WO2005064539A1

WO2005064539A1 - Image area extraction and image processing method, device, and program

Info

Publication number: WO2005064539A1
Application number: PCT/JP2004/018816
Authority: WO
Inventors: Takeshi Nakajima; Hiroaki Takano; Tsukasa Ito; Daisuke Sato
Original assignee: Konica Minolta Photo Imaging, Inc.
Priority date: 2003-12-26
Filing date: 2004-12-16
Publication date: 2005-07-14

Abstract

There are provided an image area extraction method, an image area extraction device, an image area extraction program, an image processing method, an image processing device, and an image processing program for extracting an image area from an image. When extracting a desired image area from a source image, an image adjustment/processing unit firstly extracts an edge and creates a low-frequency image corresponding to the source image. For this low-frequency image, pixels satisfying a predetermined extraction condition are extracted as initial pixels and by using the low-frequency image signal, an image area is extended from the initial pixels. After this, when the image area has reached the image edge pixels, the extension is terminated and pixels of the image area after the extension are extracted.

Description

Specification

Method, apparatus, and program for image region extraction and image processing

Technical field

The present invention relates to an image region extraction method, an image region extraction device, an image region extraction program, an image processing method, an image processing device, and an image processing program for extracting an image region from an image.

Background art

Conventionally, a technique of photoelectrically reading an image formed on a color photographic film with a CCD (Charge Coupled Device) sensor or the like and converting it into an image signal has been widely used.

[0003] Such image signals are subjected to various image processing such as negative / positive inversion, luminance adjustment, color balance adjustment, grain removal, sharpness enhancement, and the like, and then are subjected to CD-R (CD-Recordable), CD-RW. (CD—Rewritable), FD (Floppy (registered trademark) disk), distributed via recording media such as memory cards and the Internet. The distributed image signal is output as a hard copy image using silver halide photographic paper, an ink jet printer, a thermal printer, etc., or displayed on a CRT (Cathode Ray Tube), liquid crystal display, plasma display, etc. for viewing. You.

[0004] In recent years, digital still cameras (including those incorporated in devices such as mobile phones and personal computers; hereinafter, abbreviated as DSCs) have become widespread, and, like color photographic films, have been used as hard copy images. It is output and displayed on a CRT or the like for viewing.

[0005] Here, when an image to be viewed includes a person's face, the person's face is most watched at the time of viewing. For this reason, in order to output a high-quality image, it is necessary to impart appropriate color, brightness, sharpness, noise, three-dimensionality, and the like to a person's face.

[0006] However, when photographing an image, it is generally not an environment in which a person's face is photographed as an image having an appropriate color, brightness, sharpness, noise, stereoscopic effect, and the like. is there.

[0007] For example, in shooting against backlight, the face is shot dark, and it is difficult to output a high-quality image. In addition, shooting with an inexpensive camera 'DSC' is a fixed focus shooting, Since the image is shot with the whole object in focus, an image in which the face of the person does not have a three-dimensional effect will be output.

[0008] In order to solve such a problem, for example, a specific image area is extracted from an image using a simple area expansion method, and it is determined whether the extracted image area is a main part (face) or not. In addition, a technique for performing a blurring process on an area other than the main part has been proposed (see Patent Document 1).

[0009] Here, the simple area extension method is to extract an image area by expanding adjacent pixels whose data difference between pixels is equal to or less than a threshold value as belonging to the same image area, and extracting the image area. (For example, see Non-Patent Document 1). In other words, starting from an initial pixel that satisfies a specified specific condition, if the data difference between pixels adjacent to the initial pixel (either 8-connected or 4-connected) is less than or equal to the threshold, It is assumed that the adjacent pixel and the initial pixel belong to the same image area, and the same judgment is performed for the pixel adjacent to the pixel belonging to the same image area, and the same area is determined starting from the initial pixel. This is an image processing method that extracts an image area by gradually expanding it.

Patent Document 1: Japanese Patent Application Laid-Open No. 2001-57630

Non-Patent Document 1: Mikio Takagi and Director of Director Hisahisa Shimoda, "Image Analysis Handbook", First Edition, The University of Tokyo Press, January 17, 1991 However, as a result of detailed examination, It has been concluded that it is difficult to accurately extract a desired image region from an image and perform appropriate image processing on the extracted image region by the method for extracting an image region as disclosed in Japanese Patent Application Laid-Open No. H11-157163. Was.

[0010] For example, when an image signal is generated by reading a color photographic film by a CCD or the like, a data difference between adjacent pixels is larger than a threshold value due to the presence of granular noise in the photographic film. And it becomes difficult to accurately extract a desired image region.

[0011] Even if the image signal is generated by a relatively inexpensive DSC or the like, it becomes difficult to accurately extract a desired image area. In other words, in such a DSC, the pixel pitch of the image sensor is short, so that shot noise is easily generated at low sensitivity, and the image The current noise generated by the heating of the sensor is likely to occur. Further, in the case of a DSC equipped with a CMOS image sensor, noise due to the leak current generated from the image sensor is also likely to occur. Through such noise processing S, color filter array interpolation and edge enhancement image processing, mottle-like granular unevenness is likely to be formed. For this reason, due to the presence of such granular unevenness, there is a possibility that the data difference between adjacent pixels becomes larger than a threshold value, and it becomes difficult to accurately extract an image region.

[0012] Furthermore, even if the threshold is set to a larger value in order to solve the above-described problem, an inconvenience such that an image area extracted based on the threshold becomes an area larger than a desired image area. , It is still difficult to accurately extract a desired image area.

[0013] Further, if the extraction of the image area is not accurate, the image processing may not be properly performed. That is, if image processing is performed on a specific image area, accurate extraction of the image area is premised. Even if applied, the desired effect cannot be obtained, and the possibility is high.

Disclosure of the invention

An object of the present invention is to appropriately extract a desired image area from an image and perform appropriate image processing on the extracted image area.

[0015] In order to solve the above-mentioned problems, the mode described in Item 1

Obtaining an image signal composed of signals of a plurality of pixels;

Extracting a pixel corresponding to an image edge from the plurality of pixels;

Creating a low frequency image signal from the image signal;

Extracting a pixel satisfying a predetermined extraction condition from the plurality of pixels as an initial pixel;

The image area is extended from the initial pixels using the low-frequency image signal. Thereafter, when the image area reaches the pixel at the image edge in the extension process, the extension is stopped and the image area after the extension is extended. Extracting pixels of

It is characterized by including.

[0016] Further, as in the form described in Item 2, in the form described in Item 1,

In the step of extracting pixels corresponding to the image edge, the luminance of the image signal between the pixels is determined. The rate of change is normalized by either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated, and when the normalized value is equal to or greater than a predetermined threshold, the pixel to be expanded is Is preferably extracted as a pixel corresponding to the image edge.

[0017] Further, as in the embodiment described in Item 3, in the embodiment described in Item 1 or 2,

The extraction condition is preferably a condition for extracting a pixel representing the skin of a person.

[0018] In order to solve the above-mentioned problems, the mode described in Item 10 is

Obtaining an image signal composed of signals of a plurality of pixels;

Extracting a pixel corresponding to an image edge from the plurality of pixels;

Creating a low frequency image signal from the image signal;

Extracting a pixel that satisfies an extraction condition for extracting a pixel representing a human skin from the plurality of pixels as an initial pixel;

A step of determining whether or not the image area after the expansion represents a person's face; a step of performing predetermined image processing on the image area determined to represent a person's face;

It is characterized by including.

[0019] Further, as in the mode described in Item 11, in the mode described in Item 10,

Extracting the initial image,

It is a step of extracting, as initial pixels, a plurality of pixels each satisfying a plurality of types of extraction conditions for extracting a pixel representing a person's skin from the plurality of pixels, and extracting pixels of the expanded image area. The process of extracting

The image area is expanded from the initial pixels using the low-frequency image signal, and thereafter, when the image area reaches the pixel at the image edge in the expansion process, the expansion is stopped and expanded. It is a step of extracting pixels of each subsequent image area,

The step of determining whether or not the face of the person is represented, A step of determining whether or not each of the image areas after the expansion represents a person's face.

[0020] Further, as in the mode described in [12], in the aspect described in [10] or [11], in the step of extracting a pixel corresponding to the image edge, a luminance change rate of an image signal between pixels is determined by: Normalization is performed using either the luminance value of the target pixel or the average value of the luminance of the pixel whose luminance change rate is to be calculated.If the standardized value is equal to or greater than a predetermined threshold, the pixel to be expanded is set as the image edge. It is preferable to extract as corresponding pixels.

Brief Description of Drawings

FIG. 1 is a perspective view showing an external configuration of an image processing apparatus to which the present invention is applied.

FIG. 2 is a block diagram showing an internal structure of an image processing device to which the present invention is applied.

FIG. 3 is a block diagram mainly showing an internal configuration of an image processing unit shown in FIG. 2.

FIG. 4 is a flowchart illustrating image processing contents to which the present invention is applied. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, another preferred embodiment of the present invention will be described.

[0023] The form described in Item 4 is

Acquiring means for acquiring an image signal composed of signals of a plurality of pixels;

Edge extracting means for extracting a pixel corresponding to an image edge from the plurality of pixels, creating means for creating a low-frequency image signal from the image signal,

Initial pixel extracting means for extracting a pixel satisfying a predetermined extraction condition from the plurality of pixels as an initial pixel,

The image area is extended from the initial pixels using the low-frequency image signal. Thereafter, when the image area reaches the pixel at the image edge in the extension process, the extension is stopped and the image area after the extension is extended. Expanded region extracting means for extracting pixels of

It is characterized by having.

[0024] Further, the form described in Item 5 is similar to the form described in Item 4,

The edge extraction means normalizes the luminance change rate of the image signal between pixels by using either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated. Is greater than or equal to a predetermined threshold, the pixel to be expanded is paired with the image edge. Preferably, it is extracted as the corresponding pixel.

[0025] Further, the form described in Item 6 is similar to the form described in Item 4 or 5,

[0026] The form described in Item 7 is

In the computer that performs image processing,

A function of acquiring an image signal composed of signals of a plurality of pixels,

A function of extracting a pixel corresponding to an image edge from the plurality of pixels,

A function of creating a low-frequency image signal from the image signal;

A function of extracting a pixel satisfying a predetermined extraction condition from the plurality of pixels as an initial pixel;

The image area is extended from the initial pixels using the low-frequency image signal. Thereafter, when the image area reaches the pixel at the image edge in the extension process, the extension is stopped and the image area after the extension is extended. Function to extract the pixels of

To realize.

[0027] Further, as in the mode described in Item 8, as in the mode described in Item 7,

To the computer,

The luminance change rate of the image signal between pixels is normalized by either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated, and the normalized value is equal to or greater than a predetermined threshold value. In this case, it is preferable to further realize a function of extracting the pixel to be extended as a pixel corresponding to an image edge.

[0028] Further, as in the form described in Item 9, as in the form described in Item 7 or 8,

[0029] The form described in Item 13 is:

Edge extracting means for extracting a pixel corresponding to an image edge from the plurality of pixels, creating means for creating a low-frequency image signal from the image signal, Initial pixel extracting means for extracting a pixel satisfying an extraction condition for extracting a pixel representing a human skin from the plurality of pixels as an initial pixel;

Determining means for determining whether or not the image area after the expansion represents a person's face; processing means for performing predetermined image processing on the image area determined to represent a person's face;

It is characterized by having.

[0030] Further, as in the form of Item 14, in the form of Item 13,

The means for extracting the initial image,

Initial pixel extraction means for extracting, as initial pixels, a plurality of pixels each satisfying a plurality of types of extraction conditions for extracting a pixel representing a human skin from the plurality of pixels,

Means for extracting pixels of the image area after the expansion,

The image area is expanded from the initial pixels using the low-frequency image signal, and thereafter, when the image area reaches the pixel at the image edge in the expansion process, the expansion is stopped and expanded. A post-expansion area extracting means for extracting pixels of each subsequent image area,

The means for determining whether or not the face of the person is represented,

A determination unit configured to determine whether each of the extended image areas represents a person's face.

[0031] Further, as in the embodiment described in Item 15, in the embodiment described in Item 13 or 14,

The edge extraction means normalizes the luminance change rate of the image signal between pixels by using either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated. When is greater than or equal to a predetermined threshold, it is preferable to extract the pixel to be expanded as a pixel corresponding to an image edge.

[0032] As in the form described in Item 16, In the computer that performs image processing,

A function of creating a low-frequency image signal from the image signal;

A function of extracting a pixel that satisfies an extraction condition for extracting a pixel representing a human skin from the plurality of pixels as an initial pixel;

A function of determining whether or not the image area after the expansion represents a human face; a function of performing predetermined image processing on the image area determined to represent a human face;

To realize.

[0033] Further, as in the form described in Item 17, in the form described in Item 16,

The function of extracting the initial image,

A function of extracting, as initial pixels, a plurality of pixels each satisfying a plurality of types of extraction conditions for extracting a pixel representing a person's skin from the plurality of pixels, and extracting pixels of the image area after the expansion. The function to extract is

The image area is expanded from the initial pixels using the low-frequency image signal, and thereafter, when the image area reaches the pixel at the image edge in the expansion process, the expansion is stopped and expanded. It is a function to extract the pixels of each subsequent image area,

The function of determining whether or not the face of the person is represented,

It is preferable to realize a function of determining whether or not each of the extended image areas represents a human face.

[0034] Further, as in the embodiment described in Item 18, as in the embodiment described in Item 16 or 17,

To the computer,

The luminance change rate of the image signal between pixels is normalized by either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated, and the normalized value is determined by a predetermined threshold. When the value is equal to or larger than the value, it is preferable to further realize a function of extracting the pixel to be extended as a pixel corresponding to an image edge.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

First, the external configuration of the image processing apparatus 1 will be described with reference to FIG.

As shown in FIG. 1, the image processing apparatus 1 is provided with a magazine loading section 3 for loading a photosensitive material on one side surface of a housing 2. Inside the housing 2, there are provided an exposure processing section 4 for exposing the photosensitive material, and a print creating section 5 for developing and drying the exposed photosensitive material to create a print. On the other side of the housing 2, a tray 6 for discharging the print created by the print creating section 5 is provided.

In addition, a CRT 8 as a display device, a film scanner unit 9 that is a device for reading a transparent original, a reflective original input device 10, and an operation unit 11 are provided at an upper portion of the housing 2. Further, the housing 2 is provided with an image reading section 14 for reading images recorded on various digital recording media, and an image writing section 15 for writing (outputting) image signals on various digital recording media. Further, inside the housing 2, a control unit 7 for integrally controlling each unit constituting the image processing apparatus 1 is provided.

[0038] The image reading unit 14 is provided with a PC card adapter 14a and an FD adapter 14b, so that the PC card 13a and the FD 13b can be inserted.

[0039] The image writing unit 15 is provided with an FD adapter 15a, a MO (Magneto-Optical) adapter 15b, and an optical disk adapter 15c, and the FD 16a, M016b, and the optical disk 16c can be inserted thereinto, respectively. I have. The optical disk 16c includes CD_R, DVD-R (Digita 1 Versatile Disk—Recordable), DVD—RW (DVD—Rewritable), and the like.

In FIG. 1, the operation unit 11, the CRT 8, the film scanner unit 9, the reflection document input device 10, and the image reading unit 14 are configured integrally with the housing 2. May be provided separately.

In the image processing apparatus 1 shown in FIG. 1, an example in which a photosensitive material is exposed to light and developed to create a print is exemplified. However, the print creating method is not limited to this. A system such as an ink jet system, an electrophotographic system, a heat-sensitive system, and a sublimation system may be used. <Internal configuration of image processing device 1>

Next, an internal structure of the image processing apparatus 1 will be described with reference to FIG. As shown in FIG. 2, the image processing apparatus 1 includes a control unit 7, an exposure processing unit 4, a print generation unit 5, a film scanner unit 9, a reflection document input device 10, an image reading unit 14, a communication unit (input) 32, It has an image writing unit 15, a data storage unit 71, an operation unit 11, a CRT 8, and a communication unit (output) 33.

The exposure processing section 4 exposes the photosensitive material to an image, and outputs the photosensitive material to the print creating section 5. The print creating section 5 develops the exposed photosensitive material and dries it to create prints Pl, P2 and P3.

The control unit 7 includes a microcomputer, and various control programs such as an image processing program stored in a ROM (Read Only Memory) or the like (not shown) and a CPU (Central Processing Unit) (not shown). In cooperation with (abbreviation), the operation of each unit constituting the image processing apparatus 1 is controlled in a comprehensive manner.

The control unit 7 includes an image processing unit 70, and based on an input signal (command information) from the operation unit 11, a film scanner unit 9, a reflection document input device 10, an image reading unit 14, an external device The communication unit (input) forms an image for exposure for each image input from the input unit 32 and outputs the image to the exposure processing unit 4. The image processing unit 70 will be described later in detail.

The film scanner unit 9 reads an image recorded on a transparent original such as a developed negative film N or a reversal film captured by an analog camera.

The reflection document input device 10 reads an image formed on a print P (photo print, document, various prints) by a flatbed scanner (not shown).

The operation unit 11 has information input means 12. The information input means 12 is composed of, for example, a touch panel or the like, and outputs a press signal of the information input means 12 to the control section 7 as an input signal. The operation unit 11 may be configured to include a keyboard, a mouse, and the like. The CRT 8 displays an image or the like according to the display control signal input from the control unit 7.

[0048] The image reading unit 14 includes an image transfer unit 30, reads an image recorded on the PC card 13a or the FD 13b, and transfers the image to the control unit 7. The image transfer means 30 has a PC card adapter 14a, an FD adapter 14b, and the like. The image reading section 14 reads an image recorded on the PC card 13a inserted into the PC card adapter 14a or the FD 13b inserted into the FD adapter 14b, and transfers the read image to the image transfer means 30. Is transferred to the control unit 7 using.

[0050] The image writing unit 15 includes an image transport unit 31, and the image transport unit 31 includes an FD adapter 15a, an MO adapter 15b, an optical disk adapter 15c, and the like. The image writing unit 15 is configured such that the FD 16a inserted into the FD adapter 15a, the M〇 16b inserted into the MO adapter 15b, and the optical disk inserted into the optical disk adapter 15c, according to the write signal input from the control unit 7. Write various data to 16c.

[0051] The communication means (input) 32 receives images, various commands, and the like from another computer in the facility where the image processing apparatus 1 is installed or a distant computer via the Internet or the like.

[0052] The communication means (output) 33 transmits an image, order information, and the like to another computer in the facility where the image processing apparatus 1 is installed, or to a remote computer via the Internet or the like.

The data storage unit 71 stores data such as images and order information (information on how many prints are to be made from which frame images, print size information, etc.).

Next, the configuration of the image processing unit 70 will be described with reference to FIG. As shown in FIG. 3, the image processing unit 70 includes a film scan data processing unit 701, a reflection original scan data processing unit 702, an image data format decoding processing unit 703, and an image adjustment processing unit 704 (described in the claims). Acquisition unit, edge extraction unit, creation unit, initial pixel extraction unit, expanded area extraction unit, discrimination unit, and processing unit.), CRT specific processing unit 705, printer specific processing unit 706, printer specific processing And an image data format creation processing unit 708.

The film scan data processing unit 701 performs a calibration operation unique to the film scanner unit 9, a negative-positive inversion in the case of a negative original, a gray balance adjustment, a contrast adjustment, and the like for the image input from the film scanner unit 9. And outputs it to the image adjustment processing unit 704. Further, the film scan data processing unit 701 includes a film size, a negative / positive type, ISO (International Organization for Standardization) sensitivity optically or magnetically recorded on the film, a manufacturer name, information on a main subject, and shooting conditions. Information (example For example, APS (Advanced Photo System) information) is also output to the image adjustment processing unit 704.

The reflection document scan data processing unit 702 performs a calibration operation unique to the reflection document input device 10, negative-positive inversion for a negative document, gray balance adjustment, and contrast adjustment for an image input from the reflection document input device 10. And outputs the result to the image adjustment processing unit 704.

The image data format decryption processing unit 703 performs restoration of a compression code, conversion of a method of expressing color data, and the like in accordance with the data format of the image signal input from the image transfer unit 30 or the communication unit (input) 32. Are output to the image adjustment processing unit 704.

The image adjustment processing unit 704 performs various types of image processing on images input from the film scanner unit 9, the reflection original input device 10, the image transfer unit 30, and the communication unit (input) 32. In particular, the image adjustment processing unit 704 executes the image processing shown in the flowchart of FIG.

The image adjustment processing unit 704 sends the processed image to the CRT-specific processing unit 705, the printer-specific processing unit 706, the printer-specific processing unit 707, the image data format creation unit 708, and the data storage unit 71. Output.

[0059] The CRT-specific processing unit 705 performs processing such as changing the number of pixels and color matching on the image input from the image adjustment processing unit 704, and outputs the processed image to the CRT 8 together with various display information.

The printer-specific processing unit 706 performs printer-specific calibration processing, color matching, changing the number of pixels, and the like on the image-processed image signal input from the image adjustment processing unit 704, and an exposure processing unit. Output to 4.

[0061] The image processing apparatus 1 of the present embodiment is provided with a printer-specific processing unit 707 corresponding to the external printer 34 such as an inkjet printer. The printer-specific processing unit 707 performs an appropriate printer-specific calibration process, color matching, change of the number of pixels, and the like on the image input from the image adjustment processing unit 704.

[0062] The image data format creation processing unit 708 converts the image input from the image adjustment processing unit 704 into a JPEG (Joint Photographic Experts Group), a TIFF fagged Image File Format, an Exif (Exchangea Die image file format), or the like. (This is typical.) Conversion to various general-purpose image formats is performed, and the image is transferred to the image transport unit 31 and communication means (output) 33. Output.

Note that the above-described units of the control unit 7 (eg, a film scan data processing unit 701, a reflection original scan data processing unit 702, an image data format decoding processing unit 703, an image adjustment processing unit 704, a CRT specific processing unit 705, a printer The unique processing units 706 and 707, the image data format creation processing unit 708, etc.) do not necessarily have to be realized as physically independent devices. It may be something that is represented. Further, the image processing apparatus 1 can be applied to various modes, such as a digital photo printer, a printer dryino, and a plug-in of various image processing software, which are not limited to the above-described contents.

Next, image processing to which the present invention is applied will be described with reference to FIG. The image processing described below is executed by the image adjustment processing unit 704.

First, when an original image is acquired via the film scanner unit 9, the reflection original input device 10, the image transfer device 30, the communication means (input) 32 (Step S1), the edges of the original image are extracted and ( Step S2), a low-frequency image is created for the original image (step S3).

Here, when extracting the edge, if the luminance of the image signal is expressed as Y, the luminance change rate Δ Υ is represented by the luminance value of the target pixel (or the luminance average value of a plurality of pixels for which the change rate is calculated, etc.). ) The value Δ 値 / Δ normalized by Y is calculated, and if the calculated value ΔΥ / Υ is equal to or larger than a specific threshold, it is determined to be an edge, and information indicating the position of the edge pixel (for identifying the pixel) Is stored in the data storage means 71 or the internal memory of the control unit 7.

[0067] Although image edge extraction can be performed using a known edge extraction filter, in the present embodiment, it is appropriate to perform image edge extraction using high-frequency components obtained by binomial wavelet transform. Is preferred for extracting In addition, the generation of the low-frequency image can be performed using a known low-pass filter. In the present embodiment, it is preferable to use a low-frequency component obtained by the binomial-wavelet transform.

Next, a skin color region is extracted from the low-frequency image created in step S3 by using the simple region extension method.

[0069] That is, using one extraction condition for identifying and extracting an image signal representing a flesh color, an initial pixel (one or more pixels) is selected from the pixels of the image signal satisfying the extraction condition. It consists. ) Is specified, and simple area expansion is started from the initial pixel (step S4). In addition, not only the simple area expansion method but also a known expansion method can be used.

[0070] Here, it is assumed that the extraction conditions are set in advance and stored in the data storage means 71 or the internal memory of the control unit 7.

The extraction condition may be such that the user specifies a point (pixel) on the image with a mouse or the like and is set based on the image signal of the specified pixel, or may be predetermined. Was

Good, even if it is set based on RGB values. At this time, since the flesh color of the person has a hue (Hue) and a saturation (Satulation) within a certain range under substantially the same light source, the initial pixel is selected based on the conditions defining the hue and the saturation. Is preferred. It is preferable that the conditions defining the hue and saturation be changed depending on the type of light source at the time of photographing. Further, it is preferable that the type of light source at the time of photographing is automatically determined by a known method.

After step S4, simple area expansion for extracting a skin color area is performed while sequentially comparing the difference in image signal between adjacent pixels (step S5). That is, it is determined whether the image signal of the adjacent pixel satisfies the extraction condition (step S6). If the image signal of the adjacent pixel satisfies the extraction condition (step S6; Yes), It is further determined whether or not is an edge (step S7). If the adjacent pixel is not an edge (step S7; No), the adjacent pixel is included in the skin color area (step S8). Here, with respect to the pixels included in the flesh color region, information (information for identifying the pixels) representing each position is sequentially stored in the data storage unit 71, the built-in memory of the control unit 7, and the like.

[0073] After step S8, the process proceeds to step S5, and the above processing is repeated. In the process, if the image signal of the adjacent pixel does not satisfy the extraction condition in step S6 (step S6; No), or if the adjacent pixel is an edge in step S7 (step S7; Yes). , The extraction process of the skin color area using the simple area expansion is ended.

Here, the eyes, the mouth, and the like are not extracted as a part of the skin-color area that is simply expanded even though it is formed in the skin-color area. On the other hand, the eyes and mouth are included as closed areas in the skin color area. For this reason, when the closed region is included as a part of the skin color region that has been simply expanded, it is preferable to process the closed region so as to be included in the skin color region. Etc. can also be extracted by including it in the skin color area expanded by simple area Become.

As described above, when extracting a skin color region from an original image, the image adjustment processing unit 704 first extracts edges to create a low-frequency image for the original image. Then, the skin color region is expanded to the edge of the low-frequency image using the simple region expansion method, and the skin color region is extracted.

[0076] Therefore, by performing simple area expansion based on the low-frequency image, the influence of granular noise or the like can be excluded, so that a desired skin color area can be appropriately extracted.

[0077] Further, in the simple area expansion, when the pixel corresponding to the image edge extracted in advance is reached, the simple area expansion is forcibly terminated, so that the desired skin color area can be more appropriately extracted. It becomes.

Further, since a plurality of different extraction conditions are applied, and each face region is extracted using the simple region expansion method based on each of the extraction conditions, for example, an image in which the light source type at the time of shooting cannot be specified, Even if the image includes a plurality of faces, each of which is illuminated by a different light source, a face area can be properly extracted from the image.

The description in the present embodiment shows an example of the image area extracting method, the image area extracting apparatus, the image area extracting program, the image processing method, the image processing apparatus, and the image processing program according to the present invention. However, the present invention is not limited to this. The detailed configuration and detailed operation of the image processing device 1 according to the present embodiment can be appropriately changed without departing from the spirit of the present invention.

For example, in addition to the image processing shown in the flowchart of FIG. 4, a process of determining whether or not the skin color region extracted in the image processing can be specified as a face may be added. As a method for determining whether or not the extracted skin color area is a face, a known determination method such as a method using a neural network or pattern matching can be used. Further, image processing to be performed on the skin color area when it is determined to be a face includes processing such as color adjustment, grain removal, sharpness enhancement, and dynamic range adjustment. In particular, in order to shorten the extraction time of the flesh-color area, the flesh-color area is extracted from the reduced image obtained by reducing the image size of the original image, and the area of the original image corresponding to the flesh-color area of the extracted reduced image is extracted. It is good to perform image processing for Further, in the present embodiment, the case where simple area expansion is started from one initial pixel corresponding to one type of extraction condition has been described. However, the present invention is not limited to this. The simple area expansion may be performed independently from the initial pixels of the above. In this case, the processing power from step S4 to step S8 in FIG. 4 is performed according to the types of the initial pixels, and a plurality of skin color regions corresponding to the number of types of the initial pixels are extracted. Therefore, a plurality of different extraction conditions are applied, and each skin color region is extracted using the simple region expansion method for each extraction condition. , And even if the images are each illuminated by a different light source, the flesh color region can be accurately extracted from the images.

Further, in the present embodiment, the image area to be extracted using the simple area expansion method has been described as the skin color area. However, the present invention is not limited to this. For example, the image area to be extracted may be other areas such as the head hair. It is also applicable.

Industrial applicability

According to the present invention, by performing simple area expansion on a low-frequency image, the influence of granular noise or the like can be excluded, so that a desired image area can be properly extracted.

Further, in the case of region expansion, when the pixel corresponding to the image edge extracted in advance is reached, the region expansion is forcibly stopped, so that a desired image region can be more appropriately extracted.

[0085] Furthermore, since a plurality of different extraction conditions are applied and each face region is extracted based on the respective extraction conditions, for example, an image in which the light source type at the time of shooting cannot be specified, or a plurality of faces are included. Furthermore, even if the images are each illuminated by different light sources, the face region can be properly extracted from the images.

Claims

The scope of the claims

[1] a step of acquiring an image signal composed of signals of a plurality of pixels;

Extracting a pixel corresponding to an image edge from the plurality of pixels;

Creating a low frequency image signal from the image signal;

A method for extracting an image region, comprising:

[2] In the step of extracting a pixel corresponding to the image edge, the luminance change rate of the image signal between the pixels is determined by using either the luminance value of the target pixel or the average value of the luminance of the pixel whose luminance change rate is to be calculated. 2.The image according to claim 1, wherein the pixel to be expanded is extracted as a pixel corresponding to an image edge when the standardized value is equal to or larger than a predetermined threshold. Region extraction method.

3. The method according to claim 1, wherein the extraction condition is a condition for extracting a pixel representing a human skin.

[4] acquiring means for acquiring an image signal composed of signals of a plurality of pixels;

An image region extraction device comprising:

[5] The edge extraction means normalizes the luminance change rate of the image signal between the pixels by using either the luminance value of the target pixel or the average luminance value of the pixel for which the luminance change rate is to be calculated, 5. The image region extracting apparatus according to claim 4, wherein when the standardized value is equal to or larger than a predetermined threshold, the expansion target pixel is extracted as a pixel corresponding to an image edge.

6. The image region extraction device according to claim 4, wherein the extraction condition is a condition for extracting a pixel representing a human skin.

[7] Computers that perform image processing

A function of creating a low-frequency image signal from the image signal;

Image area extraction program for realizing.

[8] The computer

The luminance change rate of the image signal between pixels is normalized by either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated, and the normalized value is equal to or greater than a predetermined threshold value. 8. The image area extraction program according to claim 7, further comprising a function of extracting the pixel to be extended as a pixel corresponding to an image edge in the case of (1).

9. The image region extraction program according to claim 7, wherein the extraction condition is a condition for extracting a pixel representing a human skin.

[10] a step of obtaining an image signal composed of signals of a plurality of pixels;

Extracting a pixel corresponding to an image edge from the plurality of pixels;

Creating a low frequency image signal from the image signal;

Using the low-frequency image signal to extend the image area from the initial pixel, then Stopping the expansion when the image area reaches the pixel of the image edge in the expansion process, and extracting the pixels of the image area after expansion;

An image processing method comprising:

[11] The step of extracting the initial image,

The step of determining whether or not the face of the person is represented,

11. The image processing method according to claim 10, wherein the step of determining whether each of the image areas after the expansion represents a person's face is performed.

[12] In the step of extracting a pixel corresponding to the image edge, the luminance change rate of the image signal between the pixels is calculated by using the luminance value of the target pixel or the average value of the luminance of the pixel whose luminance change rate is to be calculated. 11.The image according to claim 10, wherein the pixel to be expanded is extracted as a pixel corresponding to an image edge when the standardized value is equal to or larger than a predetermined threshold. Processing method.

[13] acquiring means for acquiring an image signal composed of signals of a plurality of pixels;

Initial pixel extracting means for extracting a pixel satisfying an extraction condition for extracting a pixel representing a human skin from the plurality of pixels as an initial pixel;

The image area is extended from the initial pixel using the low-frequency image signal. Thereafter, when the image area reaches the pixel at the image edge in the extension process, the extension is performed. Stopping, extracting the pixels of the image area after expansion, an expanded area extraction means, and a determination means for determining whether the expanded image area represents a human face, Processing means for performing predetermined image processing on the image area determined to be present,

An image processing apparatus comprising:

[14] The means for extracting the initial image,

Means for extracting pixels of the image area after the expansion,

The means for determining whether or not the face of the person is represented,

14. The image processing device according to claim 13, further comprising: a determination unit configured to determine whether each of the image regions after the expansion represents a human face.

[15] The edge extracting means normalizes the luminance change rate of the image signal between pixels by using either the luminance value of the target pixel or the average luminance value of the pixel for which the luminance change rate is to be calculated, 14. The image processing apparatus according to claim 13, wherein when the standardized value is equal to or larger than a predetermined threshold, the pixel to be expanded is extracted as a pixel corresponding to an image edge.

[16] Computers that perform image processing

A function of creating a low-frequency image signal from the image signal;

A function of extracting a pixel that satisfies an extraction condition for extracting a pixel representing a human skin from the plurality of pixels as an initial pixel; The image area is extended from the initial pixels using the low-frequency image signal. Thereafter, when the image area reaches the pixel at the image edge in the extension process, the extension is stopped and the image area after the extension is extended. Function to extract the pixels of

Image processing program for realizing.

[17] The function of extracting the initial image,

17. The image processing program according to claim 16, wherein the function is to determine whether each of the extended image areas represents a human face.

[18] The computer

The luminance change rate of the image signal between pixels is normalized by either the luminance value of the target pixel or the average value of the luminance of the pixel for which the luminance change rate is to be calculated, and the normalized value is equal to or greater than a predetermined threshold value. 17. The computer-readable storage medium according to claim 16, further comprising a function of extracting the pixel to be extended as a pixel corresponding to an image edge.