KR980013403A - Progressive image transmission method using adaptive block truncation coding - Google Patents

Abstract

The present invention relates to a progressive image transmission method using adaptive block truncation coding. In particular, the present invention relates to an incremental image transmission method using adaptive block truncation coding, which is simple in algorithm and can facilitate the implementation of software or hardware.

An object of the present invention is to provide a new progressive image transmission method using an adaptive block truncation coding which has a simple algorithm and excellent data compression ratio. To achieve the above object, the present invention provides an improved adaptive block truncation coding Adaptive Block Truncation Coding (ABTC) method and a new type of ABTC transmission parameter and transmission format.

The present invention presents a more improved HVS model. The equation is expressed as follows.

If? <k for? <T ₁ , then? = 0

If α <k ₂ for T ₁ ≤ D <T ₂ , α = 0

If? &lt; k ₃ for?? T ₂ ,? = 0

Where T ₁ = 50, T ₂ = 120, k ₁ = -0.085 硫 +40, k ₂ = 6, and k ₃ = 0.075 硫 + 1.742

In the above-mentioned jn.d motel, α means each standard deviation, and β means each average. The method of obtaining the improved ABTC by applying the characteristics of the HVS to the extended BTC using this equation is as follows

Let α = σ and β = X, and if α <β, α = 0, ie σ = o.

Let α ₁ = σ _h and β ₁ = X _h , and if α ₁ <β _1, then α ₁ = 0, ie σ _h = 0.

Let α ₂ = σ ₁ and β ₂ = X _1. If α ₂ <β _2, then α ₂ = 0, that is, σ ₁ = 0.

The progressive transmission of the ABTC according to the present invention all has three passes. In the first pass, the degraded image is transmitted so as to recognize only the image type, and in the second pass, the detailed image information is reproduced through the additional data transmission.

And the final image is reproduced in the third pass. Each of the second and third passes reproduces an image using the data transmitted in the previous pass, so that each pass has a hierarchical structure. Fig. 5 shows the parameters to be transmitted for each path in the case of MP and MMSE for each pattern.

The transport format of the ABTC consists largely of a header and a payload. The title includes information necessary for decoding. Format, mode-flag, pass-level, vertical-size and horizontal-size.

Description

Progressive image transmission method using adaptive block truncation coding

The present invention relates to an incremental image transmission method using adaptive block truncation coding, and more particularly, to an incremental image transmission method using adaptive block truncation coding,

Hypertext Transfer Protocol (HTTP), an application layer protocol using TCP / IP that appeared in 1990, has been used in conjunction with the World Wide Web (WWW). The Internet, which has been recognized as an existing international academic network by the spread of the WWW, has been easily used by many users and has become a commercial opportunity. The remarkable growth of the Internet makes it possible to construct this infrastructure as a basic infrastructure for the super - speeding of communication networks in each country.

The international standardization study on the compression method for storing and transmitting a vast amount of video signals, which is the most important element in the multimedia service, starts from JBIG which defines the compression method of the binary image for facsimile, H.261 for video telephony, PEG-1 for digital storage media, and MPEG-2 for compressing and transmitting high-definition TV signals.

However, in spite of the development of the technique for the compression method, since the development of the peripheral device for the terminal is insufficient, the recognition of the multimedia is not widely spread yet. The current Internet, which is providing multimedia services through computer or data communication, will be highly appreciated according to the evolutionary strategy. However, since the bandwidth is not enough to accommodate the wide range of video data, the demand is gradually increasing There is a limit to the provision of full-scale multimedia services.

There are many types of images available on the Internet. Among them, GIF (Graphics Interchange format) and JPBG are mainly used as still image formats. Although the importance of still images is present in the still image information itself, it facilitates the visual search of the image database in the provision of moving images because of the progressive image transmission. However, the progressive transmission method of JPEG There is a complicated problem, and interlaced-GIF, which shuffles and transmits each line data among GIFs, is not strictly a gradual method.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a new gradual image transmission method using adaptive block truncation coding, which has a simple algorithm and excellent data compression ratio.

In order to achieve the above object, the present invention provides an improved adaptive block truncation coding (ABTC) method and a new type of ABTC transmission parameter and transmission format.

The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Adaptive block truncation coding has been proposed as a method for solving the drawbacks caused by discrete cosine transform (DCT) on edge parts of an image in digital video coding such as computer graphics or animation.

ABTC, which is a pure intra mode non-transformed image coding method without reference image, shows excellent performance in still images as well as moving images.

Conventional block truncation coding (hereinafter referred to as BTC) reproduces blocks of 4 × 4 pixels at two levels, whereas ABTC obtains four reproduction level values, (Hereinafter referred to as &quot; jnd &quot;) of the HVS, The number of levels to be reproduced varies depending on the characteristics to be one to four.

In order to expand the BTC to play the block at four levels, input pixels should be classified into four sets as follows.

A = {X _ij : X _ij ≥ X _h }

B = {X _ij : X? X _ij <X _h } (1)

C = {X _ij : X ₁ ? X _ij <X}

D = {X _ij X _ij <X ₁ }, 0? _I , _j? 3

Where X _h means the average rate of pixels greater than the pixel average X of the input block and X ₁ means the average of pixels less than X

Let p, q, s, and r denote the number of pixels of the sets A, B, C, and D in the above equation (1)

The method of determining the reproduction level for the pixels classified from the formula (1) includes the same first and second Moment Preserving (MP) methods as those of the conventional BTC and the Minimizing Mean Squared (MMSB) method Errof) method. (2) and (3), respectively, depending on the method.

In Equation (2), σ _h and σ ₁ denote the standard deviation of pixels with X or more and less than X, respectively, and r ₁ , r ₂ , r ₃ and r ₄ in Equation (2) Quot; reproduction level &quot; The reproduction values in the case of MP can be obtained in the same manner as the conventional BTC, and the reproduction values in the case of MMSE are determined as the average value of the pixels corresponding to each set.

The expansion result of BTC obtained from the above is divided into five patterns as shown in FIG. 1 according to statistical characteristics of each sub-image.

In FIG. 1, the first pattern is a case in which pixels in a block have a very large correlation. The second pattern indicates a large correlation between pixels larger than X and pixels smaller than X, respectively. The third pattern is a case in which pixels smaller than X are relatively widely distributed, whereas pixels larger than X exhibit a large redemption relationship. The fourth pattern is the opposite of the third pattern. The fifth pattern corresponds to the case where the pixels have a wide distribution.

The data rate of the extended BTC is 4 bpp as in the case of MP, since it requires 8 bits for X _h , X ₁ , σ _h and σ ₁ for MP, and 32 bits for bit plane transmission. For MMSE, the X _hh, 32 bits are required for an 8-bit, each bit-plane transferred to the transmission of X _hl, X _m, and X ₁₁ is a so 4bpp, as in the case of MP. This is a transmission rate that is twice the recording BTC. The high data rate can be reduced by using the jnd of HVS. However, the HVS model used in the conventional adaptive block truncation coding has a disadvantage in that the blockiness phenomenon partially occurs in the existing BTC.

Accordingly, the present invention proposes solutions thereto.

If each of X, X _h, and X ₁ is a background brightness I and the reproduction level is I +? I, then ΔI, which is a jnd that can be identified by HVS, has a constant value.

Therefore, according to Weber's law, a characteristic that is more sensitive to changes in intensity in a dark region than in a bright region is obtained.

The present invention presents a more improved HVS model. The equation is expressed as follows.

If? <k ₁ for? <T ₁ ,? = 0

If? <K ₂ for Tl? P <T ₂ ,? = 0 (4)

If? &lt; k ₃ for?? T ₂ ,? = 0

Where T ₁ = 50, T ₂ = 120, k ₁ = -0.085D + 40, k ₂ = 6, and k ₃ = 0.075 硫 + 1.742

J.n.d. in equation (4). In the model, α means each standard deviation, and β means each mean. The method of obtaining the improved ABTC by applying the characteristics of the HVS to the extended BTC using this equation is as follows.

α = 0, β = X, and if α <β, α = 0, ie σ = 0.

Let α ₁ = σ _h and β ₁ = X _h . If α ₁ <β _1, then α ₁ = 0 and σ _h = 0.

Let α ₂ = σ ₁ and β ₂ = X _1. If α ₂ <β _2, then α ₂ = 0, that is, σ ₁ = 0

FIG. 2 shows the pattern of occurrence of ABTC obtained by applying j.n.d of HVS in equation (4) to extended BTC using the above method. In the ABTC transmission, the necessary parameters are different for each pattern, which is shown in FIG.

4 shows a case where the result pattern of the MMSB-ABTC coding of one block of the girl 1 is frequently used as a test image. In FIG. 2, a pixel having a value smaller than X is reproduced as X ₁ , Pixels are reproduced with two values. Therefore, as shown in FIG. 4, the BPL (Lower Bit Plane) needs to be transmitted, but the transmission of the BPH (High Bit Plane) can be restored by transmitting only the bit plane information of the pixel whose BPL value is "1" Do. Therefore, the BPH information indicated by "X" does not need to be transmitted. In the case of the third pattern, transmission can be performed on the same principle as described above.

Hereinafter, the progressive image transmission using the ABTC will be described.

The progressive transmission of the ABTC according to the present invention all has three passes. In the first pass, the degraded image is transmitted so as to recognize only the image type, and in the second pass, the detailed image information is reproduced through the additional data transmission.

And the final image is reproduced in the third pass. Each of the second and third passes reproduces an image using the data transmitted in the previous pass, so that each pass has a hierarchical structure. Fig. 5 shows the parameters to be transmitted for each path in the case of MP and MMSE for each pattern.

Referring to FIG. 5, MP-ABTC will be described as follows.

The first pattern is a case in which one block is reproduced only by an average value, and reproduction is possible if only X is used. No further information is needed for the second and third passes.

The second pattern is a block reproduced at two levels as in the conventional BTC. In the first pass, only X is sent. In the second pass,? And BPL information are transmitted.

The third pattern is a case where a region of pixels smaller than X, which is the average of the blocks, is reproduced at two levels. Therefore, _Xh is required in the first pass, and the entire block is reproduced at this value. In the second pass, X, and BPL information are transmitted. At this time, only the pixel whose BPL value is "0" is changed to X ₁ . In the third pass,? ₁ and BPH are transmitted. In the third pass, the blocks are finally reproduced using the time received in the second pass and? ₁ .

The fourth pattern is reproduced through a process similar to that of the third pattern when compared with the third pattern.

In the fifth pattern, the block is reproduced only by the entire average value in the first pass, the two levels are reproduced in the second pass, and finally the four levels are reproduced in the third pass. In FIG. 5, the numbers in parentheses indicate the number of bits required for transmission, and V means the number of variable bits.

6 is a transmission format. The transport format of the ABTC consists largely of a header and a payload. The title includes information necessary for decoding. Format, mode-flag, pass-level, vertical-size and horizontal-size.

The format means the color signal format of the input image (8 bits).

1: 420 YIQ format

2: 22 YIQ format

3: 444 YIQ format

4: RGB format

5: GRAY format

Other undefined

The mode-indication tells you how to calculate the playback level (4 bits).

0: MP ABTC

1: MMSE ABTC

Other undefined

The pass-level means the number of passes to be decoded (4 bits) and is currently fixed at 3.

Vertical-size means the vertical size of the image (32 bits).

Horizontal-size refers to the horizontal size of the image (32 bits).

In the payload, the color signal component of the block is transmitted for each pass. The values transmitted at this time are as shown in FIG. Here, the transmission of the block parameter according to the color signal format is the same as the format of the macro block defined in MPEG-2.

As described above, the present invention provides an algorithm for progressive transmission of still images using adaptive block truncation coding. The method of the present invention is simpler than JPEG in terms of algorithm when compared with conventional JPEG, and thus it is easy to implement software and hardware. In addition, the present invention can efficiently provide image information even in an independent system, and can be particularly useful for a still image transmission in the Internet or a video database search.

It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the appended claims and their equivalents. .

FIG. 1 shows an extended block truncation coding pattern,

FIG. 2 shows an adaptive block truncation coding pattern,

FIG. 3 is a diagram showing parameters of each pattern for adaptive block truncation coding transmission;

4 shows an embodiment of adaptive block truncation coding,

FIG. 5 is a diagram showing parameters for progressive transmission for each pass,

FIG. 6 is a diagram showing a transmission format of adaptive block truncation coding.

Claims (5)

A progressive image transmission method using adaptive block truncation coding, wherein a just noticeable difference motel of the human visual system satisfies the following equation: If? <k ₁ for? <T ₁ ? = 0 If α <k ₂ for T ₁ ≤ β <T ₂ , α = 0 If? &lt; k ₃ for?? T ₁ ,? = 0 (? is the standard deviation of the pixel,? is the average, T ₁ = 50, T ₂ = 120, k ₁ = -0.085D + 40, k ₂ = 6, k ₃ = 0.075D + 1.742) A progressive image transmission method using adaptive block truncation coding, wherein the adaptive block truncation coding satisfies the following equation. If α <β, α = 0 If α _h <β _h, σ _h = 0 If α ₁ <β _1, σ ₁ = 0 (?: standard deviation of each pixel, ? _h : standard deviation of pixels having a value larger than the average value of the pixels, ? ₁ : standard deviation of pixels having a value smaller than the average value of the pixels, ?: average interval between pixels, ? _h is an average value of pixels having a value larger than the average value of the pixels, ? ₁ : average value of pixels having a value smaller than the average value of the pixels) In the progressive image transmission method using adaptive block truncation coding, there are three paths for transmission, and in the case of the first pattern, the parameter assignment for each path is X for the first path, no information for the second path and the third path Quot; In the case of the second pattern, MP indicates X in the first pass,? And BPL in the second pass, X _h in the first pass, X ₁ and BPL information in the second pass, MP in the third pass, Without assigning any information to both MMSEs, For the third pattern MP is X ₁₁ to X _h, the second pass, the X ₁ and BPL in the first pass, a is σ ₁ and BPH third pass, MMSE are the X _h in the first pass, the second pass and the BPL, the third pass in the case of assigning the X _1h and BPH, and a fourth pattern, MP is the X _h, the second pass, the X ₁ and BPL in the first pass, the third pass has a σ _h and BPH , MMSE in the case of the X _h1 and BPL the X _hh in the first pass, the second pass, the third pass assign X _h1 and BPH, and the fifth pattern, MP is X _h, the second to the first path X ₁ and BPL for the third pass, σ _h and σ ₁ and BPH for the third pass, X times for the first pass, X _1h and BPL for the second pass, X _h1 and X _{11 for} the third pass, And BPH. Progressive image transmission method using adaptive block truncation coding with DMF feature. (Average of greater resolution than the pixel average of X _h X input block, X ₁ Average of pixels smaller than X, sigma: standard deviation of pixel, σ _h , σ _{1 is} the standard deviation of the pixels with X or more and pixels with less than X, A = {x _ij &lt; x _ij &gt; X _h } B = {x _ij : X? X _ij <X _h }, (1) C = {x _ij : X ₁ &lt; x _ij &lt; X} D = {x _ij : x _ij <X ₁ }, 0? I, j? 3 x _ij input pixel) A progressive image transmission method using adaptive block truncation coding is characterized in that the transmission format is composed of a header and a payload for transmitting color signal components for each path. Video transmission method 5. The method of claim 4, wherein the title comprises a format, a mode-flag, a pass-level, a vertical-size and a horizontal-size, Mode, 444 YIQ format, RGB format, and GRAY format, and the mode-indication for calculating the reproduction level is 0 for MP ABTC and 1 for MMSB ABTC, and the pass- Wherein the vertical-size means a vertical size of the image, and the horizontal-size indicates a horizontal size of the image. ※ Note: It is disclosed by the contents of the first application.