JPH04241844A - Image interpolating method - Google Patents

Abstract

PURPOSE:To reduce the unnaturalness generated in an interpolation image at the time of generating the interpolation image to be positioned spatially or hourly between original images from two pieces of original images. CONSTITUTION:On an original image (22), a partial area (5) corresponding to a targeted small area (4) is set. Subsequently, a small part (6nn) on an original image (32) opposed to a small part (511) in the partial area 5 by holding the notice small area (4) therebetween is determined, and with regard to images of those small parts (511), (6nn), a distance d11 is calculated. With regard to all the small parts (511)-(5nn) in the partial area (5), distance d11-dnn are calculated in the same way as the calculating method, and the shortest one of those distances d11-dnn is selected. Next, by an interpolating operation using the small part image which becomes the root of the shortest distance, the image of the targeted area (4) is derived. By setting the targeted small area (4) extending over the whole frame (12), and applying the interpolating method to each of them, an interpolation image N is derived.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] This invention is used for interpolation of CT images, playback of moving images, etc.
The present invention relates to an image interpolation method that creates an interpolated image that should be located between these images spatially or temporally.

0002

2. Description of the Related Art FIG. 6 is an explanatory diagram showing a conventional image interpolation method. In the figure, 10 is a range in which an interpolated image N is created by interpolation, and interpolated frames 20 and 30 are the sources of interpolation, respectively. In the original image, 40 is an element on the interpolation frame 10, and 50 and 60 are elements on the original images 20 and 30, respectively.

Next, the operation will be explained. Here, an image (image element) of element 40 in the interpolation frame 10 is assumed to be nij, and images of elements 50 and 60 on the original images 20 and 30 are respectively aij and bij. For example, the image element nij in the interpolated frame image 10 is obtained by linear interpolation shown in Equation 1.

0004

[Math 1]

[0005] Then, by performing the calculation according to formula 1 on all elements constituting the interpolated frame 10,
An entire interpolated image is created. Therefore, original images 20, 30
Let A and B represent the interpolated image N, and the interpolated image N can be expressed by Equation 2.

[0006]

[Math 2]

For example, when a human body tomographic image by CT scan is obtained as two original images 21 and 31 as shown in FIG. , an interpolated image 11 is obtained.

[0008]

[Problem to be Solved by the Invention] Since the conventional image interpolation method is performed as described above, two original images 21 and 31
If there is a shift in the position of the pattern inside, interpolated image 1
In No. 1, there was a problem in that the pattern was doubled, resulting in an unnatural pseudo-image.

[0009] This invention was made to solve the above-mentioned problems, and aims to provide an image interpolation method that can prevent patterns from becoming duplicated and create an interpolated image that is close to a natural image. shall be.

0010

[Means for Solving the Problems] The image interpolation method according to the present invention provides a method for interpolating small partial images in two original images, which are opposite to each other and centering on a small portion of interest in an interpolation frame. The distance is calculated for each small partial image in the partial area of the two original images, the combination of small partial images corresponding to the smallest distance among the calculated distances is selected, and the combination of the selected small partial images is Perform interpolation calculations such as linear interpolation based on the combination, set the result of the calculation as the image of the small part of interest, set the small part of interest over the entire interpolation frame, and repeat the process of calculating the image of the small part of interest as described above. , an interpolated image within an interpolated frame is created.

[0011]

[Operation] In this invention, the selection of a small partial image from a partial area for the interpolation calculation of the image of the small portion of interest is performed by calculating the distance between the small partial images facing each other with the small portion of interest as the center. Since it is executed to find a combination of sub-images with the minimum , the sub-image that is most similar between the two original images is selected, reducing the unnaturalness of the interpolated image .

0012

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, 12 is an interpolation frame that is the range in which the interpolated image N is generated, and 22 and 32 are the original images (
Hereinafter, they will be referred to as original image A and original image B. ), 4 is a small region of interest in which an image is created by interpolation, and 5 and 6 are partial regions on original images A and B. In addition, 511 to 5nn are partial areas 5
7 is the distance d1 for each small part image.
It is a set of 1 to dnn. Note that the partial regions 5 and 6 have, for example, the same center coordinates as the coordinates of the small region of interest 4 in the interpolated frame 12, and are sized according to the maximum predicted pattern shift amount in the two original images A and B. (When the amount of deviation is large, the size is large; when the amount of deviation is small, the size is small). Further, FIG. 2 is a flowchart showing processing when this image interpolation method is executed by a computer or the like.

Next, the operation will be explained with reference to the flowchart shown in FIG. First, a small region of interest 4 and a partial region 5 for it are determined (steps ST11 and ST12).
). Then, the partial area 5 is divided into small parts 511 to 5nn (step ST12). Furthermore, two original images A
, B (if you want to obtain an interpolated image N that should be exactly in the middle of the two original images A and B spatially, then A small portion 5uv (a position equidistant from the two original images A and B), a small portion 5uv (u
= 1 to n, v = 1 to n)
The upper small portion 6xy (x=1 to n, y=1 to n) is determined. Here, if the contents of one original image A are the contents of the other original image B translated in parallel, the coordinates of the small portion 5uv are (u, v), and the coordinates of the small region of interest 4 are (i
, J), the coordinates (x, y) of the small portion 6xy are (
zi-u, zj-v).

Next, the distance between the small portion images auv and bxy is determined from the small portion image auv of the small portion 5uv and the small portion image bxy of the small portion 6xy using Equation 3. Formula 3
Specifically, auv and bxy are concentrations.

[0015]

[Math 3]

[0016] In Equation 3, the two integral symbols represent, for example, adding the squared difference in density of each image constituting the small partial images auv, bxy for all pixels constituting the small partial images auv, bxy. means. distance d
When calculating uv, instead of calculating the sum of squared differences as shown in Equation 3, the sum of absolute differences may be calculated (∫
∫｜auv-bxy｜).

In the example shown in FIG. 1, for the partial area K, distances d11 to dnn as shown in FIG. 1(C) are obtained corresponding to the small partial images a11 to ann shown in FIG. 1(B). . Therefore, as expressed by formula 4, the distance d11
Select the smallest one among ~dnn.

[0018]

[Math 4]

[0019] If the small partial images auv and bxy, which are the basis of the minimum distance dm, are expressed as am and bm, then
The small image of the small region of interest 4 is expressed by Equation 5. In this way, the most approximate partial images am, b
A small partial image n can be obtained from m.

[0020]

[Math 5]

In the process according to the flowchart of FIG. 2, in step ST13, a11 to bnn are set as initial values for am and bm, and initial values are set for parameters u and v for loop processing. And original image A
A small partial image bxy in the original image B corresponding to the small partial image auv in is determined in step ST14. Then, in step ST15, the calculation of Equation 3 is executed.

When the obtained distance duv is smaller than the already set value dm, the value dm is updated with the distance duv, and further am and bm are updated (
Steps ST16, ST17). Then, the distance duv is calculated for all the small partial images auv in the partial area K (steps ST18 to ST21). Finally, based on the remaining values dm, am, bm, an image of the small region of interest 4 is calculated in step ST22.

By setting this small region of interest 4 over the entire interpolation frame 12 and repeating the process of calculating the image of the small region of interest described above (step ST23), an interpolated image N can be obtained. For example, as shown in Figure 3, C
An interpolated image N can be created in an interpolated frame 13 set between original images 21 and 31 obtained by T-scanning.

The interpolated image N shown in FIG. 3 is created from two original images at spatially different positions, such as a CT image, and has a high correlation, but it is not suitable for animation images or video transmission images. It is also possible to obtain an interpolated image from original images at temporally different positions. In the example shown in FIG. 4, the original image 24 at time ti-1 and the time t
It shows how an interpolated image N at time ti is obtained from the original image 34 at i+1. The process of obtaining the interpolated image N is the same as that of the method described above.

Note that the distance duv (u=
1 to n, v=1 to n) are all greater than a predetermined value, that is, if there is no combination of small partial images with high intercorrelations, the two small parts obtained for the partial area K1 adjacent to the partial area K. The relationship between partial images am1 and bm1 may be applied. In other words, the small partial image am1 corresponding to the shortest distance dm1 found for the adjacent partial region K1
, bm1, and the small partial images auv and bxy with respect to the partial area K.
A small partial image of interest n is calculated as m and bm. For example, in the adjacent sub-basin K1, am1=a11,
When bm1=bnn, in the partial area K, am=a11 and bm=bnn are also set. However, small partial image a11 in area K1
, bnn and the small partial images a11, bnn in the area K have the same positional relationship, and the original images A, B
They do not refer to the same thing within.

Furthermore, instead of the interpolation formula expressed by Equation 5, n=am may be used. In this case, original image A
The pair with the highest correlation is selected from the small partial image auv in the original image B and the small partial image bxy in the original image B, and one of the pairs is set as the image of the small region of interest 4. Significant values can be stored.

Note that if the relationship between the original images A and B is not a relationship of parallel movement but a relationship that involves image transformation f such as rotation, enlargement, reduction, etc., the original image B
The upper small portion (6xy) may be set.

[0028]

[Math 6]

Furthermore, in the above embodiment, a case has been described in which an interpolated image of a two-dimensional image is created, but the present invention can also be applied to one-dimensional signal processing.

[0030]

As described above, according to the present invention, the image interpolation method is applied between small partial images on two original images that are opposite to each other with the small area of interest of the interpolation frame as the center. The interpolated image of the small region of interest was obtained using the small partial image with the shortest distance between them, and this small region of interest was set over the entire interpolation frame. It is possible to use a small partial image with the highest correlation, and there is an effect that an interpolated image close to a natural image can be created. In other words, when applied to CT images, the number of sections that actually need to be imaged by CT can be reduced, which reduces the burden on the subject, and when applied to animation images and image transmission, it is possible to It becomes possible to eliminate the number of frames that need to be created and the number of images that need to be transmitted.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram for explaining processing of an image interpolation method according to an embodiment of the present invention.

FIG. 2 is a flowchart showing processing of an image interpolation method according to an embodiment of the present invention.

FIG. 3 is an explanatory diagram showing an example of an original image and an interpolated image obtained by CT scanning.

FIG. 4 is an explanatory diagram showing an example of an original image and an interpolated image of an animation image.

FIG. 5 is an explanatory diagram for explaining processing of a conventional image interpolation method.

FIG. 6 is an explanatory diagram showing an example of an original image obtained by a CT scan and an interpolated image obtained by a conventional image interpolation method.

[Explanation of symbols]

4. Small areas of interest 5 Partial area 12 Interpolation frame 22 Original image 32 Original image In addition, in the figures, the same reference numerals indicate the same or equivalent parts.

Claims (1)

[Claims] 1. In an image interpolation method that creates an interpolated image from two original images in an interpolated frame to be located between these original images, a small area in the two original images of the same size as a small area of interest in the interpolated frame is provided. The distance between the small part images which are images of the parts and are opposed to each other with the small part of interest as the center is set by the second part larger than the size of the small part.
calculate each small partial image existing in a partial area in the two original images, select a combination of the small partial images corresponding to the smallest distance among the calculated distances, and select the combination of the selected small partial images. The result of an interpolation calculation such as linear interpolation targeting is set as an image of the small portion of interest, and the small portion of interest is set over the entire interpolation frame,
An image interpolation method characterized by creating an interpolated image of the entire interpolated frame.