KR20070099911A - Method for compressing digital image, and a medium having computer readable program executing the method - Google Patents

Abstract

A method for compressing digital image, and a medium having computer readable program executing the method are provided to reduce a size of an image, compute a DCT(Discrete Cosine Transform) constant using the reduced image value, and quantize the computed result using the quantization table, thereby effectively compressing an image by reducing entropy. A method for compressing digital image comprises the following steps of: blocking digital images(S1); generating changed block values by dividing the blocked original value by a predetermined number and removing the remainders(S2); performing DCT conversion of the changed block value(S3); and quantizing the DCT(Discrete Cosine Transform) conversion value(S4); and using a quantization table multiplied by a predetermined number for recovering a compressed image through the quantization step(S5,S6).

Description

Method for compressing digital image, and a medium having computer readable program executing the method

1 is a schematic flowchart for performing a digital image compression method according to the present invention.

2A to 2D illustrate a process of compressing image data according to a conventional method of compressing a digital image using DCT and quantization.

3A to 3D illustrate a process of compressing image data according to the method of FIG. 1.

TECHNICAL FIELD The present invention relates to digital imaging technology, and more particularly, to a method of compressing a digital image.

Digital image compression is a technique for displaying an image with as few bits as possible while maintaining image quality at a required level in order to overcome limitations in transmission time and storage capacity when transmitting or storing an image with a large amount of data.

Thanks to the development of digital signal processing and ultra-large scale integrated circuit (VLSI) technology that can realize it, digital image compression technology has been applied to many fields such as facsimile, videophone, video conference, digital VCR, digital camera, etc. It is one of the HDTV (High Definition TV) methods underway.

Recently, in the multimedia field that pursues various applications such as education and entertainment by adding a TV and VCR function to the conventional computer (especially a personal computer) which processed only text and graphics, and processing audio, still picture, and even video by integrating it. Digital image compression technology is becoming an important core technology.

In general, digital image compression involves a process of transformation and quantization. Transformation is the process of finding the redundancy of the information in the image. In order to remove the redundancy of the image data, various kinds of mathematical transformations are used to compress the image by transforming and analyzing the image from the spatial domain to another domain. A representative method is a discrete cosine transform (DCT).

Quantization is a process of dividing a DCT coefficient obtained through DCT and the like into certain constants to reduce the number of bits of significant digits. JPEG is a representative image compression method including DCT and quantization.

In the image compression algorithm using the DCT algorithm, the image is blocked in small units, and then DCT transformed and quantized each block. Then, the quantized matrix values are converted into integer strings through zigzag scanning, and then sorted by Huffman coding.

In general, each pixel value is assigned with 8 bits for red, 8 bits for green, and 8 bits for blue, and has a total color value of 24 bits, which is represented by about 1678 million colors.

Algorithms using such DCT generally separate the RGB values of blocks of (8x8 in the case of JPEG) and make matrix values using the following formula.

는 0부터 255까지의 정수값을 가지게 된다. Generally

Has an integer value from 0 to 255.

값 중,

값은 DC(고주파)값, 나머지 값들은 AC(저주파)값이라고 하며,

에 가까운 값들은 이미지의 전체적인 특성을 표현하는 중요한 정보를 담고 있게 된다.Through DCT process

Of the values,

The values are called DC (high frequency) values, and the rest are called AC (low frequency) values.

Values close to contain important information representing the overall characteristics of the image.

을 이용하여, DCT와 양자화 과정을 거쳐서 새로운 행렬이 만들어지게 된다. 이렇게 하여 만들어진 결과값을

라 하고, 그 값은 다음과 같다.Quantization table

By using, new matrix is created through DCT and quantization process. The resulting value

The value is as follows.

를 지그재그 스캐닝해 필요 없는 부분은 없애고, 중요한 부분을 정수열만으로 이미지를 표현하게 된다.So obtained

By zigzag scanning, unnecessary parts are eliminated, and important parts are represented by integer columns only.

Image compression using such DCT coefficients is widely used. Especially, in the case of JPEG (Joint Photographic Experts Group) format, DCT coefficients are calculated and quantized to compress the image.

An object of the present invention is to provide a digital image compression method that can further improve the compression efficiency by further improving the conventional digital image compression method using the above-described DCT transform and quantization.

In order to achieve the above object, a digital image compression method according to the present invention comprises a video block step of blocking a digital image, a modified block value generation step of generating a correction block value by dividing the original block value by a predetermined number and removing the remainder; A DCT transform step of DCT transforming the block value, and a quantization step of quantizing the DCT transform value.

The value produced by uniformly reducing the size of the image value and calculating the DCT coefficient using it and quantizing it using the quantization table can reduce the entropy, which enables more efficient compression of the image.

The method may further include a reconstruction step of using the quantization table multiplied by the predetermined number to reconstruct the compressed image through the quantization step.

When decoding, a new quantization table is used to compensate for the reduction in the size of the image value during compression, so that an image value close to the original value can be obtained, thereby easily increasing the compression rate without large loss of the original image.

In addition, there is also claimed a medium invention in which a computer readable program for executing the method is recorded.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a schematic flowchart for performing a digital image compression method according to the present invention.

The encryption method using the DCT algorithm according to the present embodiment is an algorithm made on the assumption that it is difficult for a person to recognize the changed part when the value of the image is slightly changed, and when the value is insignificant, it is similar or the same color. Regulate.

First, the original image is blocked (S1), and the blocked image value is divided by 2 (S2). In this case, the rest is removed so that the block value after division does not take more bits than the block value before division.

Before the DCT process, if the color was transformed to the value obtained by subtracting each value divided by the predetermined number 2, the restored value is the same or different from 1 when the product is subsequently multiplied by 2 to restore compression. As a result, the color of the original image is hardly different. For example, 135 => 134, 214 => 214, 215 => 214, and so on.

In the present embodiment, 1 is subtracted to remove the remainder divided by the predetermined number 2. However, in another embodiment, the remainder may be removed by adding 1. Also, the predetermined number may be other than two.

Through this process, if the color block deformed from the original f (i, j) is called f '(i, j), f' (i, j) may be expressed as follows.

This value satisfies the following equation.

를 이용하여 DCT알고리즘으로

를 구하면 다음과 같다.Next, a DCT matrix is created from the blocked image (S3).

Using DCT algorithm

Is obtained as follows.

을 이용하여 양자화시킨 값을

라 하고, 그 값을 구하면 다음과 같다.Next, a quantization process is performed (S4). Quantization table

Quantized using

If the value is obtained, it is as follows.

　　　　　　　　　　

　　　　　　　　　　

는

를 DCT과정을 거친 것이므로, 구한 값에 2를 곱하게 되면, 원래의 값과 비슷한 값이 나오게 된다.From here

Is

Since DCT has been processed, multiplying the result by 2 yields a value similar to the original value.

를 이용하여 양자화를 시키고, 복원할 때, Q(u, v)*2를 이용한다면 원본과 큰 차이 없이 압축을 시킬 수 있게 된다.When you quantize

When using Q (u, v) * 2 to quantize and reconstruct using, we can compress without significant difference from the original.

Comparing F and F ', F' compresses the size of each element 1/2 so that the entropy of F 'is lower than that of F, and the entropy can be improved if the entropy is reduced.

Therefore, if this method is applied to the JPEG algorithm, the image compression is performed by expressing the color value of 24 bits in 21 bits.

Finally, zigzag scanning, alignment and compression are performed (S5 and S6).

For more specific example, compare the result of making the image block value with integer string as follows.

2A to 2D illustrate a process of compressing image data according to a conventional method of compressing a digital image using DCT and quantization.

2A is a block value of an original image. If this value is made DCT coefficient, it becomes like FIG. 2B. When the quantization value is obtained using the values of the quantization table shown in FIG. 2c, it is equal to 2d.

69, -1, 16, 11, 0, 0, 0, 0, 0, 6, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, -1 00

3A to 3D are diagrams illustrating a process of compressing image data according to the method of FIG. 1.

A value obtained by dividing the value shown in FIG. 2A by 2 is 3A. If this value is made DCT coefficient, it becomes like figure 3b. If the quantization value is obtained using FIG. 3c, the value is equal to 3d.

55, 0, 8, -5, 0, 0, 0, 0, 0, 0, 3, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 00

Comparing the two values, the former is not the zero value is eight kinds, if using the present invention is six kinds, when compressing it can be reduced in capacity.

This embodiment describes a method for effectively performing DCT conversion in an image compression method including a quantization process by using a DCT algorithm such as JPEG.

In the compression process of the image after DCT algorithm and quantization process, the value is slightly modified in advance, which lowers the number of bits of color, and compresses the image with the lower value, thereby lowering the entropy in the DCT process. It is possible to have an image compression ratio.

The quantization table in the decompression process of compression should be subjected to arithmetic conversion of the contents of the quantization table so that the original image values are obtained, not the quantized values during compression.

According to the present invention, the size of an image value is uniformly reduced, a DCT coefficient is calculated using the same, and the quantization is performed using a quantization table, so that the entropy of the generated value is reduced to enable more efficient compression of the image.

In addition, by using a new quantization table for compensating for the reduction of the size of the image value during compression when decoding, an image value close to the original value can be obtained, thereby increasing the compression rate without large loss of the original image.

Although the present invention has been described in terms of some preferred embodiments, the scope of the present invention should not be limited thereby, but should be constrained by modifications or improvements of the embodiments supported by the claims.

Claims (3)

An image blocking step of blocking the digital image; Generating a modified block value by dividing the blocked original value by a predetermined number and removing a remainder; A DCT conversion step of DCT converting the correction block value; And And a quantization step of quantizing the DCT transform value. The method of claim 1, And a reconstruction step of using a quantization table multiplied by the predetermined number to reconstruct the image compressed by the quantization step. A medium having a computer readable program recorded thereon for executing the method of claim 1.

Images