 12 is a flowchart for explaining the operation of the electronic file generation apparatus 200 of the present invention. 13 and 14 are flowcharts for explaining the operation of the part detection device 25. FIG. 15 is a flowchart for explaining the operation of the appropriate angle-of-view calculation device 30. FIG. 16 is a diagram for explaining the operation of the appropriate field angle calculation device 30.

First, the operation of the electronic file generation apparatus 200 will be described with reference to FIG. An image is input by the image input device 1 (step S5-1), and the first face region is detected by the face region detection device 2 (step S5-2). Next, both eyes and mouth are detected from the face image relating to the first face region by referring to the part image DB 28 (step S5-3). Then, the angle of view is adjusted according to the position of the part detected by the appropriate angle-of-view calculating device 30 (step S5-4), and the face image relating to the second face area is extracted with the adjusted angle of view and registered in the electronic file. The face image is enlarged / reduced so as to match the size to be used (step S5-5). Finally, the face area image is registered in the electronic file by the electronic file registration device 4 (step S5-6).

Next, the operation of the site detection device 25 will be described with reference to FIGS. 13 and 14. First, the part detection device 25 generates a multi-scale image of the face image relating to the first face region detected by the face region detection device 2 (step S6-1). Next, a search block that has a certain size (for example, 24 (pixels) × 24 rectangle), moves on the face image covering the first face area, and searches for the left eye, right eye, and mouth is generated as described above. All the multi-scale images that have been scanned are scanned and a matching process is performed to detect each part.

Specifically, the following processing is repeated for the multiscale image obtained in step S6-1 (step S6-2). First, it is checked whether or not the left eye has already been detected (step S6-3). If it has not been detected, matching processing is performed on the data of the part image DB 28 in which the left eye search block and the left eye feature amount are registered. (Step S6-4) When it is determined that the search block resulting from the matching process is the left eye (Step S6-5), the position of the left eye is stored (Step S6-6). If the search block is not the left eye, the left eye search block is moved by one pixel (step S6-7).

pv,nvは重み

Here, the matching process is assumed to be quantified based on the Rectangle Feature, for example, the part data registered in the part image DB 28 and learned the feature of the facial part. The feature of the face part is expressed using a face area discriminant function called Classifier that combines this Rectangle Feature. Specifically, a weight (pv or nv) is given according to whether or not each Rectangle Feature value (Rf (i)) exceeds a threshold (th), and then the sum of the weights of all Rectangle Features ( When Cls) exceeds a certain threshold value (th2), it is determined whether or not it is a part image.

pv and nv are weights

Next, the same processing is performed for the right eye and mouth (steps S6-8 to S6-17), and the detection processing is repeated until all the parts are detected or the entire image is scanned.

The angle of view is adjusted in the appropriate image calculation device 30 based on the position of the part obtained as described above. For example, there are the following methods. First, the distance between both eyes is obtained (step S7-1). Then, a vector formed at the detection points of both eyes is obtained (step S7-2), and the both eye vectors are expanded from the center point of both eyes to the left and right according to the distance between both eyes along the obtained eye vector (step S7). -3). Assume that the detected eyes and mouth are 31 in FIG. The eye vector at this time is 32, and 32 is expanded according to the distance between the eyes. Next, a vector in which the mouth detection point and the both eyes vector intersect perpendicularly is obtained (step S7-4). In FIG. 16, 33 is the obtained vector. This vector is first expanded downward (step S7-5) and further expanded upward (step S7-6) according to the distance between the eye vector and the mouth detection point. Reference numeral 33 denotes a vector obtained as described above.

The face area rectangle 34 is obtained from the binocular vector and the mouth-binocular vector obtained as described above, and the area is extracted by the area extracting device 11 in accordance with the rectangle. As described above, it is possible to further accurately determine the appropriate angle of view by further detecting the face part, and it is possible to appropriately cut out the face region.

In the present embodiment, the face image is described as both eyes and mouth. However, the present invention is not limited to both eyes and mouth as the part of the face image. That is, there are a nose, an ear, and the like as a part of the face image, and the face image concerning the second face region may be extracted based on such a part.
Embodiment 3

The configuration of the electronic file generation device 300 according to this embodiment of the present invention will be described with reference to FIG. Note that description of portions common to the first or second embodiment is omitted. In FIG. 17, an electronic file generation apparatus 300 includes an image input apparatus 1 that inputs an image, a face area detection apparatus 2 that detects a first face area that is a face area from the input image, and a first face. A face area extracting device 29 for extracting a face image relating to the second face area from the area based on the features of the face image; an electronic file registering apparatus 4 for registering the face image extracted by the face area extracting device 29 in an electronic file; , Is composed of. The electronic file registration device 4 includes an image composition device 35 that edits a face image by superimposing a decoration image such as an icon on the extracted face image.

Next, the operation of the electronic file registration apparatus 4 including the image composition apparatus 35 will be described with reference to FIGS. FIG. 18 is a flowchart for explaining the operation. FIG. 19 is a diagram illustrating screen transition during operation.

The electronic file registration apparatus 4 displays a list of face images for several or all of the face images for a plurality of people extracted from the image as shown in 36 of FIG. 19, and the user selects one of them. Is selected and the face image editing menu is selected (step S8-1), the image composition device 35 displays a list of icons that can be composed as in 37 (step S8-1). When the user selects an icon from the icon images displayed in a list (step S8-3), the image composition apparatus enlarges or reduces the icon image according to the size of the face, and composes the icon image according to the position of the part. (Step S8-4). Here, scaling the icon image in accordance with the size of the face and synthesizing the icon image in accordance with the position of the part may be based on part detection, for example. Note that the icon image holds transparency information for each pixel, and is an image having an alpha channel, for example. The image composition device superimposes the face image and the icon based on the transmission information. The combined result is displayed as 38 (step S8-4). If the user presses the OK button (step S8-5), for example, the screen changes to the normal electronic file registration screen 22 or 24, and the electronic file is displayed. Registration is possible (step S8-6).

In this way, by making it possible to decorate the cut face photograph, the characteristics of the person can be exaggerated. For example, when a call is received, the person can be easily identified.

BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 1 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a conceptual diagram of the electronic file production | generation apparatus which shows Embodiment 2 of this invention. It is a figure which shows the basic form of RectangleFeature shown in Embodiment 1 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image input device 2 Face area detection apparatus 3 Face area extraction apparatus 4 Electronic file registration apparatus 5 Multiscale image generation apparatus 6 Image value normalization apparatus 7 Face image database (face image DB)
DESCRIPTION OF SYMBOLS 8 Face area determination apparatus 9 Overlapping area integration apparatus 10 Appropriate view angle calculation apparatus 11 Area extraction apparatus 40 List display apparatus 41 Face image selection apparatus 12 Input image 13 Multiscale image 14 Search block 25 Region detection apparatus 26 Both-eye area detection apparatus 27 Mouth Region detection device 28 Part image database (part DB)
DESCRIPTION OF SYMBOLS 30 Appropriate angle of view calculation apparatus 200 Electronic file generation apparatus 201 Face area extraction apparatus 29 Face area extraction apparatus 35 Image composition apparatus 300 Electronic file generation apparatus

Claims (8)

In an electronic file generation device that generates an electronic file in which facial images are registered,
An image input device for inputting an image;
A face image database in which face image data is registered;
The feature amount of the image inside the search block moving on the image input by the image input device is compared with the feature amount of the face image registered in the face image database, and the image inside the search block is a face region. And a plurality of images determined to be a part or all of the face area by the face area determination device are determined as the face area of the same person. A plurality of images that are determined to be the same person's face area in the event that the face image of the person copied in the input image exists from the input image, the first face area is generated A face area detection apparatus that includes an area integration device and detects the first face area for a number of persons by comparing the feature quantity of the input image with the feature quantity of the face image registered in the face image database. When,
Based on the characteristics of the face image of the first face area detected by the face area detection device, an appropriate angle of view when registering in the electronic file is obtained, and the first face area is considered in terms of the face characteristics. Then, a face area extraction device that extracts a face image related to the second face area expanded vertically and horizontally at a predetermined extension rate so that the entire face is included,
An electronic file registration device for registering the face image extracted by the face region extraction device in an electronic file;
An electronic file generation device comprising: The face area detecting device is
The electronic file generation device according to claim 1 , further comprising a pixel value normalization device that normalizes a pixel value inside the search block before performing the face area determination. The face area extracting apparatus is
A part image database in which the part data of the face image is registered;
A part region for detecting a part region of a face image from the first face region by comparing a face image relating to the first face region detected by the face region detection device and data registered in the part image database A detection device;
An appropriate angle-of-view calculating device that adjusts the angle of view suitable for registration in the electronic file based on the region of the region detected by the region of the region detection device;
Electronic file generating device according to any one of claims 1 or claim 2, further comprising a. The face area extracting apparatus is
A part image database in which data of both eyes and mouth are registered;
By comparing the face image of the first face area detected by the face area detection device with the data registered in the part image database, the region of both eyes and mouth is determined from the first face region. A region detection device for detecting;
An appropriate angle-of-view calculating device that adjusts the angle of view suitable for registration in the electronic file based on the region of the region detected by the region of the region detection device;
The electronic file generation device according to any one of claims 1 to 3, further comprising: The electronic file registration device
When the face image has been extracted several servings double by the face region extraction unit, and the list display unit for displaying a list of several persons or all portion of the face image among the plurality of persons,
A face image selection device that allows a user to select a face image that the user wants to register in an electronic file from a list of face images displayed by the list display device;
The electronic file generation device according to any one of claims 1 to 4, further comprising: The electronic file registration device
When selecting a face image displayed by the list display device, an icon list display device that displays a list of icons that can be combined with the selected face image;
When the user selects from the icon images displayed by the icon list display device, the icon image is enlarged or reduced according to the size of the face, and the icon image is combined with the selected face image according to the position of the part of the face image. An icon image composition device to
The electronic file generation device according to any one of claims 1 to 5, further comprising: 2. The phone book generation device according to claim 1, wherein the electronic file registration device is a phone book generation device. In an electronic file generation device that generates an electronic file in which an object is registered,
An image input device for inputting an image;
An object database in which object data is registered;
The feature amount of the image inside the search block moving on the image input by the image input device is compared with the feature amount of the object image registered in the object database, and the image inside the search block is the object region. An object area determination device that determines whether or not the object area is a part or the whole, and a plurality of images that are determined to be a part or all of the object area by the object area determination apparatus are determined as object areas of the same object In this case, a plurality of images determined as object regions of the same object are integrated, and an overlapping region integration device that generates a first object region that is a region where an object image copied to the input image exists from the input image In the feature quantity of the input image and the object database By comparing the feature amount of the object image being recorded, the object area detection device for detecting the first object region by the number of objects,
Based on the characteristics of the object image relating to the first object area detected by the object area detection device, an appropriate angle of view when registering in the electronic file is obtained, and the characteristics of the object are taken into consideration for the first object area. An object area extracting device for extracting an object image relating to the second object area expanded vertically and horizontally at a predetermined extension rate so that the entire object is included;
An electronic file registration device for registering an object image extracted by the object region extraction device in an electronic file;
An electronic file generation device comprising:

Images