KR100208010B1 - Image coding method using spectral duplication - Google Patents

Abstract

The present invention relates to an image encoding method using spectral redundancy, and an object thereof is to provide an image encoding method capable of effectively providing spectral redundancy.

In order to achieve the object of the present invention, any one of the luminance image (Y), red image (R), green image (G), blue image (B) of the same image as a reference image (reference image) After generating the reference image data through encoding and decoding, decode the first reference decoded image to generate a first reference decoded image, select two images from the remaining images, set one of the two images as the first estimation target image, and generate another image. 2 is a block-based estimation and encoding of a proportional factor, which is stretching information, and an acceleration / decrease factor, which is slicing information, which is set as an estimated target image so that a minimum square error between the first reference decoded image and the estimated target image is minimized. According to the present invention, since spectral redundancy can be directly removed, the conventional method of indirectly removing spectral redundancy through low resolution analysis. Compared to the method it can talk to the relative improvement in the high coding efficiency.

Description

Image coding method using spectral redundancy.

The present invention relates to an image encoding method, wherein in the same image, stretching information in terms of chromaticity correlation with respect to a red image (R), a green image (G), and a blue image (B) on the basis of the luminance image (Y) The scaling factor, which is stretching information, and the offset factor, which is slicing information, are calculated on a block-based basis, and then encoded and provided to the decoding unit. The present invention relates to an image encoding method using spectral redundancy that can effectively remove spectral redundancy by reproducing color images.

In the 1990s, interest in multimedia began to increase rapidly. This is the result of the phenomenon that the boundary between the existing computer industry, the communication industry, and the broadcasting and film industries is broken down, and elements that were recognized as characteristics of each field are expanded and integrated into other fields.

In other words, multimedia computers supporting video, voice, communication, etc. are leading the personal computer market, and services of high interactivity video such as Video On Demand (VOD) are the communication industry and the film industry. Has become a major concern.

Increasing interest in multimedia is rapidly increasing the interest and need for a video encoding method that can obtain a good image quality even at a low transmission speed. As a result, the International Standardization Group Moving Picture Experts Group (MPEG) Committee has followed the Moving Picture Experts Group-1 (MPEG-1) and the Moving Picture Experts Group-2 (MPEG-2) video at rates from 8 kbps to about 64 kbps. The company is in the process of standardizing Moving Picture Experts Group-4 (MPEG-4) for transmitting audio.

MPEG-4 is a standard for communication, access, and manipulation of digital video and audio that is defined for ease of use in multimedia applications such as video on demand, next-generation mobile communications, and video telephony. Currently, there is a lot of progress toward standardization, and a content-based coding scheme is adopted instead of the simple block-oriented coding used in H.261, MPEG-1, and MPEG-2. Is likely to be.

In the case of conventional MPEG-1 and MPEG-2, Discrete Cosine Transform (DCT) coding and motion estimation / compensation (Motion) in block or macro block units using mathematical statistical characteristics of an image Estimation / Compensation) can be used to achieve a compression ratio of about 40: 1, and a maximum compression ratio of 240: 1 can be achieved in case of H.261, which suffers a slight deterioration in consideration of limited applications such as video telephony. It is possible, but MPEG-4 requires additional compression because it wants to obtain image quality similar to that of conventional H.261 at a bit rate from 8kbps to about 64kbps.

However, even if the conventional image standard techniques are optimized as much as possible, it is judged that no additional compression of about 2: 1 is possible. MPEG-4 overcomes this problem and at the same time, it is suitable for human vision, which is the subject of image quality evaluation. To reconstruct, image coding that can actively utilize Human Visual System (HVS), that is, object-based coding, segmentation-based coding, and model-based coding Second-generation coding techniques, such as -based coding and fractal coding, are likely to be adopted as core coding techniques.

In addition to the second generation coding techniques, first generation coding techniques such as vector quantization, subband coding, and Motion Compensated Discrete Cosine Transform (MC-DCT) coding It is also a complex study.

In view of the current trend for image coding, it is apparent that additional compression must be performed in order to meet recent demands even in performing color image coding.

Hereinafter, a description of the color system will be briefly described to help understand the present invention reflecting such technical trends.

Any color can quantitatively define the amount of the three primary colors satisfying the isochronous condition by mixing the three primary colors, which is called a color specification.

In 1931, the Commission International de l'Eclairage (CIE) decided to improve the discomfort of RGB color systems where the red (R) value was negative in the range of 440 to 545 nm. An XYZ colorimetric system was defined in which the spectral colors of 700 nm (red), 546.1 nm (green), and 435.8 nm (blue) were the three primary colors.

In other words, CIE's XYZ color system selects the new primary colors X, Y, and Z, eliminates quantification, and selects the three primary colors so that the triangles created by the new three primary colors contain all of the spectral color trajectories, which are unrealistic primary colors. .

Since each primary color of the XYZ color system is an imaginary one, select the following conditions to be satisfied.

All colors can be equalized by quantitative mixing of the primary colors, i.e. triangles by the X, Y, and Z primary colors will surround the spectrum and are located on a plane of RGB space with zero luminance, Select the Z primary color, that is, only the Y primary color has a stimulus with respect to luminance, whereas the X and Z primary colors have only hue and saturation.

At this time, the color conversion equation is the same as Equation 1,

은 RGB 신호로 휘도 신호(Y)로 합성할 시에 이용되는 가장 대표적인 변환 관계식이 되고 있다.here,

Is the most representative conversion relation used in synthesizing the luminance signal Y into an RGB signal.

Hereinafter, the image encoding method according to the prior art will be briefly described based on existing image standards.

As is well known, due to the human visual characteristics, the chromaticity signal is visually insensitive compared to the luminance signal (Y), and the fact that there is a high color intensity correlation between color images of the same image is known in the art. It is common knowledge.

To reflect this fact in color image coding, they are represented by MPEG-1, MPEG-2, Joint Picture Experts Group (JPEG), Federal Communication Committee / High Definition Television (FCC), H.261, H.263, etc. In the existing image coding standards, a color image is encoded using a luminance signal Y and chrominance signals Cb and Cr. In the image standards, the luminance signal Y and the chrominance signal are encoded. the relationship between (chrominance; Cb, Cr)

[Equation 2]

Equation 2

은 각각 청색 영상(B)과 휘도 영상(Y)간의 색차 신호이고,

은 적색 영상(R)과 휘도 영상(Y)간의 색차 신호를 나타내는 것이다.Here, R, G, and B are pixel values of the red image R, the green image G, and the blue image B, respectively, and Y is the luminance image Y synthesized by the above-described R, G, and B signals. Is the pixel value of

Are the color difference signals between the blue image B and the luminance image Y, respectively.

Denotes a color difference signal between the red image (R) and the luminance image (Y).

In a general color image, the chromaticity correlation between each color image is high enough to be easily recognized in the spatial domain through the naked eye as can be seen in FIGS. 1A to 1D without referring to the conversion relation of Equation 2.

In this case, FIG. 1A is a gray level representation of a luminance signal of a table tennis image (Tabletennis Image), which is a representative test image, and FIG. 1B is a gray level representation of a red signal of a table tennis image (Tabletennis Image). The image, FIG. 1C is an image representing the green signal of a table tennis image (Tabletennis Image) in gray level, FIG. 1D is an image representing the blue signal of a table tennis image (Tabletennis Image) in gray level.

의 비율이 4:2:0, 4:2:2, 4:4:4 등으로 대표되는 표본화 주파수비에 따라 색차 신호에 대한 저해상도적인 분석을 통해 부호화 대상이 되는 영상 데이터를 압축시키고 있다.In order to reflect this fact in video encoding, conventional video coding standards such as MPEG-1, MPEG-2, JPEG, FCC / HDTV, H.261, H.263, etc., subtract the luminance signal Y and subsampling. The chrominance (Cb, Cr) is encoded through the Discrete Cosine Transform (DCT) to remove the spectral redundancy inherent in the video to be encoded.

Based on the sampling frequency ratios represented by 4: 2: 0, 4: 2: 2, 4: 4: 4, etc., the image data to be encoded is compressed through low resolution analysis of the chrominance signal.

For a more detailed description of this, refer to ISO / IEC JTC1 / SC29 / WG11, 'Coding of Moving Pictures and Associated Audio', and further description thereof will be omitted.

However, the conventional video encoding method is not a method of directly removing spectral redundancy, but rather is a video encoding method for reducing data to be encoded through low resolution analysis based on subsampling of color difference signals. Since there is still a problem that many spectral redundancies remain in the encoded color image, the appearance of image encoding that can more effectively remove it and increase the encoding efficiency is expected.

Accordingly, the present invention has been made to solve such a problem, and the color image is encoded by detecting a block-based scale factor and an offset factor used in fractal coding to improve coding efficiency. That is, the proportional factor and the decrement factor for any two of the red image (R), the green image (G), and the blue image (B) are estimated from the luminance image (Y) of normal resolution, and the other image is It is an object of the present invention to provide a method of encoding an image using chromaticity correlation that can be efficiently calculated by calculating spectral redundancy by calculating a mathematical calculation through Equation 1, which is a colorimetric conversion equation.

1A is an image diagram representing a luminance signal of a table tennis image (Tabletennis Image) in gray level,

1B is an image diagram representing a red signal of a table tennis image (Tabletennis Image) in gray level,

1C is an image diagram representing a green signal of a table tennis image (Tabletennis Image) in gray level,

1D is an image diagram representing a blue signal of a table tennis image (Tabletennis Image) in gray level,

2 is a flowchart of a preferred embodiment of an image encoding method using spectral redundancy according to the present invention.

<Description of the symbols used in the main parts of the drawing>

S100: Color image estimation encoding step S110: Reference image encoding step

S111: Reference video setting step S112: Reference video code / decoding step

S120: First estimation video encoding step S121: First estimation target video setting step

S122: First factor estimation encoding step S130: Second estimation image encoding step

S131: second estimation target image setting step S132: second factor estimation encoding step

S200: Color image estimation decoding step S210: Reference image decoding step

S220: Estimated image decoding step S221: First estimated image decoding step

S222: Second estimated image decoding step S230: Third color image calculation step

In order to achieve the object of the present invention, color image coding using spectral redundancy according to the present invention,

In the image encoding method for improving the coding efficiency by removing the spectral redundancy inherent between color images of the same image:

After setting any one of the luminance image (Y), red image (R), green image (G), blue image (B) of the same image as a reference image and generates the reference image data through encoding, The first reference decoded image is decoded to select one of the remaining images, and one of the two images is set as the first estimation target image, and another image is set as the second estimation target image. A block-based estimation and encoding of a first proportional factor, which is stretching information, and a first acceleration / decrease factor, which is slicing information, to minimize the least square error between the decoded image and the first estimated target image, and the first criterion Block-based second scaling factor, which is stretch information, and second acceleration / decrease factor, which is slicing information, to minimize the least square error between the decoded image and the second estimated target image. A color image estimation encoding step of estimating and encoding the data;

Receives and decodes the reference image data, reproduces a second reference decoded image corresponding to the reference image, receives the first proportional factor and the first decrement factor, and receives a first estimation target image from the second reference decoded image. Estimating a first estimated decoded image having a spectral band, and receiving the second proportional factor and the second subtraction factor, and having the same spectral band as the first estimated target image in the second reference decoded image. And a color image estimation decoding step of estimating the second estimated decoded image.

Here, the reference image is preferably set to any one of the luminance image (Y) and the green image (G). In particular, the reference image includes a luminance image (Y), which includes most of the information about the image and is sensitive to human visibility. It is even more preferable to set as the reference image.

In this case, when the reference image is set as the luminance image Y, the green image G and the red image R having a relatively high chromaticity correlation with the luminance image Y may be compared with the first estimation target image and the second estimation image. If the reference image is set to the green image G, the first and second estimated target images are set as the luminance image Y and the red image R, or the red image R is set as the target image. ) And a blue image (B).

A preferred embodiment of an image encoding method using spectral redundancy according to the present invention will be described with reference to FIG. 2.

2 is a flowchart of a preferred embodiment of a video encoding method using spectral redundancy according to the present invention.

According to a preferred embodiment of the video encoding method using spectral redundancy according to the present invention, as shown in FIG. 2, in the video encoding method for improving coding efficiency by removing spectral redundancy inherent between color images of the same video,

The luminance image (Y) synthesized from the red image (R), the green image (G), and the blue image (B) of the same image is defined as the reference image, the reference image data is generated through encoding, and then decoded to generate the first reference decoded image. A minimum squared error between the first reference decoded image and the first estimated target image by setting the green image G as the first estimation target image and the red image R as the second estimation target image the first proportional factor, which is the stretching information, and the first acceleration / decrease factor, which is the slicing information, are encoded based on a block, and the least square error between the first reference decoded image and the second estimated object image is minimized. A color image estimation encoding step (S100) of estimating and encoding the second proportional factor, which is stretching information, and the second acceleration / decrease factor, which is slicing information, on a block basis, and the reference zero; Receives and decodes data to reproduce a second reference decoded image corresponding to the reference image, receives the first proportional factor and the first decrement factor, and specifies the second estimated decoded image as a first estimation target image. Estimating a first estimated decoded image having a parallel band, and receiving the second proportional factor and the second decrement factor, a second estimation having the same spectral band as the first estimated target image in the second reference decoded image. Color image estimation decoding step of estimating the decoded image (S200).

In the color image estimation encoding step S100, the luminance image Y synthesized from the red image R, the green image G, and the blue image B of the same image is determined as the reference image, and the reference image is encoded. After generating the data, the reference image encoding step 110 of decoding the first reference decoded image and setting the green image G as a first estimation target image to generate the first reference decoded image and the first estimation target image A first estimated image encoding step (S120) of estimating and encoding the first proportional factor, which is stretching information, and the first acceleration / decrease factor, which is slicing information, on a block basis to minimize the least square error between the blocks; The second proportional factor and the slice information, which is stretching information for setting (R) as the second estimation target image to minimize the least square error between the first reference decoded image and the second estimation target image. Information, the second consists of the acceleration factor as a second estimated image encoding step (S130) of encoding to estimate a block-based.

At this time, the reference image encoding step 110 is a reference image setting step of determining a luminance image (Y) synthesized from a red image (R), a green image (G), a blue image (B) of the same image as a reference image (S111). And a reference image code / decoding step (S112) of generating the first reference decoded image by decoding the reference image after generating the reference image data.

In addition, the first estimated image encoding step (S120) may include setting a first estimated target image (S121) of setting the green image (G) as a first estimated target image, and performing the first reference decoded image and the first estimation. A first factor estimation encoding step (S122) of estimating and encoding on a block basis a first proportional factor that is stretching information and a first acceleration / decrease factor that is slicing information so that a minimum square error between target images is minimized. The second estimated image encoding step (S130) may include setting a second estimation target image (S131) of setting the red image R as a second estimation target image, and performing the first reference decoded image and the second estimation. A second factor that estimates and encodes a second proportional factor, which is stretching information, to minimize the least square error between the target image, and a second additive factor, which is slicing information, on a block basis; Estimated encoding step (S132).

On the other hand, the color image estimation decoding step (S200) is a reference image decoding step (S210) for receiving and decoding the reference image data to reproduce a second reference decoded image corresponding to the reference image, the first proportional factor and The first estimated decoded image having the same spectral band as the first estimated target image from the second reference decoded image is received from the second reference decoded image, and the second proportional factor and the second added / decreased factor are input. An estimated image decoding step of receiving a second estimated decoded image having the same spectral band as the second estimated target image from the second reference decoded image (S220), and the first reference decoded image and the first estimated decoded image. The image and the first estimated decoded image are substituted into a colorimetric conversion equation, and the rest of the three color images of the same image except for the first and second estimated target images It consists of an image having a spectral band, such as the image in the third color image calculation step (S230) for calculating from the calculation.

In this case, the estimated image decoding step (S220) may be performed by receiving the first proportional factor and the first subtraction factor to decode the first estimated decoded image having the same spectral band as the first estimated target image from the second reference decoded image. A second estimated image decoding step (S221) for estimating and a second estimation having the same spectral band as the second estimated target image in the second reference decoded image by receiving the second proportional factor and the second added / subtracted factor; A second estimated image decoding step (S222) of estimating the decoded image is performed.

The procedure of the preferred embodiment of the video encoding method using spectral redundancy according to the present invention configured as described above will be described with reference to FIG. 2.

First, when a color image is input, in the reference image encoding step 110, the luminance image Y synthesized from the red image R, the green image G, and the blue image B of the same image is determined as the reference image. After encoding the reference image data through encoding, the first reference decoded image is decoded.

Subsequently, in the first estimation video encoding step S120, the green square G is set as a first estimation target image so that a least square error between the first reference decoded image and the first estimation target image is obtained. The first proportional factor, which is the stretching information to be the minimum, and the first acceleration / decrease factor, which is the slicing information, are estimated and encoded on a block basis, and in the second estimation image encoding step S130, the red image R is converted into the second estimation target image. Set to be a block-based estimation of the second proportional factor, which is stretching information, and the second acceleration / decrease factor, which is slicing information, so that the least square error between the first reference decoded image and the second estimation target image is minimized, and then decoded. Or transfer to a storage medium.

Looking at the process of the present invention in more detail through a mathematical expression to help understand the present invention, in the image encoding method using spectral redundancy according to the present invention, the red image (R), green image (G After calculating the luminance image (Y) image using Equation 1 or 2 using the blue image (B), instead of directly encoding the green image (G) and the red image (R), a proportional factor And extracting and subtracting factors,

의 두 화소 집합을 정의할 경우,

의 두 화소 집합을 최적으로 근사화시키기 위해, 수학식 4에서

값을 최소화시키는 비례 인자 '

'와 가감 인자 '

'를 구함으로써

의 화소 집합과 비례 인자 '

' 및 가감 인자 '

'만을 이용하여

의 두 화소 집합을 표현할 수 있다.As in Equation 3

If you define a two-pixel set of,

In order to optimally approximate two pixel sets of

Proportional factor to minimize the value '

'Addition and deduction factor'

By saving

Pixel set and proportional factor of '

'And subtraction factor'

Using only

Can represent two sets of pixels.

은 화소 집합의 크기를 나타내는 값이며,

는 화소 집합의 원소, 즉, 화소값을 나타내는 인덱스 변수이다. 이때, 상기와 같이 최소 자승 오차를 만족시킬 수 있는 비례 인자 '

'와 가감 인자 '

'는 비례 인자 '

'와 가감 인자 '

'을 변수로 하는 편미분을 통해 산출할 수 있는 데, 그 결과를 수학식 5에 제시한다.here,

Is a value that indicates the size of the pixel set,

Is an index variable representing an element of a pixel set, that is, a pixel value. At this time, the proportional factor that can satisfy the least square error as described above

'Addition and deduction factor'

'Is a proportional factor'

'Addition and deduction factor'

It can be calculated through the partial derivative with 'as a variable, and the result is shown in Equation 5.

'의 분모항이 수학식 6과 같이 영(zero)이되면,At this time, in Equation 5

If the denominator of 'becomes zero like Equation 6,

'가 영으로 나누어짐에 따라 연산 결과가 발산하게 되는데, 이와 같은 경우에는 수학식 7을 이용함으로써 이와 같은 문제를 해결할 수 있다.'

'Is divided by zero, and the calculation result is diverged. In such a case, this problem can be solved by using Equation 7.

'와 가감 인자 '

'는 각각

의 두 화소 집합을 최적으로 근사화되기 위한 스트레칭 정보와 슬라이싱 정보가 된다. 즉, 비례 인자 '

'와 가감 인자 '

'는 각각

의 두 화소 집합을 최적으로 근사화되기 위한 비례값(scale value)과 가감값(offset value)이 된다.As above, the proportional factor '

'Addition and deduction factor'

'Is each

Stretch information and slicing information for optimally approximating two sets of pixels. That is, the proportional factor '

'Addition and deduction factor'

'Is each

It becomes a scale value and an offset value for optimally approximating two sets of pixels.

The process of extracting the proportional factor and the subtractive factor to be applied to the present invention in which the proportional factor and the subtractive factor extraction process as described above are directly applied to the red image (R), the green image (G), and the blue image (B) is carried out using a mathematical expression. It will be briefly described.

First, for convenience of explanation, two images in which the degree of correlation between the luminance image Y and the red image R, the green image G, and the blue image B, which are set as the reference image, have priority. That is, a pixel set for the green image G and the red image R is defined as in Equation 8.

는 각각 수평 방향 및 수직 방향의 화소 위치를 나타내는 인덱스 변수이며,

은 각각 수평 방향의 총 화소수 및 수직 방향의 총 화소수임에 따라

는 각각 휘도 영상(Y), 녹색 영상(G), 적색 영상(R)의 공간좌표

위치의 화소값을 나타내게 된다.here,

Are index variables representing pixel positions in the horizontal direction and the vertical direction, respectively.

Are the total number of pixels in the horizontal direction and the total number of pixels in the vertical direction, respectively.

Are spatial coordinates of the luminance image (Y), the green image (G), and the red image (R), respectively.

The pixel value of the position is shown.

First, in the reference image encoding step 110, the luminance image Y synthesized from the red image R, the green image G, and the blue image B of the same image in the reference image setting step S111 is a reference image. In the reference video code / decoding step (S112), the reference video is encoded to generate reference video data. In this case, any encoding scheme known so far may be adopted as the encoding scheme used. Typical examples include a number of coding techniques including differential pulse code modulation (DPCM) coding, discrete cosine coding, wavelet transform coding, subband coding, fractal coding, and object-based coding. Then, the first reference decoded image is generated by decoding the reference image data in the reverse order of encoding.

Here, the basis for setting the reference image as the luminance image (Y) includes most of the information about the image and is not only sensitive to human visual characteristics, but also the red image (R), green image (G), and blue image of the same image. This is because the compound can be easily synthesized by substituting (B) into the color conversion equation.

At this time, in another embodiment of the present invention, the green image G, which is a component of the luminance image Y, may be used as the reference image, and if necessary, a red image R or a blue image B may be used. May be set as the reference image, but as in the preferred embodiment of the present invention, it is most preferable to set the reference image as the luminance image (Y).

In addition, the encoding end encodes the original reference image and then decodes the first reference decoded image to generate the actual reference image data. After estimating the subtraction factor, it is to be encoded and decoded using this. That is, the first reference decoded image and the second reference decoded image may be regarded as the same image when an error that may occur on a channel (transmission path, storage medium, etc.) is excluded. In this case, although the original reference image which is not decoded may be used as the actual reference image data, it is preferable to use the decoded first reference decoded image.

Subsequently, in the first estimation image encoding step S120, the green image G is set as the first estimation target image in the first estimation target image setting step S121, and in the first factor estimation encoding step S122. The first proportional factor, which is the stretching information for minimizing the least square error between the first reference decoded image, and the first estimation target image, and the first decrement factor, which is the slicing information, are estimated and encoded on a block basis.

Subsequently, in the second estimation image encoding step S130, the red image R is set as the second estimation object image in the second estimation target image setting step S131, and in the second factor estimation encoding step S132. A second proportional factor, which is stretch information for minimizing a least square error between the first reference decoded image, and a second estimated object image, and a second additive factor that is slicing information, are estimated based on a block basis and encoded.

In fact, the procedure of performing the first estimated image encoding step S120 and the second estimated image encoding step S130 is independent of the encoding performance and does not cause an error in the performing process. That is, since the first estimated image encoding step S120 and the second estimated image encoding step S130 are performed independently of each other, after performing the second estimated image encoding step S130, the first estimated image is performed. The encoding step S120 may be performed.

Hereinafter, a process of performing the first estimated video encoding step S120 and the second estimated video encoding step S130 through a formal expression will be described as follows.

As described above, the proportional factor and the subtractive factor may be calculated for the entire image. However, as a serious local error occurs in the predicted image according to the above method, a very visually disturbing distortion of the reconstructed image occurs. In order to overcome this problem, it is necessary to divide the image into a plurality of blocks and calculate a proportional factor and a subtractive factor that can replace each block. That is, it is necessary to calculate proportional factor and decrement factor respectively by block-based.

Accordingly, in a preferred embodiment of the present invention, the pixel set for the luminance image Y as the reference image, the green image G as the first estimation target image, and the red image R as the second estimation target image is expressed by Equation 8 as follows. When defined together, each image is divided into block units defined as in Equation (9).

However, in the present invention, since the luminance image Y and the first luminance decoded image after decoding the luminance image Y can be used in the same sense as broadly, for convenience of mathematical expression, The first luminance decoded image and the luminance image Y will be represented by the expression 'luminance image Y'.

은 각각 수평 방향 및 수직 방향의 블럭의 순서적인 위치를 나타내는 인덱스 변수이고,

는 각각 수평 및 수직 방향의 총 블럭수임에 따라

는 각각 휘도 영상(Y)과 녹색 영상(G) 및 적색 영상(R)의 블럭의 순서적인 위치를 나타나게 된다. 예를 들어,

일 경우, 이것은 휘도 영상(Y)에서 수평 방향으로 3번째, 수직 방향으로 10번째에 위치하는 블럭을 나타낸다.here,

Are index variables representing the ordered positions of the blocks in the horizontal direction and the vertical direction, respectively.

Is the total number of blocks in the horizontal and vertical directions, respectively.

Denotes the ordered positions of blocks of the luminance image Y, the green image G, and the red image R, respectively. E.g,

In this case, this represents a block located third in the horizontal direction and tenth in the vertical direction in the luminance image Y.

On the other hand, in the relationship between the luminance image (Y), the green image (G) and the red image (R), the equation representing the process of obtaining the proportional factor and the decrement factor for each block is expressed by Equation (10).

,

는 각각 수평 방향으로

번째, 수직 방향으로

번째에 위치한 블럭에 대한 제 1 비례 인자 및 제 1 비례 인자이고,

,

는 각각 수평 방향으로

번째, 수직 방향으로

번째에 위치한 블럭에 대한 제 1 가감 인자 및 제 2 가감 인자이며,

는 각각 단위 블럭의 수평 방향 및 수직 방향의 총 화소수를 나타낸다.here,

,

Are each in the horizontal direction

In the vertical direction

A first proportional factor and a first proportional factor for the block located at the first;

,

Are each in the horizontal direction

In the vertical direction

A first acceleration factor and a second acceleration factor for the block located at the first;

Denotes the total number of pixels in the horizontal and vertical directions of the unit block, respectively.

는 사각형을 이루는 여하의 크기로 설정 가능하며, 통상, 8화소×8화소, 16화소×16화소, 32화소×32화소 중 어느 하나로 정하는 것이 적절하며, 본 발명의 바람직한 실시예에서는 16화소×16화소로 정하는 것이 바람직하며, 블럭의 크기를 너무 크게 설정하면 압축율은 증가시킬 수 있으나 화질이 열화되고, 블럭의 크기를 너무 작게 설정하면 압축율이 저하되므로 블럭의 크기에 따라 압축율과 화질은 상호 타협(tradeoff) 관계를 형성한다.At this time, the size of the block

Can be set to any size forming a rectangle, and it is generally appropriate to set one of 8 pixels × 8 pixels, 16 pixels × 16 pixels, and 32 pixels × 32 pixels, and in the preferred embodiment of the present invention, 16 pixels × 16 If you set the block size too large, the compression rate can be increased, but the image quality deteriorates. If the size of the block is set too small, the compression rate is lowered. tradeoff) form a relationship.

번째, 수직 방향으로

번째에 위치한 블럭과 제 1 추정 대상 영상인 녹색 영상(G)에서 수평 방향으로

번째, 수직 방향으로

번째에 위치한 블럭을 최적으로 근사화시키기 위해, 수학식 10에서

값을 최소화시키는 제 1 비례 인자

와 제 1 가감 인자

를 구하는 과정을 전체 블럭에 적용함으로써 휘도 영상(Y)과 제 1 비례 인자

와 제 1 가감 인자

만을 이용하여 휘도 영상(Y)과 녹색 영상(G)을 부호화할 수 있다.In the first estimated image encoding step S120, the luminance image Y, which is a reference image, is moved in a horizontal direction.

In the vertical direction

In the horizontal direction from the first block and the green image G, which is the first image to be estimated

In the vertical direction

In order to optimally approximate the block located at first,

First proportional factor that minimizes the value

And the first accelerating factor

By applying the process to find the entire block to the luminance image (Y) and the first proportional factor

And the first accelerating factor

Luminance image (Y) and green image (G) may be encoded by using only.

및

를 구하는 방법은 수학식 11에 나타낸 바와 같다.At this time,

And

The method of obtaining is as shown in Equation (11).

에 대하여 각각

및

를 변수로 편미분한 결과가 영이되도록 하는

및

를 구하면 되는 데, 수학식 11과 같은 조건을 만족시키는

및

에 대한 결과식은 수학식 12와 같다.In other words,

Against each

And

Which results in the partial differential of

And

To obtain a condition that satisfies the condition

And

The resulting equation for is given by Equation 12.

의 분모항이 수학식 13과 같이 영이되면,At this time, in Equation 12

If the denominator of is zero as in Equation 13,

이 영으로 나누어짐에 따라 연산 결과가 발산하게 되는데, 이와 같은 경우에는 수학식 14를 이용함으로써 이와 같은 문제를 해결할 수 있다.

As this is divided by zero, the calculation result is diverged. In such a case, this problem can be solved by using Equation (14).

번째, 수직 방향으로

번째에 위치한 블럭과 제 2 추정 대상 영상인 적색 영상(R)에서 수평 방향으로

번째, 수직 방향으로

번째에 위치한 블럭을 최적으로 근사화시키기 위해, 수학식 10에서

값을 최소화시키는 제 2 비례 인자

와 제 2 가감 인자

를 구하는 과정을 전체 블럭에 적용함으로써 휘도 영상(Y)과 제 2 비례 인자

와 제 2 가감 인자

만을 이용하여 휘도 영상(Y)과 적색 영상(R)을 부호화할 수 있다.In addition, in the second estimation image encoding step S130, the luminance image Y, which is a reference image, is moved in a horizontal direction.

In the vertical direction

In the horizontal direction from the block located at the first position and the red image R as the second estimation target image

In the vertical direction

In order to optimally approximate the block located at first,

Second proportional factor that minimizes the value

And the second acceleration and decrease factor

The luminance image Y and the second proportional factor are obtained by applying the process of

And the second acceleration and decrease factor

The luminance image Y and the red image R may be encoded using only.

및

를 구하는 방법은 수학식 15에 나타낸 바와 같다.At this time,

And

The method of obtaining is as shown in (15).

에 대하여 각각

및

를 변수로 편미분한 결과가 영이되도록 하는

및

를 구하면 되는 데, 수학식 11과 같은 조건을 만족시키는

및

에 대한 결과식은 수학식 12와 같다.In other words,

Against each

And

Which results in the partial differential of

And

To obtain a condition that satisfies the condition

And

The resulting equation for is given by Equation 12.

의 분모항이 수학식 13과 같이 영이되면,

이 영으로 나누어짐에 따라 연산 결과가 발산하게 되는데, 이와 같은 경우에는 수학식 17을 이용함으로써 이와 같은 문제를 해결할 수 있다.At this time, in Equation 16

If the denominator of is zero as in Equation 13,

The result of the calculation is diverted by dividing by zero, and in this case, this problem can be solved by using Equation (17).

Subsequently, in the reference image decoding step (S210), the reference image data is received and decoded to reproduce a second reference decoded image corresponding to the reference image.

및 상기 제 1 가감 인자

를 입력받아 상기 제 2 기준 복호 영상에서 제 1 추정 대상 영상과 같은 스펙트럴 대역을 갖는 제 1 추정 복호 영상을 추정해내고, 상기 제 2 비례 인자

및 상기 제 2 가감 인자

를 입력받아 상기 제 2 기준 복호 영상에서 제 2 추정 대상 영상과 같은 스펙트럴 대역을 갖는 제 2 추정 복호 영상을 추정해낸다.Accordingly, in the estimated image decoding step (S220), the first proportional factor

And said first accelerating factor

Estimates a first estimated decoded image having the same spectral band as the first estimated target image from the second reference decoded image, and receives the second proportional factor.

And the second adding and decreasing factor

The second estimated decoded image having the same spectral band as the second estimated target image is estimated from the second reference decoded image.

, 제 1 가감 인자

, 제 2 기준 복호 영상을 수학식 18에 대입하여 제 1 추정 대상 영상 즉, 녹색 영상(G)과 동일한 스펙트럼 대역을 갖는 제 1 추정 복호 영상을 블럭 기반으로 산출하고,That is, in the first estimated image decoding step (S221), the first proportional factor is applied to all blocks.

, First adding / decreasing factor

, By substituting the second reference decoded image into Equation 18, calculating a first estimated decoded image having the same spectral band as the first estimated target image, ie, the green image G, on a block basis,

, 제 2 가감 인자

, 제 2 기준 복호 영상을 수학식 19에 대입하여 제 2 추정 대상 영상 즉, 적색 영상(R)과 동일한 스펙트럴 대역을 갖는 제 2 추정 복호 영상을 블럭 기반으로 산출한다.In the second estimated image decoding step (S222), a second proportional factor is applied to all blocks.

2nd accelerating factor

, By substituting the second reference decoded image into Equation 19, a second estimated decoded image having the same spectral band as that of the red image R is calculated on a block basis.

Subsequently, in the calculating of the third color image (S230), the first reference decoded image, the first estimated decoded image, and the first estimated decoded image are substituted into the colorimetric conversion equation represented by Equation 1 or Equation 2 to perform the same operation. An image having the same spectral band as the other image except for the first and second estimation target images among the three color images of the image is calculated through calculation.

As described above, in the preferred embodiment of the present invention, the reference image is defined as the luminance image Y, the first estimation target image is determined as the green image G, and the second estimation target image as the red image R, As a result of computer simulation, one of the luminance image (Y), the red image (R), the green image (G), and the blue image (B) of the same image is defined as the reference image and the remaining images When two images are selected from among two images, one of the two images is set as the first estimation target image, and another image is set as the second estimation target image, that is, the luminance image (Y), the red image (R), and the green color. Even when the reference image, the first estimation target image, and the second estimation target image are determined by a combination determined by extracting any three images from the image G and the blue image B, there is no significant difference in encoding performance. Can not confirm Eoteuna, the preferred embodiment of the present invention could be obtained when the best coding efficiency.

In addition, even when the original reference video that was not decoded as the actual reference video data and the first reference decoded video that were decoded, it was confirmed that there was no significant difference in encoding performance. However, according to the preferred embodiment of the present invention, In this case, the best coding efficiency can be obtained.

As described above, although the present invention may generate various cases through synthesis and combination using the luminance image (Y), the red image (R), the green image (G), the blue image (B), and the color difference image of the same image, Even if any combination is used, this will be subject to the spirit of the present invention as it will be readily available to those skilled in the art.

Meanwhile, the present invention can divide an image into blocks having various sizes and shapes by estimating a color image based on a block basis, but such modification is also easy for those who have acquired general knowledge in the art. It is therefore apparent that the invention is subject to the spirit of the invention.

If the above-mentioned result of calculating the proportional factor and the subtractive factor is represented as a real number having a decimal point and directly encoded, the data processing technique commonly known as a scaling technique is used to encode the integer form to avoid this as the allocated bits increase. I use it. In other words, to reflect the value below the decimal point while processing integers, the proportional factor and decimal factor having decimal point are multiplied and encoded, and then divided by the same value during decoding. This is a conventional data processing technique. Therefore, further detailed description will be made weak.

The idea of the present invention can obtain high coding efficiency when applied to images having high cross-correlation in addition to color image coding, and the proportional factor and the subtraction factor can be easily applied as pattern recognition parameters in the field of image recognition. have.

As described in detail above, any one of the luminance image (Y), the red image (R), the green image (G), and the blue image (B) of the same image is defined as a reference image and encoding is performed. After generating the reference image data through the decoding, to create a first reference decoded image, select two images from the remaining images to set one of the two images as the first estimation target image and another image to the second estimation target Block-based estimation of the first proportional factor, which is stretching information, and the first accelerating / reducing factor, which is slicing information, to minimize the least square error between the first reference decoded image and the first estimated target image. And a second proportional factor and a slice, which are stretching information for minimizing the least square error between the first reference decoded image and the second estimation target image. According to the video encoding method using spectral redundancy according to the present invention, which estimates and encodes information-based second subtraction factor based on block, spectral redundancy can be directly removed through indirect resolution through low resolution analysis. Compared with the conventional method of removing the PB, it is possible to improve the coding efficiency relatively high and, as it is composed of a repetitive procedure as a whole, the hardware structure is simple, so it is easy to implement, and it is easily combined with the known video encoding method. It can be used to create a high technical ripple effect.

Claims (26)

In the image encoding method for encoding by removing the redundancy inherent between the first image and the second image in order to improve the encoding efficiency: Stretching information for setting any one of the first image and the second image as a reference image, and then designating another image as an estimation target image to minimize an error between the reference image and the estimation target image. an estimation encoding step of estimating and encoding a scale factor which is stretching information and an offset factor which is slicing information; And And an estimation decoding step of receiving the reference image, the data encoding the proportional factor and the subtraction factor, and decoding the data in the reverse order of the estimation encoding step to reproduce the first image and the second image. Image coding method using parallel redundancy. The method of claim 1, wherein the estimation encoding step, A reference image encoding step of setting one of the first image and the second image as the reference image, generating reference image data through encoding, and decoding the first image to generate a first reference decoded image; And Another image is set as the estimation target image, and the proportional factor and the slicing information, the stretching factor and the slicing information, are stretched to minimize the error between the first reference decoded image and the estimated target image. And an estimated video encoding step of encoding after estimating. The method of claim 2, wherein the estimation decoding step, A reference image decoding step of receiving and decoding the reference image data and reproducing a second reference decoded image corresponding to the reference image; And And an estimated image decoding step of receiving and decoding the data obtained by encoding the proportional factor and the subtraction factor and estimating and decoding the estimated decoded image corresponding to the estimated target image in the block unit from the second reference decoded image. Image coding method using spectral redundancy. The image using spectral redundancy according to claim 2 or 3, wherein the block unit has any one of 8 pixels x 8 pixels, 16 pixels x 16 pixels, and 32 pixels x 32 pixels in the horizontal and vertical directions. Coding method. The method according to claim 4, wherein the estimation encoding step, And estimating the proportional factor and the decrement factor equal to the total number of blocks and the same number of the estimated target image, respectively. The method of claim 1, wherein the error is A video encoding method using spectral redundancy, characterized in that the least square error (least square error). The method of claim 1, wherein the reference image, The image encoding method using spectral redundancy, characterized in that the image containing a relatively large amount of the luminance component of the first image and the second image. In an image encoding method of encoding by removing spectral redundancy inherent between color images of the same image photographing the same subject to improve encoding efficiency: After setting any one of the luminance image, the red image, the green image, and the blue image of the same image as a reference image, reference image data is generated through encoding, and among the remaining images except the reference image. A scale factor, which is stretching information, and an offset that is slicing information, which are set to at least one image as an estimation target image, so that an error between the reference image and the estimation target image is minimized. A color image estimation encoding step of estimating and encoding the; And Receives and decodes the reference image data to reproduce a reference decoded image corresponding to the reference image, and receives and decodes data encoded by the proportional factor and the subtraction factor to correspond to the estimated target image in the reference decoded image. And a color image estimation decoding step of estimating an estimated decoded image. The method of claim 8, wherein the color image estimation encoding step comprises: A reference image encoding step of setting any one of the red image, the green image, the blue image, and the luminance image as the reference image and generating the reference image data through encoding; And By selecting two images from the remaining images except for the reference image, one of the two images is set as a first estimation target image, and another image is set as a second estimation target image, so that the reference image and the first estimation target are selected. The first proportional factor, which is the stretching information for minimizing the error between the images, and the first acceleration / decrease factor, which is the slicing information, are estimated and encoded, and the stretching information is such that the minimum square error between the reference image and the second estimated target image is minimized. And a factor estimating encoding step of estimating and encoding a second proportional factor and a second acceleration / decrease factor, which is slicing information. The method of claim 9, wherein the factor estimation encoding step comprises: By selecting two images from the remaining images except for the reference image, one of the two images is set as the first estimation target image, and a minimum square error between the reference image and the first estimation target image is minimum. A first estimated image encoding step of estimating and encoding on a block basis the first proportional factor, which is stretching information, and the first acceleration / decrease factor, which is slicing information; And The second proportional factor, which is stretch information, which sets another image as the second estimated object image, and minimizes the least square error between the reference image and the second estimated object image; And a second estimated video encoding step of estimating and encoding the block on a block basis. The method of claim 8, wherein the color image estimation decoding step comprises: A reference image decoding step of receiving and decoding the reference image data and reproducing a reference decoded image corresponding to the reference image; And The first estimation target image and the second estimation target image from the reference decoded image are decoded by receiving data obtained by encoding the first proportional factor, the first acceleration factor, the second proportion factor, and the second acceleration factor. And an estimated image decoding step of estimating, on a block basis, the first estimated decoded image and the second estimated decoded image respectively corresponding to the image encoding method using spectral redundancy. The method of claim 11, wherein the color image estimation decoding step comprises: The reference decoded image, the first estimated decoded image, and the first estimated decoded image are substituted into a colorimetric conversion equation, and one image of the three color images of the same image, except for the first and second estimated target images. And calculating a third color image by calculating an image corresponding to the third color image. The method of claim 11, wherein the decoding of the estimated image comprises: Block-based estimation of the first estimated decoded image having the same spectral band as the first estimated target image from the reference decoded image by receiving and decoding data obtained by encoding the first proportional factor and the first subtraction factor. A first estimated image decoding step; And Block-based estimation of the second estimated decoded image having the same spectral band as the second estimated target image from the reference decoded image by receiving and decoding data obtained by encoding the second proportional factor and the second subtraction factor. And a second estimated video decoding step. The method of claim 8, wherein the reference image, The image encoding method using spectral redundancy, characterized in that the luminance image. 10. The method of claim 8, wherein the first estimation target image is one of the green image and the red image, and the second estimation target image is another image. . The method of claim 10, wherein the factor estimation encoding step comprises: 12. A video encoding method using spectral redundancy, wherein the number of pixels in the horizontal and vertical directions is based on one of 8 pixels x 8 pixels, 16 pixels x 16 pixels, and 32 pixels x 32 pixels. The method of claim 16, wherein the color image estimation encoding step comprises: And estimating the proportional factor and the decrement factor equal to the total number of blocks and the same number of each of the estimated target images, respectively. In an image encoding method of encoding by removing spectral redundancy inherent between color images of the same image photographing the same subject to improve encoding efficiency: One of the luminance image, the red image, the green image, and the blue image of the same image is set as a reference image to generate reference image data through encoding, and then decoded to generate a first reference decoded image. A scale factor, which is stretching information, to minimize the error between the first reference decoded image and the estimated target image by setting at least one image from among the remaining images except the reference image as an estimated target image; A color image estimation encoding step of estimating and encoding an offset that is slicing information; And Receives and decodes the reference image data to reproduce a second reference decoded image corresponding to the reference image, receives and decodes the data obtained by encoding the proportional factor and the subtraction factor, and decodes the estimated object in the second reference decoded image. And a color image estimation decoding step of estimating an estimated decoded image corresponding to the image. The method of claim 18, wherein the color image estimation encoding step comprises: A reference image encoding step of deciding one of the red image, the green image, the blue image, and the luminance image as the reference image, generating the reference image data through encoding, and then decoding the image to generate the first reference decoded image ; And By selecting two images from the remaining images except the reference image, one of the two images is set as the first estimation target image, and another image is set as the second estimation target image, so that the second reference decoded image and the first image are selected. A first proportional factor, which is stretching information, and a first acceleration / decrease factor, which is slicing information, are estimated and encoded so that an error between an estimated object image is minimized, and a least square error between the first reference decoded image and the second estimated object image is minimum. And a factor estimating encoding step of estimating and encoding a second proportional factor that is stretching information and a second acceleration / decrease factor that is slicing information. 20. The method of claim 19, wherein the factor estimation encoding step comprises: Least square error between the first reference decoded image and the first estimation target image by selecting two images from the remaining images except the reference image and setting one of the two images as the first estimation target image A first estimated image encoding step of estimating and encoding on a block basis the first proportional factor, which is stretching information, and the first acceleration / decrease factor, which is slicing information, so that N) is minimized; And The second proportional factor and slicing information, which is stretching information that sets another image as the second estimation target image to minimize the least square error between the first reference decoded image and the second estimation target image. And a second estimated video encoding step of estimating and encoding the two additive factors on a block-based basis. The method of claim 20, wherein the color image estimation decoding step comprises: A reference image decoding step of receiving and decoding the reference image data to reproduce a second reference decoded image corresponding to the reference image; And Block-based the first estimated decoded image having the same spectral band as the first estimated target image in the second reference decoded image by receiving and decoding data obtained by encoding the first proportional factor and the first subtraction factor. Estimating a first estimated image; And Block-based the second estimated decoded image having the same spectral band as the second estimated target image in the second reference decoded image by receiving and decoding data obtained by encoding the second proportional factor and the second subtraction factor. And a second estimated image decoding step of estimating. The method of claim 21, wherein the color image estimation decoding step comprises: The second reference decoded image, the first estimated decoded image, and the first estimated decoded image are substituted into the colorimetric conversion equation, and the remaining three images of the same image except the first and second estimated target images are excluded. And a third color image calculating step of calculating an image corresponding to one image through calculation. The method of claim 18, wherein the reference image, The image encoding method using spectral redundancy, characterized in that the luminance image. 19. The method of claim 18, wherein the first estimation target image is one of the green image and the red image, and the second estimation target image is another image. . The method of claim 20, wherein the factor estimation encoding step comprises: 12. A video encoding method using spectral redundancy, wherein the number of pixels in the horizontal and vertical directions is based on a block of any one of 8 pixels x 8 pixels, 16 pixels x 16 pixels, and 32 pixels x 32 pixels. The method of claim 25, wherein the color image estimation encoding step comprises: And estimating the total number of blocks of each of the estimated target images, the same proportional factor and the added and subtracted factors on a block basis, respectively.

Images