KR101659377B1 - Method and system for data encoding - Google Patents

Abstract

An encoding method and a system therefor are provided. The encoding method is characterized in that the encoding system quantizes the quantization object data to obtain quantization result data, the encoding system obtains the reconstruction data of the quantization result data, and the reconstruction data obtained when the reconstruction data does not belong to the normal range And storing the correction data in which the quantization result data is corrected by the encoding system as the encoding object data in the storage device.

Description

[0001] The present invention relates to a method and system for data encoding,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data encoding method and system, and more particularly, to an encoding method and system capable of increasing a compression rate by not storing sign bits of data to be coded .

1 is a diagram showing an example of a conventional data encoding method.

Referring to FIG. 1, it is possible to encode predetermined object data (for example, a ₁ to a _n ) of conventional data. The object data may be, for example, data corresponding to pixels included in a frame of an image or a moving image.

The encoding system is predictive of the predetermined target data (e.g., a ₂₎ in order to not to encode each of the predetermined target data (e.g., a ₁ to a _n), encoding a particular target data (e.g., a ₂₎ (E.g., a ₂ -a ₁ ) between the data (e.g., a ₁ ) and the specific object data (e.g., a ₁ ). It is well known that the prediction data (e.g., a1) may be a value corresponding to any one of the pixels adjacent to the target data (e.g., the left pixel) in the case of an image, and various embodiments of what to use as prediction data It is also known that such a method can be performed.

Thus, in each of the target data (e.g., a ₁ to a _n) rather than directly encoding the difference data corresponding to the target data (e.g., a ₁ to a _n) (e. G., A _1, a ₂ -a ₁ , ..., a _n - a _n-1 ), it is well known that data having the same value can appear more frequently and thus have a higher compression ratio. Further, when the target data (e.g., a ₂ ) is restored through decoding, the encoded difference data (e.g., encoded a ₂ -a ₁ ) is decoded and decoded data (e.g., a ₂ -a ₁ ) and prediction data (e.g., a ₁₎ (by giving, for example, adding the two values) through it is possible to restore the target data (e. g., a _2).

On the other hand, when the difference data (e.g., a ₁ , a ₂ -a ₁ , ..., a _{n -an-1} ) is encoded, the target data (e.g., a ₁ through a _n ) Values may be less than lossless.

In this lossless compression, even if the encoded data is decoded, the decoded result data is the original target data, so that it has a value within a predetermined range. That is, the target data is n-bit, represented by (n is a natural number, e.g., 8 bits), the destination data (e.g., a ₁ to a _n), each value of the normal range between 0 2 ⁿ -1 .

That is, even if the data to be encoded that is, difference data (for _{example, a1, a 2 -a 1,} ..., a n - a n-1) , even if this has a negative value of the difference between the data (for example, a1, a (For example, a ₁ to a _n ) is performed using the target data (e.g., _{2-a 1} , ..., a _n - a _n-1 ) For example, a ₁ to a _n itself, so that it always has the value of the normal range. In such a case, said difference data (for _{example, a1, a 2 -a 1,} ..., a n - a n-1) of the at least one even if he is negative, said difference data (for example, a1, a ₂ -a ₁ , A, ..., a _n - a _n-1 ), the result of acquiring the object data (for example, a ₁ through a _n ) always has the value of the normal range.

If this is the case, and the said difference data (for _{example, a1, a 2 -a 1,} ..., a n - a n-1) the sign bit of said destination data normally do not need to store the (sign bit) (e.g., a ₁ To < RTI ID _{= 0.0} > a &_lt; / _RTI > Therefore, in lossless compression of conventional encoding solutions (e.g., JPEG-LS (Lossless), PNG (Portable Network Graphics), etc.), the sign bit may be stored for each solution, Various methods are optionally used, such as changing or ignoring some bits.

However, in the case of lossy compression (Lossy compression), the coding object data, that is, the difference between the data-in (for _{example, a1, a 2 -a 1,} ..., a n a n-1) is the quantized data, that is, the quantization result When decoding is performed using data (e.g., Q (a ₁ ), Q (a ₂ -a ₁ ), ..., Q (a _n -an _-1 ), where Q is a quantization function) , The decoded result may have a certain error, not the target data (for example, a ₁ to a _n ) itself. And the decoded data is not guaranteed to be within the normal range due to the error. Thus, these include the coding object data, if _{((e.g., Q (a 1), Q} (a 2 -a 1), ..., Q (a n -a n-1)) decoding is performed to be stored in the sign bit And the conventional encoding solution also stores up to the sign bit of the data to be coded in lossy compression.

In the case of storing the sign bits, at least one or more bits are required to be stored for the sign bits themselves as compared with the case where the sign bits are not stored, thereby causing a loss in the compression rate.

SUMMARY OF THE INVENTION It is therefore a primary object of the present invention to provide a method and system for performing encoding without storing sign bits of data to be coded even in lossy compression.

According to another aspect of the present invention, there is provided a method of encoding data, the encoding system comprising: quantizing quantization target data to obtain quantization result data; obtaining the restoration data of the quantization result data by the encoding system; And storing the correction data in which the quantization result data is corrected by the encoding system as encoding object data in the storage device when the restoration data does not belong to the normal range.

The encoding method may further include the step of the encoding system storing the quantization result data as the encoding object data when the decoding prediction data belongs to the normal range.

Wherein the encoding system quantizing the quantization target data to obtain quantization result data comprises the steps of the encoding system acquiring difference data between the predicted data of the target data and the target data, and the encoding system converts the difference data into quantization target data As shown in FIG.

Wherein the step of storing the correction data in which the quantization result data is corrected by the encoding system in the storage device as the encoding object data includes a step of, when the restoration data is obtained using the correction data, And correcting the quantized data with the correction data.

The normal range may be a value equal to or greater than 0 and equal to or less than 2 ⁿ -1 when the object data is represented by n (n is an integer) bits.

The step of correcting the quantized data to the correction data so that the restored data belongs to the normal range may include the step of the encoding system adding or subtracting a predetermined correction value from the quantization result data to generate the correction data .

Wherein the step of the encoding system obtaining the reconstruction data of the quantization result data includes a step of the encoding system dequantizing the quantization result data and a step of reconstructing the reconstruction data based on the inversely quantized data and the prediction data corresponding to the quantization result data. And acquiring the data.

Wherein the step of storing the correction data corrected by the encoding system as the encoding target data in the storage device stores only bits other than the sign bit even when the encoding target data is negative.

The encoding method may be stored in a computer-readable recording medium on which the program is recorded.

According to an aspect of the present invention, there is provided an encoding system including a quantization module for quantizing quantization target data to obtain quantization result data, a decoding module for obtaining reconstruction data of the quantization result data, And a correction module for storing the correction data, which is obtained by correcting the quantization result data, as the encoding object data in the storage device when the data does not belong.

The encoding system may further include a preprocessing module for obtaining difference data between the predicted data of the target data and the target data, and specifying the difference data as quantization target data.

The correction module may correct the quantized data to the correction data so that the restored data belongs to the normal range when the restored data is obtained using the correction data.

The decoding module may obtain the restored data based on an inverse quantization degree in which the encoding system dequantizes the quantized result data, and the predicted data corresponding to the inversely quantized data and the quantized result data.

The correction module may store only bits other than the sign bit even if the encoding target data is negative.

According to the technical idea of the present invention, when the coding target data is stored while the lossy compression is performed, if the target data is decoded by decoding the coding target data, the correction is performed so that the restored data always belongs to the normal range, Even when the coding target data has a negative value, the sign bit is not stored.

In addition, since the sign bit is not stored, the compression ratio can be remarkably increased.

BRIEF DESCRIPTION OF THE DRAWINGS A brief description of each drawing is provided to more fully understand the drawings recited in the description of the invention.
1 is a diagram showing an example of a conventional data encoding method.
FIG. 2 shows a flowchart for explaining an encoding method according to an embodiment of the present invention.
FIG. 3 is a diagram illustrating a process of encoding according to an embodiment of the present invention. Referring to FIG.
4 is a diagram showing a schematic configuration of an encoding system according to an embodiment of the present invention.

In order to fully understand the present invention, operational advantages of the present invention, and objects achieved by the practice of the present invention, reference should be made to the accompanying drawings and the accompanying drawings which illustrate preferred embodiments of the present invention.

Also, in this specification, when any one element 'transmits' data to another element, the element may transmit the data directly to the other element, or may be transmitted through at least one other element And may transmit the data to the other component.

Conversely, when one element 'directly transmits' data to another element, it means that the data is transmitted to the other element without passing through another element in the element.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in detail with reference to the preferred embodiments of the present invention with reference to the accompanying drawings. Like reference symbols in the drawings denote like elements.

FIG. 2 shows a flowchart for explaining an encoding method according to an embodiment of the present invention. 3 is a diagram for explaining a process of encoding according to an embodiment of the present invention.

2 and 3, the encoding system in accordance with the technical features of the present invention (100 in Fig. 4) is predetermined for the target data (e.g., a ₁ to a _n), the difference data (for example, a1, d _1, from ... , d _n-1 ) (S100).

The object data (e.g., a ₁ through a _n ) may mean meaningful data that the encoding system 100 stores and uses. According to one example, the object data (e.g., a ₁ through a _n ) may be data corresponding to each pixel of an image or moving picture frame. However, the present invention is not limited to this, and any meaningful data to be compressed through the encoding system 100 may be the object data (e.g., a ₁ to a _n ).

Furthermore, the encoding system 100 may encode the target data of the difference data for encoding (e.g., a ₁ to a _n) (for _{example, a1, d 1, ...,} d n-1). Of course, any one (e. G., A ₁₎ of the difference data (for _{example, a1, d 1, ...,} d n-1) may be a target data itself. Where d _n = a _{n + 1} - a _n .

By using the difference data (for _{example, a1, d 1, ...,} d n-1), is between the data sets (set) of the actual encoding to be performed are the values that are repeated appearance it can be increased.

Then, the encoding system 100 may perform a quantization on the difference data (for example, a1, d _1, ..., d _n-1) to carry out the lossy compression (S110). In this case, the difference data (for _{example, a1, d 1, ...,} d n-1) there can be a quantization target data to be quantized.

The quantization may be performed through a predetermined quantization function (e.g., Q ()). Various quantization schemes can be applied to the technical idea of the present invention.

For example, the encoding system 100 may be a value obtained by rounding a value obtained by dividing data to be quantized, that is, data to be quantized, by a predetermined quantization level value (e.g., q). In this case, the quantization function can be defined as follows.

[Equation 1]

Q = [(d + q / 2) / q]

Here, d is data to be quantized, and [] is a function taking an integer value.

Then, the encoding system 100 comprises the quantized target data (for _{example, a1, d 1, ...,} d n-1) to the perform quantization through a predetermined quantization function quantized result data (e. G., Q (a _1), Q (d ₁ ), ..., Q (d _n-1 )) (S110).

The quantized target data (for _{example, a1, d 1, ...,} d n-1) that can have a value of negative. That is, when the prediction data (for example, a ₁ ) of the target data has a larger value than the predetermined target data (for example, a ₂ ), the value of the quantization target data (for example, d ₁ ) . This can have a result of the quantization data (for example, Q (d ₁₎₎ is also the negative of the quantization target data (e.g., d _1).

At this time, in the case of conventional lossy compression, the sign bit of the quantization result data (for example, Q (d ₁ )) has to be stored. That is, when restoration data corresponding to the object data (for example, a ₂ ) is obtained using the quantization result data (e.g., Q (d ₁ )), the restored data has a normal range due to an error due to quantization I can not.

For example, restoration data restored using the quantization result data (e.g., Q (d ₁ )) may be expressed by the following equation.

&Quot; (2) "

Q ^-1 (Q (d ₁ )) + pred (d ₁ )

Here, Q ^-1 may be an inverse quantization function, and pred (d ₁ ) may mean predicted data of d ₁ (for example, a ₁ ). The inverse quantization function Q ^-1 (x) may be x x q.

Thus, data _{^{(Q -1 (Q (d 1}} ))) that is obtained when the compression is lost if the inverse quantization quantization target data (e.g., d ₁₎ the quantization result of the quantization data obtained (e. G., Q (d1)) (E.g., Q ^-1 (Q (d ₁ )) + pred (d ₁ ) corresponding to the target data (for example, a2) There is an error with the data (for example, a2). Due to such an error, the reconstructed data (for example, Q ^-1 (Q (d ₁ )) + pred (d ₁ ) may not have a normal range value, (d1)).

To solve these problems, the encoding system 100 according to the technical idea of the present invention has conventionally used quantization result data (for example, Q (a ₁ ), Q (d ₁ ), ..., Q (d _n-1 ) (Q ₁ ), Q (d ₁ ), ..., Q (d _n-1 )) different from the quantization result data (for example, Huffman coding, etc.).

For this purpose, the encoding system 100 includes a quantization target data (for example, a1, d _1, ..., d _n-1) of the quantizing, respectively, and the quantization result of the quantization data (for example, Q (a _1), Q (d ₁ ), ..., Q (d _n-1 ), respectively, at step S120.

For example, the encoding system 100 obtains the result of the quantization data (for example, Q (d ₁₎₎ to restore the data (e.g., Q ^-1 (Q (d ₁₎₎ + pred (d ₁₎₎ corresponding to each can do.

When the restored data (for example, Q ^-1 (Q (d ₁ )) + pred (d ₁ )) is in the normal range, the quantization result data (for example, Q (d ₁ )) is coded It can be specified by target data.

If the recovered data _{^{(e.g., Q -1 (Q (d 1}} )) + pred (d 1)) that is not a normal range, the recovered data _{^{(e.g., Q -1 (Q (d 1}} )) + pred ( d ₁ )) is less than 0 or greater than 2 ⁿ -1. In this case, the encoding system 100 may generate the quantization result data (for example, Q ¹ (Q (d ₁ )) + pred (d ₁ ) For example, Q (d ₁ ) can be corrected.

For example, when the restored data (for example, Q ^-1 (Q (d ₁ )) + pred (d ₁ ) has a value smaller than the normal range, a predetermined correction value c is stored in the quantization result data Q (d ₁ )). When the restored data (for example, Q ^-1 (Q (d ₁ )) + pred (d ₁ )) has a large value in the normal range, a predetermined correction value c is stored in the quantization result data (d ₁ )).

The correction value (c) is the correction value (c) the calibration data (e. G., Q (d ₁₎ - c, or Q (d ₁₎ + c) with restoration data (e.g., using a Q ^-1 (Q ( _{d 1) - c) + pred} (d 1) or _{^{Q -1 (Q (d 1)}} + c) + pred (d 1)), the obtained restored data (e.g., Q ^-1 when obtaining (Q ( any value may be possible such that the value of d ₁ ) -c) + pred (d ₁ ) or Q ^-1 (Q (d ₁ ) + c) + pred (d ₁ ) has a value in the normal range.

According to an example, as the correction value c is smaller, the restored data using the correction data (for example, Q (d ₁ ) -c or Q (d ₁ ) + c) is smaller than the original quantization result data (d ₁ )) of the restored data.

Therefore, in the case of using the quantization function as in Equation (1), since the error range may be q / 2, even if the correction value (c) is selected as 1, . Of course, the correction value c may be selected depending on the quantization function.

This process can be expressed as the following equation as shown in FIG.

The encoding system 100 quantizes the quantization object data (e.g., a ₁ , d ₁ , ..., d _n-1 ) to obtain quantization result data (e.g., Q (a ₁ ), Q (d ₁ ) d _n-1 )) (S110).

The encoding system 100 may then obtain the corresponding reconstructed data for each of the obtained quantization result data (e.g., Q (a ₁ ), Q (d ₁ ), ..., Q (d _n-1 ) (S120).

Then, the encoding system 100 may optionally result of the quantization data (for _{example, Q (a 1), Q} (d 1) , depending on whether each of the recovered data has the value of the normal range, ..., Q (d _{n- 1} )) can be corrected (S130, S140, S150).

The selective correction of the quantization result data (e.g., Q (a ₁ ), Q (d ₁ ), ..., Q (d _n-1 )) can be expressed, for example,

&Quot; (3) "

C {Q (x)} = Q (x) (if 0≤ Q -1 (Q (x)) + pred (x) ≤ 2 m -1)

= Q (x) - c ( if Q -1 (Q (x)) + pred (x)> 2 m -1)

= Q (x) + c (if Q- ¹ (Q (x)) + pred (x)

Here, m may mean a bit (e.g., 8 bits) representing each object data. The correction value c may be 1 as described above, but is not limited thereto.

As a result, if the reconstructed data of the quantization result data (for example, Q (a ₁ ), Q (d ₁ ), ..., Q (d _n-1 )) is not within the normal range, (e. _{g., Q (a 1), Q} (d 1), ..., Q (d n-1)) encoding the target data (e.g., C {Q (a ₁₎ is a correction for each of the target of encoding the correction data }, C {Q (d ₁ )}, ..., C {Q (d _n-1 )}).

Of the specified each coded object data _{(e.g., {Q (a 1)}} C, C {Q (d 1)}, ..., C {Q (d n-1)}) it is included in the encoding system 100 (S140, S150).

Thus, the corrected coding object data _{(e.g., {Q (a 1)}} C, C {Q (d 1)}, ..., C {Q (d n-1)}) generates the restored data corresponding with the respective C (Q ₁ )}, C {Q (d ₁ )}, ..., and so on, since the generated restoration data always has a value in the normal range, C {Q (d _n-1 )}).

Also, when the encoding system 100 of the present invention encodes a frame of a predetermined image or moving image, the object data (for example, a ₁ to a _n ) may be data corresponding to each pixel, quantized result data (e. _{g., Q (a 1), Q} (d 1), ..., Q (d n-1)) substituted with optionally at least one value (e.g., Q (d ₁₎₎ of the correction data The deterioration of image quality may not be large. On the other hand, the increase in the compression ratio that can be obtained by not storing the sign bit can be remarkable.

The encoding system 100 includes a stored coding object data _{(e.g., C {Q (a 1)} }, C {Q (d 1)}, ..., C {Q (d n-1)}) in a predetermined manner Can be coded. For example, Variable Length Coding (VLC) such as Huffman encoding may be performed, but is not limited thereto.

One embodiment of an encoding system 100 for implementing the technical idea of the present invention is shown in FIG.

4 is a diagram showing a schematic configuration of an encoding system according to an embodiment of the present invention.

Referring to FIG. 4, an encoding system 100 according to an embodiment of the present invention includes a quantization module 120, a decoding module 130, and a correction module 140. The encoding system 100 may further include a preprocessing module 110 and / or an encoding module 150.

The encoding system 100 may be embodied in a data processing device (e.g., a computer, a mobile terminal, etc.) and may include hardware in the data processing device and predetermined software code As shown in FIG.

In this specification, the term 'module' may mean a functional and structural combination of hardware for carrying out the technical idea of the present invention and software for driving the hardware. For example, the 'module' may refer to a logical unit of a predetermined code and a hardware resource for executing the predetermined code, and may be a code physically connected to the module, or a kind of hardware Or may be easily deduced to the average expert in the field of the present invention.

According to an embodiment, the encoding system 100 may not be any physical device, but may be distributed over a plurality of physical devices. If necessary, if each of the configurations included in the encoding system 100 is implemented as independent physical devices, these physical devices may be organically combined through a wired / wireless network so that the encoding system 100 according to the technical idea of the present invention 0.0 > 100 < / RTI >

The quantization module 120 may quantize the quantization target data to obtain quantization result data. The quantization target data may have various implementations according to an encoding solution (for example, JPEG, H.264, etc.), and may vary depending on the source data (or source data) to be encoded and also depending on the encoding solution have. For example, if the source data is an image (or a frame of a moving picture), a predetermined transform (e.g., DCT, etc.) may or may not be performed according to an encoding method, Or may be data corresponding to each of the macroblocks. In addition, there may be various embodiments of the quantization target data depending on what source data is to be encoded or what encoding solution is to be used.

Then, the decoding module 130 can obtain restoration data of the quantization result data. The decoding module 130 may include a predetermined dequantization module (not shown). The decoding module 130 dequantizes the quantization result data, and based on the prediction data corresponding to the dequantized data and the dequantized data, the decoding data, i.e., the decoded data after the predetermined target data is encoded Can be obtained.

The correction module 140 can specify the coding object data based on the obtained restoration data. The correction module 140 may selectively correct the quantization result data as described above.

According to an example, the correction module 140 may store the correction data, which is obtained by correcting the quantization result data, as the coding object data (not shown) when the obtained restoration data does not belong to the normal range. Of course, when the restored data belongs to the normal range, the correction module 140 can directly specify the quantization result data as the coding target data.

When the coding target data is specified, the encoding module 150 can perform coding through a predetermined coding algorithm (e.g., VLC such as Huffman coding).

In addition, the encoding system 100 may acquire source data (for example, an image or the like) to be encoded or predetermined target data obtained from the source data, and generate prediction data of the target data and difference data of the target data And a preprocessing module 110 for specifying the difference data as data to be quantized.

In other words, the target data may mean meaningful data for the encoding system 100 to perform encoding. According to an example, the target data may mean a pixel value of a predetermined image or the like, but is not limited thereto. . It will be readily apparent to one of ordinary skill in the art that any series of data may be encoded in accordance with the teachings of the present invention and that the set of data may be set to the target data .

In addition, as described above, when the correction module 140 acquires the reconstruction data using the correction data, the correction module 140 may correct the quantization data to the correction data such that the reconstruction data belongs to the normal range. For this, the correction module 140 may generate the correction data by adding or subtracting a predetermined correction value from the quantization result data. It should be noted that the correction value may depend on the embodiment of the quantization performed by the quantization module 120, and the correction value may be 1 in the example described above.

When the quantization result data is selectively corrected by the correction module 140, the correction module 140 can store only the remaining bits except for the sign bits in a predetermined storage device even when the coding target data is negative , And the improvement of the compression rate can be remarkably increased due to not storing the sign bit.

The image encoding method according to the embodiment of the present invention can be implemented as a computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a hard disk, a floppy disk, an optical data storage device, and the like in the form of a carrier wave (for example, . In addition, the computer-readable recording medium may be distributed over network-connected computer systems so that computer readable codes can be stored and executed in a distributed manner. And functional programs, codes, and code segments for implementing the present invention can be easily inferred by programmers skilled in the art to which the present invention pertains. BRIEF DESCRIPTION OF THE DRAWINGS The present invention will now be described more fully hereinafter with reference to the accompanying drawings, It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the scope of the present invention. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

Claims (14)

A method for encoding data,
The encoding system quantizing the quantization object data to obtain quantization result data;
The encoding system obtaining reconstruction data of the quantization result data; And
Correction data obtained by correcting the quantization result data by the encoding system when the obtained restoration data does not belong to the normal range; and when the restoration data is obtained using the correction data, And data to be included in the normal range, as the coding object data,
Wherein the step of storing, as coding target data, correction data in which the encoding system has corrected the quantization result data,
And a step of correcting the quantization result data with the correction data by adding or subtracting a correction value that is a predetermined constant that allows the restored data to belong to the normal range when the restored data is obtained using the correction data, Way.
The method according to claim 1,
And if the restored data belongs to the normal range, the encoding system stores the quantized result data as the coding target data.
delete delete delete delete delete 2. The method of claim 1, wherein the step of storing, as coding target data, correction data in which the quantization result data is corrected by the encoding system,
Wherein only the remaining bits excluding the sign bits are stored even if the coding target data is negative.
A computer-readable recording medium recording a program for performing the method according to any one of claims 1, 2, and 8.
A quantization module for quantizing quantization target data and obtaining quantization result data;
A decoding module for obtaining reconstruction data of the quantization result data; And
The correction data obtained by correcting the quantization result data when the obtained restoration data does not belong to the normal range, and the correction data, when the restoration data is obtained using the correction data, And a correction module for storing, as data to be coded, in the storage device,
Wherein the correction module comprises:
Wherein the correction data is corrected by adding or subtracting a correction value that is a predetermined constant that allows the restored data to belong to the normal range when the restored data is obtained by using the correction data.
delete delete delete 11. The apparatus of claim 10, wherein the correction module comprises:
Wherein only the remaining bits excluding the sign bits are stored even when the coding target data is negative.

Images