JP5532004B2 - Image processing apparatus, image processing method, and program - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and a program which are processed by changing the degree of conversion of a person's face part when, for example, converting a photographic image including a person into a painting.

  In recent years, with the widespread use of digital cameras, the way to enjoy photos has become diversified. In particular, since it is digital image data, various image processing is possible.

  For example, a technique has been proposed that can generate and display an image with a different tone (such as a painting tone) based on an original photograph by adding image processing (for example, a patent) Reference 1).

  The technology of Patent Document 1 is developed, and features such as color information and brushing information are extracted from the image of the painting actually drawn by the painter. Considering this, a technique for converting to a painting-like image with high artisticness has also been proposed (see, for example, Patent Document 2).

  In addition, a technique for changing a picture quality such as a painting style has been proposed (see, for example, Patent Document 3).

JP-A-8-44867 JP 2004-213598 A JP-B-1-46905

  However, when converting a photo into a painting-like work, it is possible to obtain artistic effects even when bold, such as landscapes, flowers, and buildings, but in the case of a human face, There was a problem that it might become a work that looks strange when it is subjected to various transformations.

  The present invention naturally detects a human face by performing a different tone conversion between a face area and another area by detecting a face area when performing a tone conversion such as a painting style. It is an object to provide an image processing apparatus, an image processing method, and a program that can be expressed.

  In order to solve the above problem, an image processing apparatus according to claim 1 determines a face detection unit for detecting a face from image data and a region corresponding to the face detected by the face detection unit. A region determination unit performs a first tone conversion process on the image data of the region corresponding to the face detected by the region detection unit, and converts the image data to image data other than the region corresponding to the face. And a tone conversion processing means for performing the second tone conversion processing.

In order to solve the above problem, the invention according to claim 1 is an image processing apparatus capable of converting image data including a face into a plurality of types of image tone, and selects one type from the plurality of types of image tone. A selection unit, a face detection unit for detecting a face from image data, a region (sq) corresponding to the face detected by the face detection unit, and inner and outer boundary regions (el, el5) are determined. In converting the image data into the type of image tone selected by the selection unit and the region determination unit, the image data of the region (sq) corresponding to the face determined by the region determination unit is changed according to the parameter. The first image tone conversion processing is performed, the second image tone conversion processing is performed on the image data of the inner boundary region (el) determined by the region determination unit according to the parameter, and the region determination unit determines The third image tone conversion processing is performed on the image data of the outer boundary region (el5) that has been made according to the parameters, and image data other than the region (sq) corresponding to the face and the boundary region (el, el5) And a tone conversion processing means for performing a fourth tone conversion process according to the parameter, and the tone conversion processing means includes means for forming a brushstroke by a group of a plurality of pixels of the image data. The pixel of the image data is represented by A (RGB) (x, y), where (x, y) is a two-dimensional coordinate, (RGB) is the RGB value of the pixel A, and the parameter is represented by P When the influence degree of the parameter is represented by α, A (RGB) (x, y) = A (RGB) (x, y) * P * α1 is added to the image data of the region (sq) corresponding to the face. The process represented by (x, y) is executed, and the inside Image data in A (RGB) (x, y ) of the boundary region = A (RGB) (x, y) * P * α2 (x, y) and executes the processing represented by the boundary area of the outer A process represented by A (RGB) (x, y) = A (RGB) (x, y) * P * α3 (x, y) is executed on the image data, and α1>α2> α3. An image processing apparatus characterized by the above is provided.

According to a fourth aspect of the present invention, there is provided an image processing method capable of converting image data including a face into a plurality of types of image tone, a selection step of selecting one type from the plurality of types of image tone, A face detecting step for detecting a face from the inside; a region determining step for determining a region (sq) corresponding to the detected face and inner and outer boundary regions (el, el5); and the selection of the image data In converting to the type of image tone selected by the process, the image data of the area (sq) corresponding to the face determined by the area determining process is subjected to a first image tone conversion process according to a parameter, and the area determined by the determining step that said inner boundary region (el) performs second image tone conversion processing in accordance with the against parameter to the image data of said region determining step said outer boundary region determined by (el5) A third tone conversion process is performed on the image data according to the parameter, and a fourth tone conversion is performed on the image data other than the area (sq) corresponding to the face and the boundary area (el, el5) according to the parameter. A tone conversion processing step for performing processing, wherein the tone conversion processing step includes a step of forming a brushstroke by a group of a plurality of pixels of the image data, and the pixels of the image data are represented by A (RGB) (X, y) (where (x, y) is a two-dimensional coordinate, (RGB) is the RGB value of the pixel A) The parameter is represented by P, and the influence of the parameter is represented by α The image data of the area corresponding to the face is subjected to processing represented by A (RGB) (x, y) = A (RGB) (x, y) * P * α1 (x, y) A (RGB) (x, y) is added to the image data of the inner boundary region. = A (RGB) (x, y) * P * α2 (x, y) is executed, and A (RGB) (x, y) = A (RGB) is applied to the image data of the outer boundary region. ) (X, y) * P * α3 (x, y) is executed, and an image processing method is provided that satisfies α1>α2> α3.
According to a fifth aspect of the present invention, there is provided a computer of an image processing apparatus capable of converting image data including a face into a plurality of types of image tone, and a selection means for selecting one type from the plurality of types of image tone. , A face detection unit for detecting a face from the image data, and a region determination unit for determining a region (sq) corresponding to the face detected by the face detection unit and inner and outer boundary regions (el, el5) And converting the image data into the type of image tone selected by the selection means, the image data of the area (sq) corresponding to the face determined by the area determination means is used as a parameter. In response, a first tone conversion process is performed, and a second tone conversion process is performed on the image data of the inner boundary area (el) determined by the area determination unit according to a parameter. Third image tone conversion processing is performed on the image data of the outer boundary area (el5) determined by the means according to the parameters, and the area other than the area (sq) corresponding to the face and the boundary area (el, el5) The image data of the image data is made to function as a tone conversion processing unit that performs a fourth tone conversion process according to the parameters, and the tone conversion processing unit forms a brushstroke by a group of a plurality of pixels of the image data. And the pixel of the image data is represented by A (RGB) (x, y) (where (x, y) is a two-dimensional coordinate, and (RGB) is the RGB value of the pixel A) When the parameter is represented by P and the influence of the parameter is represented by α, A (RGB) (x, y) = A (RGB) (x, y) is added to the image data of the region (sq) corresponding to the face. ) * P * α1 (x, y) And the process represented by A (RGB) (x, y) = A (RGB) (x, y) * P * α2 (x, y) is performed on the image data of the inner boundary region, A process represented by A (RGB) (x, y) = A (RGB) (x, y) * P * α3 (x, y) is performed on the image data of the outer boundary region, and α1> An image processing program characterized by α2> α3 is provided.

  According to the present invention, it is possible to suitably perform image conversion processing for converting a photographic image showing a person into a painting-like image.

  In particular, according to the present invention, by applying different image tone processing to the face portion and the non-face portion of the person, the unnaturalness in the whole picture-like image is eliminated, and the face of the person looks natural. There is an excellent effect that picture conversion can be realized.

The block diagram which shows the structure of the network system of the image display system by embodiment of this invention The block diagram which shows the structure of the image service site 10 by the embodiment The figure which shows the attribute information of the user ID by the embodiment The figure which shows the data structure of the work by the embodiment Flowchart for explaining a picture conversion process according to the embodiment Flowchart for explaining face detection processing according to the embodiment The figure for demonstrating creation of the face area | region by the embodiment (A), (b), (c) is a figure for demonstrating the importance map by the embodiment. The figure for demonstrating the boundary of the area | region of the face by the embodiment The figure which shows the exchange of the data between the terminal and server by the embodiment Flowchart for explaining download in the embodiment

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
Embodiment

  FIG. 1 is a block diagram showing the configuration of the network system of the image display system according to the present embodiment.

  Reference numeral 10 in the figure denotes an image service site having an SNS (social network service) function, and is connected to a plurality of terminals 1-1, 1-2, 1-3,. The terminal may be a normal personal computer.

  FIG. 2 is a block diagram showing the configuration of the image service site 10.

  A server has a member area 100, a common area 200, and a control area 300 in the server. Reference numeral 20 denotes a CPU for controlling the whole, and reference numeral 30 denotes a work memory when the CPU 20 performs work.

  The member area 100 is an area provided for each registered member, and has a user ID 102 for identifying the member and various attribute information 104 for each user correspondingly.

  As shown in FIG. 3, the attribute information 104 is user-specific information such as name, handle name, gender, age, region, membership type, and the like. The membership type includes a free member who has just registered for free, a paid member who has registered for a fee, and a premium member who satisfies a predetermined condition.

  Reference numeral 106 denotes an upload image area in which a user uploads a photo, 108 denotes a work buffer area in which a work whose picture has been converted by the image service site 10 is stored, and 110 denotes a download work in which a work of another user is downloaded and stored. It is a buffer area.

  The common area 200 is an area provided in common to all users, and includes a work area 202 in which a number of works obtained by converting an image uploaded by the user into a painting style are stored.

  Each work stored in the work area 202 stores an image ID 204 for identifying the work and attribute information 206 of the work together with image data 208.

  As shown in FIG. 4, the attribute information 206 includes a user ID indicating a contributor of the work, date information such as a creation date or a posting date, image type information such as the size of image data and a type of image, Access number information indicating the number of times of browsing, download number information indicating the number of times the work has been downloaded, point information indicating evaluation of the work, and other information unique to the work.

  The control area 300 includes a picture conversion processing unit 302 that stores an image processing program for performing picture conversion processing on image data, a parameter table 304 that is referred to during picture conversion processing and stores parameters for picture conversion, and picture conversion processing. The importance map 306 for storing the importance of each pixel in the image data, the display control unit 308 for storing the control program for displaying the work on the screen, the evaluation of the work according to the number of accesses, etc. And an evaluation control unit 310 that stores a control program for assigning points to a work, and a member management unit 312 that manages members according to user IDs and controls services for free members, paid members, and premium members. The member management unit 312 also manages the SNS function of the image service site 10.

  In addition, the control area 300 is provided with a face detection processing unit 314 that detects whether or not a face portion of a person is included in the image data and determines an area corresponding to the face by the detection.

  Hereinafter, a specific operation of the image display system of the present embodiment will be described with reference to the flowchart of FIG.

  First, the painting conversion processing technique will be described.

  Picture conversion means that each pixel constituting an image such as a photo is converted according to predetermined parameters (painting conversion parameters), oil painting, watercolor, pastel, colored pencil painting, crayon painting, illustration painting, dot drawing. This is an image processing technique for converting into a so-called painting tone, such as a tone, an airbrush, a silk screen tone, an embroidery tone, and a collage tone.

  Basically, various effect processing parameters known for photo retouching software, etc. are adjusted and combined to transform them so that they look like paintings.

  As effect processing, for example, texture processing is performed by mapping a texture to an image to give a special texture, and the image is classified into a texture portion and a flat portion such as a contour portion, a fine pattern, etc. Resolution processing that enhances resolution, HSV processing that adjusts the color by classifying it into three elements of Hue, Saturation, and Lightness (Value), R (Red) G (Green) B (Blue) ) RGB processing for adjusting the degree of each color, RGB exchange processing for exchanging in the direction of R → G → B, edge extraction processing for applying a filter called a Laplacian filter, intermediate density extraction processing for applying a filter called a median filter, adjacent pixels The RGB histogram is extracted, and the density extraction process is performed when the minimum / intermediate / maximum density is extracted, respectively, the darkest part of the image is black, The white part is white and the histogram between them is distributed as appropriate, the contrast is corrected and the histogram of the image is expanded, the gamma correction process that adjusts the intermediate brightness while maintaining the bright and dark parts, the image Shadow processing that brightens dark areas or darkens bright areas, Solarization processing that inverts RGB values when the RGB value of each pixel is equal to or higher than the threshold value, draws dots randomly, There is noise addition processing that generates noise and adjusts the amount and color of noise.

  In addition, a photo with a wide dynamic range that cannot be represented by a normal photo called HDR (High Dynamic Range) is included in a narrow dynamic range width by tone mapping to correct overexposure and overexposure blackout. There is also an effect process that increases expressiveness.

  In other words, the picture conversion process is a kind of effect process in principle, and the picture conversion algorithm creates various painting styles by combining the effect process and parameters, and the conversion algorithm is programmed.

  In the conversion algorithm, a parameter group P that is a set of the parameters for converting the pixels of the original image is prepared in advance. If there are 12 types of image tone types, the parameter groups are represented as P (1) to P (12). The tone of the converted image also changes depending on how the parameters are determined. Assuming that a parameter group for converting to an oil painting tone is P (1), and there are m parameters necessary for effect processing for converting an image so that it looks like an oil painting tone in P (1), the parameter P1 It represents -Pm.

  A pixel is generally expressed by gradations of a plurality of bits for each of RGB, but a parameter is an element for performing an operation on the bit data. For example, when the red component is emphasized to slightly suppress the green component and the blue component, the above “2”, “0.9” in the case of calculation such as R × 2, G × 0.9, B × 0.5 "," 0.5 "is a parameter.

  Alternatively, the calculation elements are programmed in advance, and R × 1.1 when the red emphasis level is 1, R × 1.2 when the emphasis level is 2, and R × 1.3 when the emphasis level is 3. The degree of emphasis when performing simple calculations is also a parameter.

  In addition, a predetermined calculation may be performed between the target pixel and adjacent pixels. For example, the pixel A1 (RGB), A2 (RGB), A3 (RGB), A4 (RGB), A5 (RGB), A6 (RGB), A7 (RGB), A8 (RGB) around the target pixel A5 (RGB) , A9 (RGB) are aligned vertically and horizontally, A5 (R) = A1 (R) × q1 + A2 (R) × q2 + A3 (R) × q3 + A4 (R) × q4 + A5 (R) The calculation of * q5 + A6 (R) * q6 + A7 (R) * q7 + A8 (R) * q8 + A9 (R) * q9 is performed, and the same calculation is performed for G and B. Here, q is a parameter, and different effect processing can be performed by changing this numerical value (coefficient).

  There is also an effect process that processes the touch (hereinafter referred to as “brush touch”) differently.

  Specifically, pixels having high correlation are collected from pixels constituting image data on the basis of color or the like to form a group. Subsequently, for each group of pixels, for the same group of pixels, the color is replaced with the representative color of the group. A group of pixels in which the colors are replaced with the representative colors in this way form each brush stroke. According to such image processing, by changing the correlation of the pixels collected as a group and the shape (length (distance), flatness) of the group composed of pixels of the same color (representative color) The formed brush can be changed, and as a result, the tone of the image represented by the image data can be changed to various painting tone. For each image tone, an optimum combination of parameters such as the degree of correlation used in each process, the length representing the shape of the group, and the flatness is set.

  Changing each of the plurality of pieces of image data to a different stroke can be realized by adjusting the shape of the group in which the pixels constituting the stroke are collected, for example, the flatness of the group when changing the image tone. When the flatness of the group composed of pixels of the same color representing the brush stroke is increased, the brush stroke becomes thick, and as a result, the touch of the image is displayed roughly. When the flatness of the group composed of pixels of the same color is reduced, a small brush can be created, and as a result, the touch of the image is displayed finely.

  The picture conversion algorithm described in detail above is stored in the image conversion processing unit 302 of the control area 300 of the image service site 10, and the parameter group is also stored in the parameter table 304 of the control area 300.

  FIG. 5 is a flowchart for performing a picture conversion process.

  The CPU 20 executes this flowchart while using the work memory 30.

  In FIG. 5, first, preprocessing is performed (step S501). This process is executed in common regardless of the type of image tone.

  Since the effect processing as described above is performed on each pixel, the image data needs to be in a bitmap format. Therefore, image data normally expressed in the JPEG format is converted into a bitmap format.

  Also, since the size of the uploaded image data is various, this is resized to the number of pixels in the display area, for example, 800 × 600 pixels. Larger images are reduced and smaller images are enlarged. This is because the parameters are fixed when the size is determined, and the processing is efficient. Of course, at the time of picture conversion of the main image, which will be described later, the picture conversion is performed for the size of the uploaded image data as it is.

  Next, in step S502, it is determined whether the photo to be converted includes a face. This is because the face becomes unnatural when subjected to extreme transformation, so the face portion is particularly finely processed for painting conversion.

  This face detection process will be described later with reference to the flowchart of FIG. 6, but the importance changes in the face area and background, the central part and the peripheral part in one image, so an importance map is created, Stored in the importance map 306 of the control area 300.

  In step S504, an image tone to be converted is selected from an oil painting tone, a watercolor painting tone, a pastel tone, a color pencil drawing tone, and the like.

  When an image tone is selected, the flow goes to the flow of each image tone conversion algorithm. For example, if it is an oil painting tone conversion, the process proceeds to step S506, and if it is a watercolor painting tone conversion, the process proceeds to step S510. Otherwise, the flow proceeds to the flow of other image tone conversion algorithm (step S520).

  When executing each algorithm, the control area parameter table 304 and importance map 306 are referred to.

  Here, the image data of the display screen size is converted into a picture while viewing the display screen. If it is determined in step S522, the process proceeds to the picture conversion of the main image (step S524).

  When redoing with another image tone, the process returns to step S504.

  In step S524, tone conversion processing is performed on the main image with reference to the parameter table and the importance map 306 using the tone conversion algorithm determined in step S522 (steps S526 and 527). Even if the image tone is the same, the appropriate parameter differs when the image size is different. Therefore, the parameter table used in step S526 is different from the parameter acquired in step S508 or step S512.

  In the post-processing in step S528, the data is converted again into the JPEG format.

  Next, face detection and importance map creation will be described with reference to FIG.

  First, in step S702, it is recognized whether or not there is a face in the target image data.

  Here, since the face recognition technique is known, it will not be described in detail. Although it is possible to recognize and detect the face, it is difficult to accurately extract the contour, so an elliptical or rectangular area including the face area is created (step S704).

  FIG. 7 is a diagram for explaining this situation, where sq is a rectangular region including the face F1, and el is an elliptical region including the face F1. That is, it is an area corresponding to the detected face.

  FIG. 8 is a diagram for explaining the importance map. When the faces F1e, F2e, and F3e are recognized from the image data as shown in FIG. 9A, face areas sq1, sq2, and sq3 are created as shown in FIG.

  FIG. 6C shows the importance given according to the area on the screen. First, the importance of the areas E1, E2, and E3 of the face areas sq1, sq2, and sq3 is set to first. Next, the importance of the central region E4 indicated by the ellipse el3 is set to the second place. Furthermore, the importance of the area E5 of the background portion indicated by the ellipse el4 on the screen is set to the third place. Finally, the importance of the peripheral part of the screen is set to the lowest.

  That is, FIG. 8C is an importance map. When picture conversion processing is performed on each pixel of the image data, the more important area is more finely converted, and the less important area is more coarsely converted. As a result, the peripheral area E5 having the lowest importance level is transformed into a picture with a blurred feeling. Of course, this is only an example, and the face regions E1, E2, and E3 may not be subjected to painting conversion processing so as not to destroy the facial atmosphere.

When this importance is indicated by α, α has the same number of pixels as the image data, and the conversion operation of the pixel A is
A (RGB) (x, y) = A (RGB) (x, y) * P1 * α (x, y)
A (RGB) (x, y) = A (RGB) (x, y) * P2 * α (x, y)
A (RGB) (x, y) = A (RGB) (x, y) * Pn * α (x, y)
(Where (x, y) is a two-dimensional coordinate). In other words, this means that the influence degree of the parameter changes depending on the value of α.

  Even if the importance is not set to the same number of pixels as the image data, six types of importance data may be prepared as long as the insides of the areas E1 to E6 are the same importance. Further, the overlapping areas may be further subdivided, for example, to be an intermediate value.

  In other words, the importance level map 306 can be said to be a filter having auxiliary data that allows the transmission information to be appropriately adjusted separately from the color expression data R, G, and B for each pixel in the image processing field. When the tone conversion process is performed using the importance map 306 in combination, the tone conversion process can be performed on the areas E1 to E6 of the image data 208 with different conversion strengths. This includes the case where the conversion intensity is not changed at all (the parameter is not affected).

  FIG. 9 is a further subdivision of the contour area (boundary area) of the face area. If the importance is changed between inside and outside the outline of the face area, there is a possibility that a step will be formed in the picture converted into the picture. Therefore, in the figure, the importance around the face area is expressed as sq> el> el5. To do. By doing so, it is possible to transform the picture of the contour portion of the face step by step with a natural feeling.

  In the image display system of the present embodiment, as shown in FIG. 10, subsequently, an image is uploaded, displayed, viewed, and downloaded between the terminal and the server of the image service site.

  This will be briefly described. First, a user who is a member accesses the image service site 10 from his / her terminal 1 (SB2). The server authenticates the user ID by logging in the user (SC2), and when it is confirmed that the user is a member, the server transmits the member's page so that the member can view it (SC4).

  When the member opens his / her page, the member uploads an image (SB4). The uploaded image is stored in the upload image area 106 of the member area 100 of the server. When the type of painting tone is selected to convert this image into a painting, the painting is converted according to the flow of FIG. 5 (SC6), and the converted work is stored in the work buffer 108. You can enjoy it as it is, but if you want to share it with other users, post it (SB8). The submitted work is transferred from the work buffer 108 in the member area 100 to the work area 202 in the common area 200 and stored. The attribute information as described in FIG. 4 is also added to the works in the work area 202.

  Since other users can view works stored in the work area 202 of the common area 200 even if they are not members (SA2), the image service site 10 displays many works so that they can be easily viewed. (SC8).

  Points are added even if the work is just viewed (SC10). If the user likes the GOOD button and presses the GOOD button (SA4), more points are added (SC12).

  When the user downloads the work (SA6), if the work is a download permission target, the download is permitted (SC14), points are further added (SC16), and the download is performed (SA8).

  Although not described in detail here, when an image that is not permitted to be downloaded is downloaded, a logo called a tag is added to the work.

Further, in the image service site 10 of the present embodiment, downloading is performed as shown in FIG. Here, membership registration is required for downloading.
Then, the member designates a desired work while viewing the display screen (step S802). The server determines whether the user is a paid member or a free member by the member management unit 312 (step S804). If the user is a paid member, the server proceeds to step S812 and can download it unconditionally. In the case of a free member, the point corresponding to the evaluation of the work is read from the attribute information of the corresponding work in step S806, and the retained points are read from the attribute information of the member who is going to download (step S808). Payment points are calculated from the difference between the points (equivalent to the price) of the work and the points held by the member (step S810). If satisfied, the process proceeds to step S812. If the work is insufficient, the work is downloaded. Since it is not possible, the process returns to step S802 to select another work. If reading of the work is permitted in step S812, points are added to the creator of the work and points are added to the point area of the attribute information of the image ID of the work (SC16 in FIG. 10). In other words, as the number of downloads increases, the evaluation of the work increases, and points are added to the point area of the attribute information of the user ID of the member who is the creator, thereby increasing the number of points owned by the member.

  Furthermore, since the image service site 10 has an SNS function, a comment on another member's work can be written, a reply to it can be written, a group of members with similar preferences can be created, and a fan of a specific author You can create a group. Since such a mechanism is already implemented in an existing SNS, it will not be described in detail, but the member management unit 312 realizes the SNS mechanism together with the member management mechanism.

  According to the above configuration, it is possible to suitably perform image conversion processing for converting a photographic image showing a person into a painting-like image. In particular, by changing the conversion intensity between the face part of the person and the part other than the face, it is possible to eliminate the whole unnaturalness in the painting-like image and realize the picture conversion in which the face of the person looks natural. Is.

  It should be noted that the present invention is not limited to the configuration described in the above-described embodiment, and the configuration of each unit can be modified or changed as appropriate.

  For example, an area corresponding to a person's face may be cut out from the image data 208, and the area corresponding to the face and other areas may be subjected to separate tone conversion and then combined.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to these, The invention described in the claim and its equal range are included.

The invention described in the scope of claims of the present application will be added below.
(Appendix)

  According to the first aspect of the present invention, a face detection unit that detects a face from image data, a region determination unit that determines a region corresponding to the face detected by the face detection unit, and the image data Image tone conversion in which the first image tone conversion process is performed on the image data of the area corresponding to the face detected by the area detection means, and the second image tone conversion process is performed on the image data other than the area corresponding to the face. And a processing means. Thereby, for example, it is possible to perform the tone conversion with different conversion strengths in the region corresponding to the face and other regions, and it is possible to perform natural tone conversion in a photograph including the face.

  According to a second aspect of the present invention, in the first aspect, the region corresponding to the face determined by the region determining means is a rectangular or elliptical region including the face. Thereby, even if the outline of the face cannot be accurately extracted, it is possible to perform the tone conversion of the area corresponding to the face.

  According to a third aspect of the present invention, in any one of the first and second aspects, the region corresponding to the face determined by the region determining means includes a boundary region. Thereby, it can be avoided that a step is formed on the inside and outside of the boundary portion of the face.

  According to a fourth aspect of the present invention, in the third aspect, the image tone converting means performs a third image tone converting process on the image data in the boundary area. This makes it possible to perform natural tone conversion in a stepwise manner on the face contour portion.

  According to a fifth aspect of the present invention, in any one of the first and second aspects, the first tone conversion processing and the second tone conversion processing of the tone conversion means are different in conversion strength of the tone conversion. Features. As a result, the area corresponding to the face and other areas can be converted to the same image tone, and only the conversion intensity can be changed.

  According to a sixth aspect of the present invention, in the fourth aspect, the first tone conversion processing, the second tone conversion processing, and the third tone conversion processing of the tone conversion means are: Are different from each other. Thus, the first image tone conversion processing, the second image tone conversion processing, and the third image tone conversion processing are the same image tone conversion processing, and only the conversion intensity can be changed.

  According to the seventh aspect of the present invention, a face is detected from the image data, a region corresponding to the detected face region is determined, and a first tone conversion process is performed on the image data of the detected region. At the same time, it is possible to obtain a method for performing the second image tone conversion process on the image data other than the area.

    According to an eighth aspect of the present invention, a computer of an image processing apparatus that performs tone conversion processing of image data includes a face detection unit that detects a face from the image data, and a region corresponding to the face detected by the face detection unit. Region determining means for determining image quality, and image processing for performing a first image tone conversion process on the image data of the region detected by the region detection unit and performing a second image tone conversion process on image data other than the region A program that functions as processing means can be obtained.

1 Terminal 10 Image Service Site 20 CPU
30 Work memory 100 Member area 102 User ID
104 Attribute information 200 Common area 202 Work area 204 Image ID
206 Attribute Information 208 Image Data 300 Control Area 302 Painting Conversion Processing Unit 304 Parameter Table 306 Importance Map 308 Display Control Unit 314 Face Detection Processing Unit 316 Face Detection Unit 317 Face Area Determination Unit 318 Tone Conversion Processing Unit 310 Evaluation Control Unit 500 Internet

Claims (5)

In an image processing apparatus capable of converting image data including a face into a plurality of types of image tone,
Selecting means for selecting one type from the plurality of types of image tone;
Face detection means for detecting a face from image data;
Area determining means for determining an area (sq) corresponding to the face detected by the face detecting means and inner and outer boundary areas (el, el5) ;
In converting the image data into the type of image tone selected by the selection means,
A first image tone conversion process is performed on the image data of the area (sq) corresponding to the face determined by the area determining means according to the parameter,
A second image tone conversion process is performed on the image data of the inner boundary area (el) determined by the area determination means according to a parameter,
A third image tone conversion process is performed on the image data of the outer boundary area (el5) determined by the area determination means according to a parameter,
A tone conversion processing unit that performs a fourth tone conversion process on image data other than the region (sq) corresponding to the face and the boundary region (el, el5) according to a parameter;
The image tone conversion processing means includes means for forming a brushstroke by a group of a plurality of pixels of the image data,
The pixel of the image data is represented by A (RGB) (x, y), where (x, y) is a two-dimensional coordinate and (RGB) is the RGB value of the pixel A)
The parameter is represented by P,
When the influence of the parameter is represented by α,
In the image data of the area (sq) corresponding to the face, A (RGB) (x, y) = A (RGB) (x, y) * P * α1 (x, y)
The process represented by
In the image data of the inner boundary area, A (RGB) (x, y) = A (RGB) (x, y) * P * α2 (x, y)
The process represented by
The outside of the image data of the boundary area A (RGB) (x, y ) = A (RGB) (x, y) * P * α3 (x, y)
And the process represented by
α1>α2> α3
An image processing apparatus characterized by the above. The image processing apparatus includes an image service site connected to a plurality of terminals via the Internet,
Further, the image service site comprises a parameter table for storing the said image-style conversion process unit parameter,
2. The image processing apparatus according to claim 1, wherein the image tone conversion processing unit converts image data uploaded from the terminal to the image service site into a selected type of image tone. The image service site has a social network service function, and a member uploads image data to the image service site and converts the image data into a selected type of image tone by the image tone conversion processing means. The converted image data is disclosed so that it can be viewed by other members, and the other members satisfying a predetermined condition can download the converted image data. The image processing apparatus according to claim 2. In an image processing method capable of converting image data including a face into a plurality of types of image tone,
A selection step of selecting one type from the plurality of types of image tone;
A face detection process for detecting a face from image data;
A region determining step for determining a region (sq) corresponding to the detected face and inner and outer boundary regions (el, el5);
In converting the image data into the type of image selected in the selection step ,
A first image tone conversion process is performed on the image data of the region (sq) corresponding to the face determined by the region determination step according to the parameter,
Performs second image tone conversion processing in accordance with the against parameter to the image data of said region determining step said inner boundary region determined by (el),
A third image tone conversion process is performed on the image data of the outer boundary area (el5) determined by the area determination step according to a parameter,
A tone conversion processing step of performing a fourth tone conversion process on image data other than the region (sq) corresponding to the face and the boundary region (el, el5) according to a parameter;
Including
The image tone conversion processing step includes a step of forming a brushstroke by a group of a plurality of pixels of the image data,
The pixel of the image data is represented by A (RGB) (x, y), where (x, y) is a two-dimensional coordinate and (RGB) is the RGB value of the pixel A)
The parameter is represented by P,
When the influence of the parameter is represented by α,
A (RGB) (x, y) = A (RGB) (x, y) * P * α1 (x, y) in the image data of the area corresponding to the face
The process represented by
In the image data of the inner boundary area, A (RGB) (x, y) = A (RGB) (x, y) * P * α2 (x, y)
The process represented by
In the image data of the outer boundary area, A (RGB) (x, y) = A (RGB) (x, y) * P * α3 (x, y)
And the process represented by
α1>α2> α3
An image processing method characterized by the above. An image processing apparatus computer capable of converting image data including a face into a plurality of types of image tone,
Selecting means for selecting one type from the plurality of types of image tone;
Face detection means for detecting a face from image data;
And functioning as area determining means for determining the area (sq) corresponding to the face detected by the face detecting means and inner and outer boundary areas (el, el5) ,
In converting the image data into the type of image tone selected by the selection means,
A first image tone conversion process is performed on the image data of the area (sq) corresponding to the face determined by the area determining means according to the parameter,
A second image tone conversion process is performed on the image data of the inner boundary area (el) determined by the area determination means according to a parameter,
A third image tone conversion process is performed on the image data of the outer boundary area (el5) determined by the area determination means according to a parameter,
A tone conversion processing means for performing a fourth tone conversion process on image data other than the region (sq) corresponding to the face and the boundary region (el, el5) according to a parameter;
To function,
The image tone conversion processing means includes means for forming a brushstroke by a group of a plurality of pixels of the image data,
The pixel of the image data is represented by A (RGB) (x, y), where (x, y) is a two-dimensional coordinate and (RGB) is the RGB value of the pixel A)
The parameter is represented by P,
When the influence of the parameter is represented by α,
In the image data of the area (sq) corresponding to the face, A (RGB) (x, y) = A (RGB) (x, y) * P * α1 (x, y)
The process represented by
In the image data of the inner boundary area, A (RGB) (x, y) = A (RGB) (x, y) * P * α2 (x, y)
The process represented by
The outside of the image data of the boundary area A (RGB) (x, y ) = A (RGB) (x, y) * P * α3 (x, y)
And the process represented by
α1>α2> α3
An image processing program.

Images