JP2005229181A - Image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing method capable of correcting an image by solving a problem of prior arts that has only magnified a person region in the image and cannot have partially adjusted the image and corrected the attitude of people in the image. <P>SOLUTION: A flowchart for indicating image adjustment processing includes: a step S4 of extracting a feature point from a reference image; a step S5 of respectively extracting feature points corresponding to the feature point of reference image data from an object image; a step S7 of determining a target point on the basis of each feature point of the object image corresponding to each feature point of the reference image; and a step S10 of applying geometrical conversion processing to the object image in a way of making each feature point of the object image in matching with the target point, wherein the target point is selectable from a point for internally dividing an interval between each feature point of the reference image and each feature point of the object image at a prescribed ratio and for applying exterior division to an interval between each feature point of the reference image and each feature point of the object image by m:n (m, n are positive integers for satisfying a relation of m>n). <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an image processing method for correcting an image.

The ID photo attached when applying for a driver's license or passport is used to identify a person, so the size of the photo, the position of the person, etc. are determined for each ID photo. In order to create a photo that conforms to such a standard, the image is trimmed and scaled based on the shape and size of the person area in the image, so that the person in the image matches the ID photo standard. Detecting the subject's head from the subject image, and trimming the part including the head from the subject image so that it is similar to the required specification of the ID photo. Techniques for scaling the size have been proposed (see, for example, Patent Document 1 and Patent Document 2).
JP 2000-196980 A JP 2001-311996 A

  However, in the above-described conventional technique, only the person area in the image is scaled, and partial adjustment and correction of the posture of the person cannot be performed.

  The present invention has been made in view of the above-described problems in the prior art, and an object thereof is to correct an image.

  In the image processing method, the feature point is extracted from a reference image prepared in advance, the feature point corresponding to the feature point of the reference image is extracted from the target image to be processed, A target point is determined based on each feature point of the reference image and each feature point of the target image, and geometric conversion processing is performed on the target image so that each feature point of the target image matches the target point. It is characterized by performing.

  According to a second aspect of the present invention, in the image processing method according to the first aspect, the target point is each characteristic point of the reference image.

  According to a third aspect of the present invention, in the image processing method according to the first aspect, the target point is a point that internally divides each feature point of the reference image and each feature point of the target image by a predetermined ratio. It is characterized by being.

  According to a fourth aspect of the present invention, in the image processing method according to the first aspect, the target point is represented by m: n (where m and n are the characteristic points of the reference image and the characteristic points of the target image). A positive number satisfying m> n).

  The invention according to claim 5 is the image processing method according to claim 1, wherein the target point is a point that internally divides each feature point of the reference image and each feature point of the target image by a predetermined ratio. Alternatively, it is possible to select from any one of points that divide each feature point of the reference image and each feature point of the target image into m: n (m, n is a positive number satisfying m> n). It is characterized by that.

  According to a sixth aspect of the present invention, in the image processing method according to the fifth aspect, an internal ratio that internally divides each feature point of the reference image and each feature point of the target image, or of the reference image An external division ratio that externally divides each feature point and each feature point of the target image can be selected.

  According to a seventh aspect of the present invention, in the image processing method according to the fifth or sixth aspect, the target point determination method can be selected for each region of the target image.

  According to an eighth aspect of the present invention, in the image processing method according to any one of the first to seventh aspects, the reference image is used after being enlarged or reduced according to a target size. .

  The invention according to claim 9 is the image processing method according to any one of claims 1 to 8, wherein the geometric transformation process is a polygon having the target image as a vertex at each feature point of the target image. And the geometric transformation is performed for each of the divided areas.

  The invention according to claim 10 is the image processing method according to claim 9, wherein the polygon is a triangle and the geometric transformation is an affine transformation.

  The invention according to claim 11 is the image processing method according to claim 10, wherein the triangle is a triangle corresponding to Delaunay triangulation with the target point as a vertex.

  According to a twelfth aspect of the present invention, in the image processing method according to any one of the first to eleventh aspects, the target image is an ID photo image obtained by photographing in a predetermined posture. The reference image is a person model image having a face position and a face size corresponding to a predetermined ID photo type.

  A thirteenth aspect of the present invention is the image processing method according to any one of the first to twelfth aspects, wherein the reference image is a plurality of pre-captured images or feature points extracted from the plurality of images. It was created based on the average position information.

  According to a fourteenth aspect of the present invention, in the image processing method according to any one of the first to thirteenth aspects, a plurality of the reference images are prepared in advance and can be selected according to the target image. It is characterized by.

  According to the first aspect of the present invention, feature points are extracted from a reference image prepared in advance, feature points corresponding to feature points of the reference image are extracted from a target image to be processed, and each feature of the reference image is extracted. The target point is determined based on the point and each feature point of the target image, and geometric conversion processing is performed on the target image so that each feature point of the target image matches the target point. it can.

  According to the second aspect of the invention, the geometric transformation process is performed so that each feature point of the target image matches each feature point of the reference image, so that the image can be standardized.

  According to the third aspect of the present invention, the geometric transformation process is performed so that each feature point of the target image matches a point that internally divides each feature point of the reference image and each feature point of the target image at a predetermined ratio. Therefore, it is possible to standardize the image while leaving the feature of the target image.

  According to the fourth aspect of the present invention, the geometric transformation process is performed so that each feature point of the target image matches a point that divides each feature point of the reference image and each feature point of the target image into m: n. As a result, the deviation of each feature point of the target image from the reference image can be emphasized.

  According to the invention described in claim 5, the target point is a point that internally divides each feature point of the reference image and each feature point of the target image at a predetermined ratio, or each feature point of the reference image and the target image By making it possible to select any one of the points that divide each feature point into m: n, the target image can be corrected according to the user's request.

  According to the sixth aspect of the present invention, an internal ratio that internally divides each feature point of the reference image and each feature point of the target image, or each feature point of the reference image and each feature point of the target image. By making it possible to select the external division ratio to be divided, it is possible to adjust the degree of standardization of the target image and the degree of emphasizing deviation from the reference image according to the user's request.

  According to the invention described in claim 7, by making it possible to select a target point determination method for each region of the target image, different processing can be performed for each region of the target image.

  According to the invention described in claim 8, since the reference image is used after being enlarged or reduced according to the target size, it is not necessary to prepare a plurality of reference images in advance.

  According to the invention described in claims 9, 10 and 11, the geometric transformation process divides the target image into polygonal regions having the feature points of the target image as vertices, and performs geometric transformation for each divided region. Therefore, each conversion can be simplified.

  According to the twelfth aspect of the present invention, for an ID photo obtained by photographing in a predetermined posture based on a person model image having a face position and a face size corresponding to a predetermined ID photo type The image can be corrected.

  According to the invention described in claim 13, since the reference image is created based on a plurality of images taken in advance or average position information of feature points extracted from the plurality of images, Can be used as a reference image.

  According to the invention described in claim 14, a plurality of reference images prepared in advance can be selected according to the target image.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the scope of the invention is not limited to the illustrated examples. In the present embodiment, an ID photo image obtained by photographing in a predetermined posture is set as a target image, and a human model image having a face position and face size corresponding to a predetermined ID photo type is used as a reference. This will be described as an image.

FIG. 1 shows a functional configuration of an image processing apparatus 1 according to an embodiment of the present invention.
As shown in FIG. 1, the image processing apparatus 1 includes a control unit 2, an operation unit 3, a display unit 4, an image input unit 5, an image output unit 6, an image data memory 7, a reference image database 8, an image processing unit 9, The storage unit 10 is provided, and each unit is connected by a bus 11.

  The control unit 2 comprehensively controls each unit of the image processing apparatus 1 according to a control program stored in the storage unit 10.

  The operation unit 3 includes various input keys such as cursor keys and numeric keys, and outputs operation signals generated by pressing these keys to the control unit 2.

  The display unit 4 includes an LCD (Liquid Crystal Display), an EL (Electro Luminescent) display, and the like, and performs a required display process according to a display control signal input from the control unit 2.

  The image input unit 5 acquires target image data to be subjected to image processing in the image processing apparatus 1. The target image data is an ID photo image obtained by photographing with a predetermined posture. As the image input unit 5, for example, a digital camera that converts an image input via a lens into an electric signal by an image sensor such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor) and acquires image data, Examples include a scanner that optically scans an image printed on paper or film and acquires image data based on the intensity of reflected light or transmitted light, and a card reader that reads image data from a recording medium such as a memory card. It is done. The target image data acquired by the image input unit 5 is stored in the image data memory 7.

  The image output unit 6 outputs an image based on the adjusted image data after image processing stored in the image data memory 7. Examples of the image output unit 6 include a printer that prints an image, a recording device that records image data on a recording medium such as a CD-R (Compact Disc-Recordable), a monitor that displays an image, and the like.

  The image data memory 7 is composed of a semiconductor memory or the like, and stores target image data acquired by the image input unit 5 and adjusted image data generated by the image processing unit 9.

  The reference image database 8 is composed of a semiconductor memory or the like, and stores reference image data prepared in advance. The reference image data is a person model image having a face position and a face size according to the ID photo type, and a plurality of reference image data are prepared according to the use and the body shape. The reference image data is created based on a plurality of images taken in advance or average position information of feature points extracted from the plurality of images.

  The image processing unit 9 performs image processing on the target image data based on the reference image data stored in the reference image database 8 and the target image data stored in the image data memory 7 to obtain the adjusted image data. Generate (see FIG. 2). The image processing unit 9 is realized by software processing in cooperation with the image processing program stored in the storage unit 10 and the control unit 2.

  The storage unit 10 stores various control programs executed by the control unit 2, data necessary for executing the control program, image processing conditions, and the like.

Next, the operation of the present embodiment will be described.
FIG. 2 is a flowchart showing the image adjustment process.

  First, target image data is acquired by the image input unit 5 (step S1). The target image data acquired by the image input unit 5 is stored in the image data memory 7.

  Next, the reference image data selection setting screen is displayed on the display unit 4, and the reference image data stored in the reference image database 8 is selected by the user instructing the selection of the identification photo type from the operation unit 3 ( Step S2). The target image data may be automatically set from a plurality of reference image data by body type by analyzing the target image data. The selected reference image data is enlarged or reduced according to the target image size (step S3).

  Next, feature points are extracted from the reference image data (step S4). A feature point is a point indicating a feature in each image. In the case of a human face image, each point indicating the center of the black eye or the position of the edge of the eye, each point indicating the contour of the mouth or the face, the head Examples include points indicating boundaries between the part and shoulder and the background. FIG. 3A shows a reference image and feature points extracted from the reference image. The feature points of the reference image data may be extracted in advance and stored in the reference image database 8 together with the reference image data.

  Then, feature points corresponding to the feature points of the reference image data are extracted from the target image data (step S5). FIG. 3C shows the feature points extracted from the target image shown in FIG.

  Next, a conversion condition selection setting screen is displayed on the display unit 4, and the conversion condition for the target image data in the image processing is selected by an instruction from the operation unit 3 by the user (step S6). The target point that is the target of each feature point in the conversion of the target image data is an internal dividing point between each feature point of the reference image and each feature point of the target image, or each feature point of the reference image and the target image The feature point may be an external division point, or the user may directly specify the internal division ratio or the external division ratio, and the degree of standardization of the target image or the deviation from the reference image is emphasized. It is also possible to set a plurality of degrees in a stepwise manner and select from among them. Then, a target point is determined based on the selected conversion condition (step S7).

Here, as shown in FIG. 4A, when each feature point A _i of the target image is matched with each feature point B _i of the reference image, each feature point B _i of the reference image is set as a target point. . Thereby, each feature point A _i of the target image can be corrected to the position of each feature point B _i of the reference image, and the image can be standardized.

Further, when each feature point A _i of the target image is brought close to each feature point of the reference image, as shown in FIG. 4B, each feature point B _i of the reference image and each of the target images corresponding to the feature point B _i respectively. A point that internally divides the feature point A _i by a predetermined ratio h: k (h and k are positive numbers) is defined as a target point C _i . Thereby, it is possible to standardize the image while leaving the feature of the target image.

When emphasizing the deviation of each feature point A _i of the target image from the reference image, as shown in FIG. 4C, each feature point B _i of the reference image and each of the target images corresponding to the feature point B _i respectively. A point that divides the feature point A _i into a predetermined ratio m: n (m, n is a positive number satisfying m> n) is defined as a target point C _i .

  Note that the conversion condition for the target image data may be selectable for each region of the target image. For example, different processing can be performed for each region of the target image, such as emphasizing only the eyes or matching only the mouth with the reference image.

  Next, Delaunay triangulation is performed with the target point as a vertex (step S8), and the target image is also divided into regions corresponding to the triangle as shown in FIG. 3D (step S9). Delaunay triangulation refers to a method of dividing a region so as to maximize the sum of the minimum angles of the triangles among the methods of dividing a region by triangles having vertices at respective points.

Then, for the target image data, an affine transformation is performed for each of the divided triangle areas so that the triangle having the vertex of each feature point of the target image matches the triangle having the corresponding target point as the vertex. Is performed (step S10), and adjusted image data is generated as shown in FIG. The affine transformation is a geometric transformation based on a combination of linear transformation and parallel movement.
This completes the image adjustment process.

  As described above, according to the image processing apparatus 1, the feature points are extracted from the reference image prepared in advance, the feature points corresponding to the feature points of the reference image are extracted from the target image to be processed, and the reference image A target point is determined based on each feature point of the target image and each feature point of the target image, and geometric conversion processing is performed on the target image so that each feature point of the target image matches the target point. can do.

  In addition, since the target image is divided into triangular regions having the feature points of the target image as vertices and affine transformation is performed for each of the divided regions, each transformation can be simplified.

  In addition, by making it possible to select a target point determination method, it is possible to standardize a target image or emphasize deviation from a reference image in accordance with a user's request.

  Further, since the reference image is enlarged or reduced according to the target size, it is not necessary to prepare a plurality of reference images in advance.

  The description in the above embodiment is an example of a suitable image processing method according to the present invention, and the present invention is not limited to this. The detailed configuration and detailed operation of each part constituting the image processing apparatus 1 can be changed as appropriate without departing from the spirit of the present invention.

It is a block diagram which shows the functional structure of the image processing apparatus 1 in embodiment of this invention. It is a flowchart which shows an image adjustment process. It is a figure for demonstrating the conversion procedure of a target image. It is a figure which shows the determination method of a target point.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Control part 3 Operation part 4 Display part 5 Image input part 6 Image output part 7 Image data memory 8 Reference | standard image database 9 Image processing part 10 Memory | storage part 11 Bus

Claims (14)

Extract feature points from a standard image prepared in advance,
Extracting feature points corresponding to the feature points of the reference image from the target image to be processed;
Determining a target point based on each feature point of the reference image and each feature point of the target image;
An image processing method, wherein a geometric transformation process is performed on the target image so that each feature point of the target image matches the target point.   The image processing method according to claim 1, wherein the target point is each feature point of the reference image.   The image processing method according to claim 1, wherein the target point is a point that internally divides each feature point of the reference image and each feature point of the target image at a predetermined ratio.   The target point is a point that divides each feature point of the reference image and each feature point of the target image into m: n (m, n is a positive number satisfying m> n). The image processing method according to claim 1.   The target point is a point that internally divides each feature point of the reference image and each feature point of the target image at a predetermined ratio, or each feature point of the reference image and each feature point of the target image. The image processing method according to claim 1, wherein m: n (m, n is a positive number satisfying m> n) can be selected from any one of the points.   An internal ratio that internally divides each feature point of the reference image and each feature point of the target image, or an external ratio that externally divides each feature point of the reference image and each feature point of the target image. 6. The image processing method according to claim 5, wherein the image processing method is selectable.   The image processing method according to claim 5 or 6, wherein a method for determining the target point can be selected for each region of the target image.   The image processing method according to claim 1, wherein the reference image is used after being enlarged or reduced according to a target size.   The geometric transformation process divides the target image into polygonal regions having respective feature points of the target image as vertices, and performs geometric transformation for each of the divided regions. The image processing method according to any one of the above.   The image processing method according to claim 9, wherein the polygon is a triangle, and the geometric transformation is an affine transformation.   The image processing method according to claim 10, wherein the triangle is a triangle corresponding to Delaunay triangulation with the target point as a vertex.   The target image is an ID photo image obtained by photographing with a predetermined posture, and the reference image is a human model image having a face position and a face size corresponding to a predetermined ID photo type The image processing method according to claim 1, wherein the image processing method is an image processing method.   13. The reference image according to claim 1, wherein the reference image is created based on a plurality of images taken in advance or average position information of feature points extracted from the plurality of images. An image processing method described in 1.   The image processing method according to claim 1, wherein a plurality of the reference images are prepared in advance and can be selected according to the target image.

Images