JPH11134494A - Method for improving quality of image, its method and recording medium stored with its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a means for improving quality of an image with high performance without being affected by blur, etc., of an original image. SOLUTION: A pre-processing to improve image quality is first performed for respective inputted images by using only image information of an individual image resolution improving means 1 and the preprocessed images are stored in a material image buffer 2 by the image resolution improving means 1. One representative image is selected from among a set of plural images after the pre-processing, the selected image is divided into small areas and stored in a representative image buffer 4 by an image dividing means 3. All the images after the pre-processing except the representative image are defined as objects, one most similar small area in to the object images is extracted and a set of similar areas is generated by a similar small area detection means 5. The small area with the best image quality is selected from the set of the similar small areas by a small area selecting means 6. An image with high image quality is generated by sticking all the selected small areas together and the generated image is stored in a synthetic image buffer 8 by an image sticking means 7.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inexpensive image input apparatus capable of obtaining a high-quality image by improving the image quality by increasing the resolution of the image without changing the hardware of input means such as a camera. The present invention relates to a technique for realizing the means. Although there are various standards for image quality, the present invention targets the degree of blur and the level of resolution.

[0002]

2. Description of the Related Art Conventionally, there are the following two types of techniques for improving image quality.

The first type is a method using information of only a single image. This includes technology that removes blur using the inverse function of the degradation function, technology that smoothly complements the pixels that make up the original image, and technology that increases the resolution by restoring the high-frequency components of the image. There is. A commentary on the former can be found, for example, in Ozaki, Taniguchi, and Ogawa, "Image Processing-From Basics to Applications" (Kyoritsu Shuppan). No. 1, 1996).

[0004] The second type is a method for realizing high resolution of an image by integrating a plurality of very similar images such as temporally close frames in a moving image. Specifically, by performing alignment in units smaller than the pixel size between images, dividing all the pixels for each image, and combining the obtained pixel fragments to synthesize a new pixel. Increase the resolution.
This method has been proposed in the document "Komatsu, Yokoi, Ito," High Resolution Video Images, "Proceedings of the IEICE General Conference, 1997, D-11-64."

[0005]

Among the above conventional methods, the first type of method has a drawback that only information contained in a single image can be used. Techniques for removing blur using an image filter can provide high-performance restoration when the image is degraded by a single point spread function. Yes, conventional methods cannot always perform high-performance restoration. In addition, this technique for increasing the resolution is ineffective when a part of the original image is deteriorated so as to be unrecoverable.

[0006] The second type of method has a drawback that it is susceptible to blurring of the original image. In the conventional method, blurring of an image to be integrated is not taken into consideration, so that blurred images are also integrated together, which adversely affects the image quality of a created image.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and has as its object to solve the above-mentioned problems in the prior art and to improve the quality of an image having high performance which is not affected by blurring of an original image. It is to provide a means.

[0008]

SUMMARY OF THE INVENTION The above-mentioned object of the present invention is as follows.
The problem is solved by the following means (1) to (6).

(1) A method for improving the quality of an image by inputting a plurality of very similar images and creating a high-quality image by integrating the plurality of images. Performing pre-processing to improve image quality using only its own image information without using information of other images, selecting one representative image from a set of a plurality of images after the pre-processing, and selecting the representative image Is divided into small areas, and for each of the divided small areas, one of the most similar small areas in the target image is set as one for all the pre-processed images other than the representative image. Create a set of similar small areas by performing the extraction process,
Performing a process of selecting a small area having the highest image quality from the set of similar small areas, and creating a high-quality image by laminating all the selected small areas. Quality method.

(2) In the method for improving the quality of an image according to the above (1), the preprocessing of each of the input images is performed as a part of the processing by increasing the resolution to increase the number of pixels. To improve the quality of the image.

(3) An image quality improving apparatus for inputting a plurality of images that are very similar and integrating the plurality of images to create a high-quality image. A preprocessing means for improving image quality using only its own image information without using information of other images, and selecting one representative image from a set of a plurality of images after the preprocessing, An image dividing means for dividing an image into small areas, and for each small area of the divided representative image, extracting a corresponding small area of all input images other than the representative image to form a set of small areas. A small area selecting means for evaluating the image quality of each small area included in the set of small areas, and selecting a small area having the highest image quality; Image stitching means to stitch all small areas High quality device for an image, characterized by comprising.

(4) In the image quality improving apparatus of the above (3), the pre-processing means performs a high-resolution process to increase the number of pixels as a part of the process. High quality equipment.

(5) A computer-executable program for inputting a plurality of images that are very similar and integrating the plurality of images to create a high-quality image, the computer-executing method comprising: For each input image, a pre-processing procedure for improving image quality using only its own image information without using information of other images, and a representative image from a set of a plurality of images after the pre-processing Is selected, and the representative image is divided into small areas. For each of the divided small areas, all the pre-processed images other than the representative image are subjected to the target image processing. The procedure of extracting one of the most similar small areas in the set to create a set of similar small areas, and evaluating the image quality of each small area included in the set of small areas, Select a small area with the best image quality from the set of areas And procedures, and procedures to create high-quality images by bonding the chosen all small regions,
A recording medium on which a method for improving image quality is recorded, wherein the program having the program is recorded on a computer-readable medium.

(6) In the recording medium on which the image quality improving method of (5) is recorded, in the preprocessing procedure for each of the inputted images, the resolution is increased by increasing the number of pixels as a part of the processing. A recording medium on which a method for improving the quality of an image is recorded.

In the method and apparatus for improving the quality of an image according to the present invention, when integrating a plurality of images, only a good area of each image is selected and integrated, whereby an image which is resistant to blurring of the original image is obtained. Realize high quality means, thereby expanding the scope of technology.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a processing block diagram according to an embodiment of the present invention. In FIG. 1, reference numeral 1 denotes preprocessing means for performing preprocessing on each input image to improve image quality using only its own image information without using information of other images; Image buffer for storing a plurality of images (hereinafter referred to as material images) resulting from the processing, and one of the representative images is selected from a set of material images stored in the material image buffer 2 as the representative image. Dividing means into small areas, 4 a representative image buffer for storing a representative image, 5 each material area stored in the material image buffer 2 using the small area as a search key for each small area of the representative image. The similar small area extracting means for extracting the most similar small area from the small area and creating a set of small areas, 6 evaluates the image quality of each small area constituting the small area set, and Small area selection hand to select area 7 is an image laminated unit stuck together all the small regions selected by the small area selecting means 6.

Hereinafter, a detailed operation example of the block in FIG. 1 will be described.

FIGS. 2A to 2C show examples of a plurality of original images to be integrated according to this embodiment.

First, the preprocessing means 1 improves the image quality of each image by using the image information of each image itself. In the present embodiment, the content of the processing is only to increase the resolution. Then, the result is stored in the material image buffer 2. Examples of the created material images are shown in FIGS.

The following is an example of a specific method for increasing the resolution.

FIG. 4A is an example for explaining a part of an image to be subjected to the above-mentioned high resolution processing. In this figure, one square represents one pixel,
The brightness of the square represents the brightness of the pixel. This resolution is increased N times vertically and M times horizontally.

First, the size of each pixel is made 1 / N times in the vertical direction and 1 / M times in the horizontal direction without changing the center position. An example of the result is shown in FIG. In this figure, N
= M = 3.

Next, the space between the reduced original pixels (referred to as original pixels) is filled with pixels whose brightness values have been emptied. This pixel is called a complementary pixel. FIG. 4C shows the processing result.

Next, the brightness of each complementary pixel is calculated from the brightness of the surrounding original pixels. There are various methods for this calculation.
Yagi, Murakami, “A Study on Complementary Filters and Image Quality for Geometric Transformation of Image Signals”, Transactions of Electronics and Communication Sciences (D), J69-D, 11, pp. 146-64. 1617-1623
(1986) ".
This is to calculate the brightness of the complementary pixel by linearly interpolating the brightness of the four original pixels near the complementary pixel.
As an example of the calculation, a calculation in a case where the brightness of the point Q indicated by “x” in FIG. The four neighboring original pixels are denoted by P ₁₁ , P ₂₁ , P ₁₂ , and P ₂₂ . Further, the brightness of an arbitrary pixel R is set to f in common with the original pixel and the complementary pixel.
Expressed by (R). In FIG. 4D, the numbers in parentheses above the four original pixels near the point Q are the brightness of each original pixel, and f (P ₁₁ ) = 10 and f (P ₂₁ ) = 8.
0, f (P ₁₂ ) = 80, and f (P ₂₂ ) = 10.

Now, the calculation of the four-point linear interpolation will be described with reference to FIG. FIG. 5 illustrates the center coordinates of Q, P ₁₁ , P ₂₁ , P ₁₂ , and P ₂₂ using a downward vertical axis Y and a right horizontal axis X. The distance between the original pixels in the vertical and horizontal directions is set to 1 in common. Also, the X coordinate of the point Q,
Let the Y coordinate be x, y. In FIG. 5, x = 1/3 and y = 2
/ 3. At this time, f (Q) is obtained by the following equation.

F (Q) = (1−x) (1−y) f (P ₁₁ ) + x · (1−y) f (P ₂₁ ) + (1−x) · y · f (P ₁₂ ) + x Y · f (P ₂₂ ) = (2/3) · (1/3) · 10 + (1/3) · (1/3) · 80 + (2/3) · (2/3) · 80 + (1 / 3) · (2/3) · 10 = 49, and 49 is calculated as lightness. By performing such brightness calculation processing on all the complementary pixels, a material image can be created.

Next, the image dividing means 3 selects one of the material images stored in the material image buffer 2 as a representative image and stores it in the representative image buffer 4. Then, the representative image is deleted from the material image buffer 2. In this example, it is assumed that FIG. 3A is selected. Then, the image dividing means 3 divides the representative image into small areas, and stores the result in the representative image buffer 4. The result of this division is shown in FIG. In the example of FIG. 6A, the image is divided into three vertically and three horizontally, and is divided into nine small areas in total. Also, the small areas in FIG. 6A are given names a1 to a9, respectively, and are shown in FIG. 7A.

Next, the image dividing means 3 performs the same processing as the division into small areas performed on the representative image for each material image stored in the material image buffer 2, and stores the result in the material image buffer 2. Store. The results of this division are shown in FIGS. 6 (b) and 6 (c).
B9 and c1 to c9, which are shown in FIGS. 7B and 7C.

Next, the process proceeds to the similar small area extracting means 5.
The similar small area extracting unit 5 extracts small areas one by one from the divided representative images stored in the representative image buffer 4, performs processing, and sends them to the small area selecting unit 6. For the sake of explanation, a small area a1 of the representative image shown in FIG.
The processing when the is extracted will be described.

The similar small area extracting means 5 extracts a small area corresponding to the small area of the representative image from each material image stored in the material image buffer 2. Specifically, based on corresponding portions (b1 and c1 in (b) and (c) of FIG. 6) in the already divided small regions of each material image,
The similarity between the small areas is determined by shifting a small number of pixels around the small area, and a process of extracting the closest small area as the corresponding small area is performed. These are the material images,
Although it should be very close to the small area corresponding to the small area of the representative image, since the material image and the representative image are very similar but do not completely match, the small image that is simply divided by the same method is used. Since the regions do not completely correspond to each other, these small regions b1 and c1 cannot be used as they are.

The similar small area extracting means 5 outputs the small area of the other material image corresponding to the small area of the representative image thus extracted to the small area selecting means 6 together with the small area of the representative image.

FIG. 8 schematically shows the operation of the small area selecting means 6. The small area selecting means 6 regards each small area received from the similar small area extracting means 5 as a candidate for a small area to be fitted into the small area a1 of the representative image. Then, the image quality of these candidates is evaluated, and the candidate determined to have the highest image quality is selected and output to the image combining means 7.

As a method for evaluating the image quality of a small area, for example, a density contrast can be used for a monochrome image. This can be represented by the variance of the density. The higher the value, the clearer the image, and the lower the value, the more blurred. In the case of FIG. 8, the small area a of the representative image which is the candidate 1
1 is selected.

The image combining means 7 fits the received small region into the corresponding small region of the representative image in the composite image buffer 8. For example, FIG. 9B shows the result of performing the processing by receiving the result of FIG. 8 and performing the processing. The composite image buffer 8 is empty in the initial state as shown in FIG.

The processing of this embodiment is completed by performing the above-described processing from the similar small area extracting means 5 to the image combining means 7 for all the small areas of the representative image.

The preprocessing for increasing the resolution of an individual image, the processing for dividing an image, the processing for detecting a similar small area, the processing for selecting a small area, and the processing for image stitching in the various processing means shown in the above embodiments. Can be appropriately executed by a computer, and a program for causing the computer to execute the procedures and the like is stored in a computer-readable recording medium such as a floppy disk or various memories. It can be recorded on a card, MO, CD-ROM or the like and distributed.

[0038]

As described above in detail, according to the present invention, when integrating a plurality of images, only the good image information of each image is selected and integrated, thereby reducing the image quality of the original image. This has a remarkable effect that a means for improving the quality of a strong image can be realized.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating processing blocks according to an embodiment of the present invention.

FIGS. 2A, 2B, and 2C are diagrams illustrating examples of a plurality of images for explaining processing contents in the embodiment.

FIGS. 3A, 3B, and 3C are diagrams of images obtained by increasing the resolution of the image shown in FIG. 2 using only its own image information.

FIGS. 4A, 4B, 4C, and 4D schematically show a process of processing for increasing the resolution of an image using only its own image information in the embodiment. FIG.

FIG. 5 is a diagram for explaining a process of calculating the brightness of a complementary pixel in a process of increasing the resolution of an image using only its own image information in the embodiment.

FIGS. 6A, 6B, and 6C are diagrams illustrating examples of a result obtained by dividing the image illustrated in FIG. 3 into small regions.

FIGS. 7A, 7B, and 7C are diagrams showing examples of names of small areas of the image shown in FIG. 6;

FIG. 8 is a diagram schematically showing the operation contents of a small area selecting means in the embodiment.

FIGS. 9A and 9B are diagrams for explaining the operation of the composite image buffer in the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Pre-processing means 2 ... Material image buffer 3 ... Image dividing means 4 ... Representative image buffer 5 ... Similar small area extracting means 6 ... Small area selecting means 7 ... Image joining means 8 ... Synthetic image buffer

 ──────────────────────────────────────────────────の Continued on the front page (72) Inventor Kenji Ogura 3-19-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Nippon Telegraph and Telephone Corporation

Claims (6)

[Claims] An image quality improving method for inputting a plurality of very similar images and creating a high quality image by integrating the plurality of images, comprising: Performs preprocessing to improve image quality using only its own image information without using information of other images, selects one representative image from a set of a plurality of images after the preprocessing, and selects the representative image. For each of the divided small areas, one of the most similar small areas in the target image is extracted for all the pre-processed images other than the representative image. Create a set of similar small areas by performing a process to perform the process of selecting a small area with the best image quality from the set of similar small areas, and paste all the selected small areas together Create high-quality images with high quality images Method. 2. The method for improving image quality according to claim 1, wherein the preprocessing of each of the input images is performed as a part of the processing to increase the resolution to increase the number of pixels. . 3. An image quality improving apparatus for inputting a plurality of very similar images and integrating the plurality of images to create a high-quality image, comprising: Preprocessing means for improving the image quality using only the image information of the image itself without using information of other images; and selecting one representative image from a set of a plurality of images after the preprocessing, and selecting the representative image Image dividing means for dividing the image into small areas, and for each small area of the divided representative image, extract a corresponding small area of all the input images other than the representative image to form a set of small areas. The similar small area extracting means to be created, the image quality is evaluated for each small area included in the set of small areas, and the small area selecting means to select the small area having the highest image quality; and the small area selecting means. Image combining means for combining all the small areas, An image quality improving device comprising: 4. The image quality improving apparatus according to claim 4, wherein said pre-processing means performs high resolution by increasing the number of pixels as a part of the processing. 5. A program for inputting a plurality of very similar images and executing a method for improving the quality of an image by creating a high-quality image by integrating the plurality of images, the program comprising: For each of the images, a pre-processing procedure for improving the image quality using only its own image information without using information of other images, and a representative image from a set of a plurality of images after the pre-processing. A procedure of selecting one image and dividing the representative image into small regions; and for each of the divided small regions, all of the pre-processed images other than the representative image are targeted. A process of extracting one of the most similar small regions to create a set of similar small regions, and evaluating the image quality of each of the small regions included in the set of small regions. A small area with the best image quality from a set of And a procedure for creating a high quality image by laminating all the selected small areas. A method for improving the quality of an image, comprising: Recording medium on which is recorded. 6. The image quality improvement according to claim 5, wherein in the pre-processing procedure for each of the input images, a high resolution is performed to increase the number of pixels as a part of the processing. A recording medium recording the method.