JP2005267454A - Image trimming device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image trimming device capable of automatically and properly trimming a portrait. <P>SOLUTION: This trimming device for photographic image including portrait comprises a face and person area specification means 14 detecting the face of a person from a photographic image and calculating a face area including all the detected face; a means 13 for setting a predetermined ratio for face position, which sets the vertical and horizontal positions of the face area in the photographic image after trimming; and a trimming means 16 for trimming the photographic image so that the face area is located in the set vertical and horizontal positions. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an image trimming apparatus and method suitable for use in performing trimming processing so that a human face is placed at an appropriate position on a photographic image including a human photograph.

  An electronic photograph taken with a digital camera or the like is appropriately trimmed in accordance with a predetermined image resolution (vertical and horizontal size of the image) and collected into an electronic album. The electrophotographic trimming is generally performed manually.

As an example of the related art regarding this, there is a method described in Patent Document 1. In this method, an electronic photograph is dragged with a mouse so that a portion to be cut out from the electronic photograph is superimposed on a trimming frame displayed on a display of a photo editing apparatus such as a personal computer. The trimming range of the electrophotographic image is determined by placing an image area desired by the editor in the trimming frame.
JP-A-10-261066

  However, in the above method, there is an inconvenience that the editor has to perform manual work to determine the trimming range in the photograph. This is an obstacle for applications such as automatically generating a photo album consisting of portraits.

  SUMMARY OF THE INVENTION In view of the problems of the conventional example, the present invention aims to simplify the image processing operation for a photographic image including a human photograph, and can automatically and appropriately trim a human photograph. An object of the present invention is to provide an image trimming apparatus and method.

  In order to solve the above-described problems, the present invention is a photographic image trimming apparatus including a person image, which detects a human face from a photographic image and calculates a face area including all detected faces. Means, image position setting means for setting the vertical and horizontal positions of the face area in the cropped photographic image, and the vertical and horizontal directions set by the image position setting means for the face area calculated by the face area calculation means Trimming means for trimming a photographic image so as to be placed at the position.

  According to the above configuration, the photographic image is trimmed by the trimming means so that the face area calculated by the face area calculating means is placed at the vertical and horizontal positions set by the image position setting means. In other words, according to the present invention, trimming is performed in accordance with the position of the face in the photographic image, so that it is possible to automatically perform trimming of a human photograph appropriately and not manually.

  The present invention is characterized in that the trimming means trims a photographic image so that the center point of the face region is located at the golden division point in the vertical direction of the image after trimming and at the center point in the horizontal direction. And According to this, it is possible to arrange the positions of human faces in the photographic image after trimming with good balance in the horizontal and vertical directions.

  The present invention further includes person area specifying means for specifying a person area that will include all persons based on the calculation result of the face area calculating means, wherein the trimming means is the person specified by the person area estimating means. The photographic image is trimmed so that the region is entirely included in the trimmed photographic image. According to this, the face of the person and the entire position of the person in the photographic image after trimming can be arranged with good balance.

  According to another aspect of the present invention, there is provided a method for trimming a photographic image including a person image, wherein a face area calculation process for detecting a human face from a photographic image and calculating a face area including all detected faces is provided. An image position setting process for setting the vertical and horizontal positions of the face area in the cropped photographic image, and the face area calculated in the face area calculation process in the vertical and horizontal directions set in the image position setting process. And a trimming process for trimming the photographic image to be placed in position.

  According to still another aspect of the present invention, there is provided a program for trimming a photographic image including a person image, detecting a human face from the photographic image, and calculating a face area including all detected faces. A face area calculation process, an image position setting process for setting the vertical and horizontal positions of the face area in the cropped photographic image, and the face area calculated in the face area calculation process is set in the image position setting process. And a description for performing by a computer a trimming process for trimming a photographic image to be placed in a horizontal position.

  Embodiments of an image trimming apparatus according to the present invention will be described below with reference to the drawings. The present embodiment includes a personal computer, a mobile phone terminal, a digital camera, a computer and peripheral devices constituting the personal information terminal, and a program executed by the computer. Photo image data subject to image processing is taken by a digital camera or the like, input to an information device constituting this embodiment via a predetermined recording medium or a communication line, and stored in a predetermined storage device. Shall.

  With reference to the block diagram showing the configuration of the present embodiment shown in FIG. 1 divided into blocks for each function and the flowchart showing the flow of processing of the present embodiment shown in FIG. 2, the configuration of the present embodiment and The operation will be described.

  The person photograph image input means 11 inputs photograph image data to be processed in accordance with a predetermined operation of the user (step S10 in FIG. 2). The input means 12 for the desired image resolution inputs the vertical and horizontal sizes of the output image after trimming (step S11 in FIG. 2). The unit is a pixel. Before image processing (trimming processing) in accordance with a user instruction or a predetermined definition so as to correspond to a predetermined output format using a user interface or other method of an application program that uses or constitutes the present embodiment ) Set before.

  The present embodiment provides a photographic image trimming function. That is, an output image is created by omitting unnecessary portions around a person's face (or the whole) from a photographic image. Therefore, in the present embodiment, the process is terminated when the photo resolution (input image size) is smaller than the desired image resolution (output image size) (“No” in step S12 in FIG. 2).

  The predetermined ratio setting means 13 for the face position sets a predetermined ratio for the position of the face area in the output image from a predetermined value or according to a user operation (step S13 in FIG. 2). The face area is an area occupied by the face (or head) of the person area. In a photographic image in which a plurality of persons are photographed, an area that covers the face areas of all persons is the face area. That is, a circumscribed rectangle of the detected face of one or more persons is a face area. The predetermined ratio setting means 13 for the face position designates the position of the face area in the vertical direction of the output image. Specifically, it is set by the ratio of the length from the upper end of the output image to the center point of the face area and the length from the center point of the face area to the lower end of the output image.

  FIG. 3 is a diagram illustrating an example of the relationship between the photographic image 1, the output image (trimmed photographic image) 2, and the like. In the example shown in FIG. 3, the faces 5 of three persons are included in the photographic image 1, and their circumscribed rectangles 4 are the face areas. The person area 3 is an area that is estimated to contain the whole image of the person according to the size and orientation of the face 5 of the person. In the example of FIG. 3, the ratio relating to the face position set by the setting means 13 for the predetermined ratio relating to the face position is the ratio of the vertical position of the face area 4 to the height h in the vertical direction of the output image 2 (38 .2% / 61.8%). This ratio is set in advance by the user interface of an application program that uses or configures the present embodiment or other methods, and is acquired before the trimming process. In this embodiment, this ratio is set to the golden division point determined by 0.382: 0.618. This golden dividing point is a dividing point determined by the golden ratio (1 + √5) / 2. In this case, the center point C of the face area 4 corresponds to the golden division point in the vertical direction of the output image 2.

  The face and person area specifying means 14 performs processing for detecting the face 5 and calculating the face area 4 (step S14 in FIG. 2) and specifying (estimating) the person area 3 based on the detection result of the face area 4. (Step S15 in FIG. 2).

  The face 5 is detected based on the detection result of the skin color area, for example. The face detection is realized by repeating the detection of a single face 5 to detect all faces of one or more persons included in the photographic image 1. The detection of a single face 5 is realized in two steps: detection of a face skin color area and comparison between the face skin color area and a face pattern template. As a method for detecting a skin color region that is a candidate for a face, there is a commonly used detection method in the HSV color space. This is a detection method that uses the feature that the H component (hue) and S component (saturation) of the skin color pixel converge within a certain range on the HS plane of the HSV color space. Here, V is lightness. Specifically, an input image expressed in RGB (Red Green Blue) format or the like is converted into an HSV color system, and skin color pixels are detected in the HSV color space. And a flesh-color area | region is formed by connecting adjacent flesh-color pixels.

  Next, it is determined whether the detected skin color area is a face area. Judgment is performed by collating with a plurality of face patterns for different orientations prepared in advance. Specifically, first, the face candidate area detected by connecting the skin color pixels is enlarged / reduced to match the size of the face pattern, and the correlation with the face pattern is obtained. Then, it determines that a face candidate region having a correlation degree equal to or greater than a predetermined threshold value among the face patterns is an image of the face 5.

  Regarding the calculation of the face region 4, based on the detection result of one or more faces 5, for example, it is calculated as a circumscribed rectangle. That is, when only one person is detected from the photographic image 1 due to the number of faces 5 detected from the photographic image 1, the face area 4 is a rectangle that circumscribes the person's face 5. On the other hand, when there are a plurality of persons (faces) 5 in the photographic image 1, the face area 4 is a minimum rectangle that covers all the detected faces 5.

  Next, the person area 3 is specified based on the calculation result of the face area 4. That is, a reference face (one of a plurality of faces 5) that is an increment value of the face area 4 is obtained from the face area 4, and the face area 4 is determined in advance in four directions, up, down, left, and right. Increment by the number of 5b. The reference face 5b having the incremented area as the person area 3 is, for example, the face 5 having the largest area in the face area 4 as the reference face. The number of reference faces 5b is determined according to, for example, the ratio of average height to face length.

  The face and person area specifying means 14 further reduces the face area 2 and the person area 3 when the size of the person area 3 is larger than the size of the output image 2 (“Yes” in step S16 in FIG. 2). Is performed (step S17 in FIG. 2). 3 shows the case where the person area 3 is smaller than the output image 2, FIG. 4 shows the case where the person area 3 is larger than the output image 2, and FIG. 5 shows the face area 2 and the person area 3 after being reduced. The person area 3 of FIG.

  Next, the trimming means 15 in FIG. 1 calculates a trimming area (step S18 in FIG. 2). The contents of the trimming process will be described with reference to FIG. However, for the photographic image 1 to be processed, the person area 3 is smaller than the output image 2 in the image data passed from the face and person area specifying means 14 by the processing in steps S16 and S17 of FIG. It shall be adjusted to.

  FIG. 3 shows a photograph showing three people in a line. As described above, the three shaded circles represent the three faces 5, and the rectangle that circumscribes them is the face area 4. A broken-line rectangle surrounding the face area 4 represents the person area 3. A rectangle surrounding the person area 3 has a desired image resolution, that is, a trimming area (output image) 2.

  Of the dimensions shown in FIG. 3, W and H, and w and h are the resolution (number of pixels in the vertical and horizontal directions) of the input image (photo image) 1 and the desired output image (trimming area) 2, respectively. It is a known value input before trimming. x and y are the vertical and horizontal sizes of the specified person area 3, and are the results obtained from the face and person area specifying means 14. v and u are the lengths from the center point of the face area 4 to the upper end and lower end of the person area 3, respectively, and the results obtained from the face detection and person area specifying means 14. y1 is the length from the upper end of the person area 3 to the upper end of the trimming area 2, and y2 is the length from the lower end of the person area 3 to the lower end of the trimming area 2.

  In order to perform trimming so that the center point of the face area 4 is placed at a specified ratio, it is necessary to calculate y1 and y2. From FIG. 3, among the various quantities u, v, y, y1, y2, and h, the simultaneous equations consisting of the following equations [1], [2], and [3] hold.

  Here, k (= 0.618) is a constant for describing the predetermined ratio. Solving this simultaneous equation yields the following equations for y1 and y2:

  In the horizontal direction, calculation is performed using x1 = x2 = (w−x) / 2. The trimming area (output image) 2 can be calculated by using the calculation formulas of y1 and y2.

  Then, the trimmed image output means 16 in FIG. 1 outputs the trimming area (output image) 2 to a predetermined output destination (or recording destination) (step S19 in FIG. 2).

  Next, a process for reducing the image and recalculating the face area and the person area in step S17 of FIG. 2 will be described. When the area obtained by the person area specifying process is larger than the desired image resolution in either the vertical or horizontal direction, a reduction operation is performed on the input image 1 so that the human area 3 is equal to or lower than the resolution of the desired image 2. And y1 and y2 are calculated so that the center point of the face region 4 is positioned at the golden division point in the vertical direction of the output image 2. With reference to FIG. 4, FIG. 5, and FIG.

  FIG. 4 is a diagram before reduction. FIG. 4 shows an example in which the length of the person area 3 in the vertical direction is longer than that of the trimming area 2. FIG. 5 is a diagram after reduction. Calculation of y1 and y2 in the vertical direction in the case shown in FIG. 4 before reduction is performed as follows.

(A) The entire photograph (the entire photograph image 1) is reduced by the ratio of the length in the vertical direction of the person area 3 to that of the trimming area. The length x0 in the horizontal direction of the reduced human area 3 is x * (h / y), and the length y ′ in the vertical direction is h (step S20 in FIG. 6).

(B) If the center point of the face area 4 is located at the golden division point of the trimming area 2 (that is, if “v0 / u0 = K” is satisfied (“Yes” in step S21 in FIG. 6), the calculation is completed. To do.

(C) If the center point of the face area 4 is not located at the golden division point of the trimming area 2 (“No” in step S21 in FIG. 6), y ′ is decremented by 2 pixels and x ′ (= x0 *) y ′ / h), u ′ (= u0 * y ′ / h) and v ′ (= v0 * y ′ / h) are recalculated (step S22 in FIG. 6). Then, y1 and y2 are calculated by the formula shown above (step S23 in FIG. 6).

(D) If the ratio of (v ′ + y1) to (u ′ + y2) is the golden division ratio, the calculation ends (“Yes” in step S24 in FIG. 6). Otherwise, the recalculation in (c) is repeated until appropriate y1 and y2 are found (from “No” in step S24 in FIG. 6 to step S22).

  As described above, in the present embodiment, when trimming a human photographic image 1 whose image resolution is higher than the trimming area 2, the face 5 is detected from the photographic image 1 and the person area 3 is specified. 4 is trimmed so that it is located at the golden dividing point in the vertical direction of the trimming region 2 and at the central point in the horizontal direction. As a result, compared to the conventional method, it is possible to automatically perform trimming with a better balance of a human photograph without human intervention.

  The embodiment of the present invention is not limited to the above-described one, and the processing steps are integrated or distributed, and the hardware for executing the processing steps is integrated, divided, or distributed. Can do. Moreover, the program which comprises embodiment of this invention can be distributed via a computer-readable recording medium or a communication line.

The block diagram for demonstrating embodiment of this invention. The flowchart for demonstrating the process of embodiment of FIG. Explanatory drawing which shows the photograph image 1 for demonstrating embodiment of FIG. The other explanatory view which shows the photograph image 1 for demonstrating embodiment of FIG. The other explanatory view which shows the photograph image 1 for demonstrating embodiment of FIG. The flowchart for demonstrating step S17 of FIG.

Explanation of symbols

1 photographic image, 2 output image (trimming area), 3 person area, 4 face area, 5 face, 11 human photograph image input means, 12 desired image resolution input means, 13 predetermined ratio setting means for face position, 14 Face and person area specifying means, 15 trimming means, 16 trimmed image output means

Claims (5)

A trimming device for a photographic image including a person image,
A face area calculating means for detecting a human face from a photographic image and calculating a face area including all the detected faces;
Image position setting means for setting the vertical and horizontal position of the face area in the cropped photographic image;
An image trimming apparatus comprising: trimming means for trimming a photographic image so that the face area calculated by the face area calculating means is placed at a vertical and horizontal position set by the image position setting means.   The trimming means trims a photographic image so that a center point of a face region is located at a golden division point in a vertical direction of the image after trimming and a central point in a horizontal direction. The image trimming apparatus according to 1. Further comprising a person area specifying means for specifying a person area that would include all persons based on the calculation result of the face area calculating means;
The image trimming apparatus according to claim 1, wherein the trimming unit trims a photographic image so that the person region specified by the person region estimation unit is included in the trimmed photographic image. A method for trimming a photographic image including a person image,
A face area calculation process for detecting a human face from a photographic image and calculating a face area including all detected faces;
An image position setting process for setting the vertical and horizontal position of the face area in the cropped photographic image;
A trimming process for trimming a photographic image so that the face area calculated in the face area calculation process is placed at the vertical and horizontal positions set in the image position setting process. . A program for trimming a photographic image including a person image,
A face area calculation process for detecting a human face from a photographic image and calculating a face area including all detected faces;
An image position setting process for setting the vertical and horizontal position of the face area in the cropped photographic image;
And a trimming process for trimming a photographic image so that the face area calculated in the face area calculation process is placed at the vertical and horizontal positions set in the image position setting process. An image trimming program characterized by that.

Images