JP2018206437A - Image processing device and image processing program - Google Patents

Abstract

To provide an image processing device that divides an image so as to reduce the number of areas as much as possible while subdividing a subject area desired to be extracted.SOLUTION: An image processing device comprises: a storage unit that stores an image; an extraction unit that extracts subject information from the image; a parameter determination unit that determines a division parameter on the basis of the subject information extracted by the extraction unit; an image processing unit that performs division processing on the image by using the division parameter; and an output unit that outputs division processing information by the image processing unit.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an image processing apparatus and an image processing program.

  Conventionally, there is an imaging device that divides a captured image into a plurality of small regions, calculates a standard deviation and an average value of image signals of pixels for each small region, and adjusts white balance (see, for example, Patent Document 1). According to this imaging apparatus, the image is divided into a plurality of areas in advance, and the subject area is specified by analyzing the state of each divided area.

JP 2010-273319 A

  However, in the above-described imaging device, if the captured image is excessively subdivided, the subject area specifying efficiency decreases. On the other hand, if the subdivision is too rough, there is a problem that a necessary subject area is not subdivided. Therefore, it is desirable to reduce the number of areas as much as possible while subdividing the subject area to be extracted.

  An object of the present invention is to provide an image processing apparatus and an image processing program for dividing an image so as to reduce the number of regions as much as possible while subdividing a subject region to be extracted.

  An image processing apparatus according to the present invention includes a storage unit that stores an image, an extraction unit that extracts subject information from the image, and a parameter determination that determines a division parameter based on the subject information extracted by the extraction unit. An image processing unit that performs a division process on the image using the division parameter, and an output unit that outputs information on a division process performed by the image processing unit.

  The image processing program of the present invention includes an image processing apparatus including a storage unit that stores an image, a step of extracting subject information from the image, a step of determining a division parameter based on the subject information, It functions as a step of performing division processing on the image using the division parameter, and a step of outputting division processing information performed using the division parameter.

  According to the present invention, it is possible to divide an image so as to reduce the number of regions as much as possible while subdividing a subject region to be extracted.

1 is a block diagram showing a system configuration of an image processing apparatus according to an embodiment. It is a figure which shows the relationship between the face size and K value which concern on embodiment. It is a figure which shows the case where the subject within the imaging | photography angle of view which concerns on embodiment, the image is subdivided by K = 60, and the image is subdivided by K = 100. It is a figure which shows the case where the subject within the imaging | photography angle of view which concerns on embodiment, the image is subdivided by K = 60, and the image is subdivided by K = 100. It is a figure which shows the moving body in the screen of the image which concerns on embodiment. It is a figure which shows the moving body in the screen of the image which concerns on embodiment.

  Hereinafter, an image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings, taking a camera as an example. FIG. 1 is a block diagram illustrating a system configuration of a camera according to an embodiment. As shown in FIG. 1, the camera 2 includes a control unit 4 that comprehensively controls each unit of the camera 2. The control unit 4 includes an imaging unit 6 that captures an image of a subject, a memory 8 that stores image data generated from an imaging signal output from the imaging unit 6, and a program such as an imaging process and a division process that are executed by the control unit 4. A program memory 10 to be stored, a memory card 12 to store image data of a photographed image, an image arranged on the back surface of the camera 2, an image captured by the imaging unit 6, a screen for setting a mode of the camera 2, and the like are displayed. The LCD display unit 14 and the operation unit 16 are connected. Here, the operation unit 16 includes a power switch, a release button, and the like.

  Next, division processing by the camera according to the embodiment of the present invention will be described. When the operator turns on the power of the camera 2 by operating the power switch of the operation unit 16, the control unit 4 images the subject at a predetermined frame rate by the imaging unit 6 and outputs the image output from the imaging unit 6. A through image is displayed on the LCD display unit 14 based on the signal. Here, the image data of the through image output from the imaging unit 6 is temporarily stored in the memory 8.

  When the operator presses the release button halfway, the control unit 4 performs a division process on the image based on the image data temporarily stored in the memory 8 to subdivide the image. In subdivision of an image, whether or not adjacent pixel groups belong to the same object region is determined using the similarity of feature amounts. As a result, adjacent pixel groups having similar characteristics are grouped, and the image is subdivided into a plurality of regions. At that time, it is possible to adjust the size of the subdivision.

For example, when the k-means method is used as a subdividing method, the k-means method divides elements into a given number of clusters K, and at this time, the average in each cluster is minimized. Split into Specifically, first, the centers of K clusters are randomly assigned (step 1). Next, each element is assigned to the nearest center (cluster) (step 2). Then, the center of each cluster is recalculated (step 3). As a result, if the centers of all the clusters have not changed, the process ends. Otherwise, the process returns to step 2. When this method is used for image segmentation, the target element is a pixel. The distance from the cluster center uses the color difference from the center pixel and the screen distance. Here, the distance from the cluster center is expressed as follows.
Distance from cluster center = color difference from center pixel + screen distance from center pixel distance from cluster center = pΔuv + qΔxy
Distance from cluster center = √ [(ui−uj) ² + (vi−vj) ² ] + √ [(xi−xj) ² + (yi−yj) ² ]
The degree of subdivision (division parameter) can be adjusted by the value of cluster K. Here, if the value of K is small, the degree of subdivision is small (the area of the subdivided region is large), and if the value of K is large, the degree of subdivision is large (the area of the subdivided region). Is smaller).

The present invention changes the value of K based on the size of the subject. In this case, the subject is a human and the value of K is changed based on the detected face size. Here, the value of K is expressed as follows. Note that f is the face size.
K = s (f)
As shown in FIG. 2, s (f) is a function determined by the size of the face size, and takes a larger value as the face size increases. In this embodiment, the function is proportional to the side length f (face size) when the face size is defined by a square rectangle. In addition, an upper limit and a lower limit are set in a range to which K = s (f) is applied, and a range outside the range is set to a fixed value. The upper limit was set to a value that would prevent the area to be subdivided from becoming too large, and the lower limit was set to the limit of the size of the face that can be detected.

  The control unit 4 detects a human face in the image based on the image data temporarily stored in the memory 8 using the face recognition function, and based on the length of the square rectangle according to the detected face size. The face size f is detected. Then, the value of K is determined by K = s (f) = α × f, and image division processing based on image data stored in the memory 8 is performed using K.

  FIG. 3 shows an example of an image (a) when the face size is large, a case where the image is subdivided with K = 60 (b), and a case where the image is subdivided with K = 100 (c). In this case, a suitable value of K is 60. The subject portion is sufficiently divided at K = 60, but is excessively divided with respect to the size of the subject at K = 100.

  FIG. 4 shows an example of an image (a) when the face size is small, a case where the image is subdivided with K = 60 (b), and a case where the image is subdivided with K = 100 (c). In this case, a suitable value for K is 100. When K = 60, the subject portion is not divided. When K = 100, the subject portion is sufficiently divided.

  The control unit 4 outputs the image data of the image subdivided by the division processing to the memory 8 and temporarily stores it, and specifies the subject area and the background area based on the stored image data of the subdivided image To do. The identified subject area is used for processing such as AF. That is, the control unit 4 performs a focusing operation on the specified subject area by an AF driving unit (not shown).

  According to the camera of the embodiment of the present invention, since the degree of subdivision of the image is determined based on the size of the subject, the number of regions is reduced as much as possible while subdividing the subject region to be extracted. Division can be performed.

  In addition, a paper (Title: Efficient Graph-Based Image Segmentation, Author: Pedro F. Felzenszwalb and Daniel P. Huttenlocher, Source: International Journal of Computer Vision, Volume 59, Number 2, September 2004, reference URL: http://people.cs.uchicago.edu/~pff/papers/seg-ijcv.pdf) is known.

The subdivision method described in the above paper is a method for sequentially determining whether or not adjacent pixel groups (regions) should be integrated. The degree of division (division parameter) is determined by a coefficient τ (C) defined by the following equation. Here, the coefficient τ (C) is easily integrated as the value increases. Note that C is a region area and k is a constant.
τ (C) = k / C
The coefficient τ (C) is the reciprocal of the area C, and the smaller the area, the more easily integrated. When τ (C) is changed based on the size of the subject, even in this case, if the subject is a person, τ (C) is expressed based on the size of the detected face as shown in the following equation. ) Can be changed to obtain the same effect as the k-means method.
τ (C) = s (f) × k / C
In the above-described embodiment, a moving part (moving object) in the screen may be regarded as a subject, and τ (C) may be changed based on the size of the moving object as shown in the following equation. . In addition, m is a moving body size.
τ (C) = s (m) × k / C
s (m) is a function determined by the size of the moving object, and takes a larger value as the moving object size increases. In this case, the function is proportional to the moving object size m, and the moving object size m is the area of a rectangular frame that includes the portion considered as a moving object. Here, FIG. 5 shows a case where the size of the moving person is small, and FIG. 6 shows a case where the size of the moving person is large.

  Any method may be used to detect the moving body portion. For example, in a plurality of images acquired at different times, a method in which a non-overlapping portion is regarded as a moving portion or a method of calculating a motion vector of sampling points in the image may be used.

  In the above-described embodiment, the dividing process is performed using the camera 2. However, the above-described dividing process may be performed using an image processing apparatus such as a personal computer.

  Alternatively, a program for causing the image processing apparatus such as a personal computer to perform the above-described image data division processing may be read to execute the image division processing based on the image data. That is, the image division processing program used in the above-described embodiment is a step of extracting information on a subject from an image, a step of determining a division parameter based on the information of the subject, and a division parameter for the image. A step of performing division processing using and a step of outputting division processing information performed using the division parameters are executed. Therefore, according to the image data division processing program according to the above-described embodiment, in order to determine the degree of subdivision of the image based on the size of the subject of the image based on the image data stored in advance in the memory, The image can be divided so that the number of regions is reduced as much as possible while subdividing the subject region to be extracted.

  In this case, the control unit of the image processing device performs division processing on the image based on the image data stored in the memory to subdivide the image, and performs white balance adjustment on the image data based on the specified subject. Also good.

  In the above-described embodiment, the result of the division process may be displayed on the display unit 14. For example, the subject area and background area specified in the subdivided image may be displayed on the LCD display unit 14 using different colors.

  DESCRIPTION OF SYMBOLS 2 ... Camera, 4 ... Control part, 6 ... Imaging part, 8 ... Memory, 10 ... Program memory, 14 ... LCD display part, 16 ... Operation part.

Claims (6)

A storage unit for storing images;
An extraction unit for extracting subject information from the image;
A parameter determination unit for determining a division parameter based on the subject information extracted by the extraction unit;
An image processing unit that performs a division process on the image using the division parameter;
An image processing apparatus comprising: an output unit that outputs division processing information by the image processing unit.   The image processing apparatus according to claim 1, wherein the subject information is information on a face size of a main subject.   The image processing apparatus according to claim 1, wherein the subject information is information on a size of a moving object region. An image processing apparatus having a storage unit for storing an image,
Extracting subject information from the image;
Determining division parameters based on the subject information;
Performing a division process on the image using the division parameter;
Outputting division processing information performed using the division parameter;
Image processing program to function as   5. The image processing program according to claim 4, wherein the subject information is information on a face size of a main subject.   The image processing program according to claim 4, wherein the subject information is information on a size of a moving object region.

Images