KR20100012149A - Quantization/inverse quantization apparatus and method and video encoding/decoding apparatus using same - Google Patents

Abstract

PURPOSE: An inverse quantization apparatus/method and an image encoding/decoding apparatus for quantizing a quantization table are provided to improve the image encoding/decoding effect by simplifying a calculation for quantization and inverse quantization. CONSTITUTION: A quantization table input unit(310) inputs a quantization table. A quantization calculator(320) quantizes the frequency coefficient by using a dead zone uniform border quantization. The quantum computing part calculates the absolute value and denotation of the quantized frequency level by using the quantization table.

Description

Quantization / Inverse Quantization Apparatus and Method and Video Encoding / Decoding Apparatus Using Same}

The present invention relates to a quantization / inverse quantization apparatus and method and an image encoding / decoding apparatus using the same. More particularly, the present invention relates to an apparatus and method for improving encoding and decoding efficiency by simplifying calculation for quantization and inverse quantization while reflecting human visual characteristics while encoding or decoding an image.

Moving Picture Experts Group (MPEG) and Video Coding Experts Group (VCEG) have developed video compression techniques that are superior to the existing MPEG-4 Part 2 and H.263 standards. The new standard is called H.264 / AVC (Advanced Video Coding) and was jointly released as MPEG-4 Part 10 AVC and ITU-T Recommendation H.264.

In such H.264 / AVC (hereinafter abbreviated as 'H.264'), integer discrete cosine transform (DCT) is called, and motion estimation and compensation of transform block size (Variable Block Size) Motion Estimation and Compensation, Quantization, Entropy Coding. Among these, motion estimation and compensation, DCT transform, inverse DCT, quantization and inverse quantization are known to be large in computational amount and additionally require large memory when actually implemented. In particular, H.264, which has twice the compression performance compared to MPEG-4, requires more computation.

Therefore, a technique for reducing the amount of computation in the process of quantization and inverse quantization, ultimately facilitating the implementation of an apparatus for encoding or decoding a video, reducing the implementation cost, and improving video encoding or decoding efficiency. Development of the situation is required.

In order to solve the above-mentioned problems, the present invention has a main object in improving encoding and decoding efficiency by simplifying calculation for quantization and inverse quantization while reflecting human visual characteristics while encoding or decoding an image.

In order to achieve the above object, the present invention, in the quantization apparatus, using dead zone uniform threshold quantization (DZUTQ) to quantize the frequency coefficient of the residual block, but at the position of the frequency coefficient of the residual block Provided is a quantization apparatus characterized by quantization using a quantization table having a corresponding quantization coefficient.

According to another object of the present invention, there is provided a quantization method comprising: a quantization table input step of receiving a quantization table having quantization coefficients according to positions of frequency coefficients; And a quantization calculation step of quantizing the frequency coefficients using dead zone uniform threshold quantization (DZUTQ), but calculating an absolute value and a sign of the quantized frequency level using a quantization table. It provides a quantization method.

According to yet another object of the present invention, in the inverse quantization apparatus, dead frequency uniform threshold quantization (DZUTQ) is used to inverse quantize the frequency coefficients of the quantized residual block, but the frequency of the residual block An inverse quantization apparatus is characterized by inverse quantization using a quantization table having quantization coefficients according to the position of coefficients.

According to still another object of the present invention, there is provided an inverse quantization method comprising: a quantization table input step of receiving a quantization table having quantization coefficients according to positions of frequency coefficients; And inverse quantization of the frequency coefficients of the quantized residual block using Dead Zone Uniform Threshold Quantization (DZUTQ), and inverse quantization by inverse quantizing the absolute value and sign of the quantized frequency level using a quantization table. And an inverse quantization calculation step of calculating the calculated frequency coefficients.

According to still another object of the present invention, there is provided an apparatus for encoding an image, comprising: a prediction unit configured to predict a current block of an image and generate a prediction block; A subtraction unit for generating a residual block by subtracting the prediction block from the current block; A transformer for converting the residual block into the frequency domain; A quantization unit for quantizing the transformed residual block using Dead Zone Uniform Threshold Quantization (DZUTQ) and quantizing the quantization table using a quantization table having a quantization coefficient according to the position of the frequency coefficient of the residual block; And an encoder to generate a bitstream by encoding the quantized residual block.

According to still another object of the present invention, there is provided an apparatus for decoding an image, comprising: a decoder configured to decode a bitstream to extract a residual block; An inverse quantizer which inversely quantizes the extracted residual block using dead zone uniform threshold quantization (DZUTQ), and inversely quantizes the quantization table having a quantization coefficient according to the position of the frequency coefficient of the residual block. ; An inverse transform unit for inversely transforming an inverse quantized residual block into a time domain; A prediction unit for predicting a current block to generate a prediction block; And an adder configured to reconstruct the current block by adding the inverse transformed residual block to the prediction block.

As described above, according to the present invention, in encoding or decoding an image, encoding and decoding efficiency may be improved by simplifying calculation for quantization and inverse quantization while reflecting human visual characteristics.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to the components of each drawing, it should be noted that the same reference numerals are designated as much as possible even if displayed on different drawings. In addition, in describing the present invention, when it is determined that the detailed description of the related well-known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

1 is a diagram illustrating frames constituting an image.

There are various techniques for encoding an image. Typically, an image may be divided into pixels and encoded in pixels, or an image may be divided into blocks and encoded in blocks. In the case of encoding in block units, one frame to be encoded among several frames constituting the image may be divided into a macroblock and a subblock constituting the macroblock, and are predicted in units of divided blocks and encoding is performed.

In general, since an image is composed of 30 frames in one second, it is very difficult to distinguish the difference with a human eye because the difference in the image between frames is small between one frame and neighboring frames. Therefore, if the image is output in 30 frames in one second, the human recognizes that each frame is continuous.

Therefore, if the image of the previous frame and the current frame is similar, the unknown pixel value of the current frame can be predicted from the pixel values of the known image constituting the previous frame. Such a prediction technique is called inter prediction. Such inter prediction is based on a motion prediction technique. Motion prediction is performed by referring to a previous frame or to both a previous frame and a future frame based on the time axis. A frame referred to to encode or decode a current frame is called a reference frame.

Referring to FIG. 1, an image is composed of a series of still images. These still pictures are divided in units of group of pictures (GOP). Each still image is called a frame. One picture group includes an I frame 110, a P frame 120, and a B frame 130. The I frame 110 is a frame that is encoded by itself without using a reference frame, and the P frame 120 and the B frame 130 are frames that are encoded by performing motion estimation and compensation using the reference frame. In particular, the B frame 130 is a frame that is predicted and encoded in a forward and a reverse direction, that is, a bidirectional direction, for a past frame and a future frame, respectively.

In addition, the next pixel may be predicted using the correlation between the pixels in the same frame, which is called intra prediction. In the case of intra prediction, the pixel value of each pixel of the current block is predicted using neighboring pixels of a neighboring block that is previously encoded and is in the vicinity of the current block according to a prediction direction determined by an optimal prediction mode to be encoded.

2 is a block diagram schematically illustrating an electronic configuration of a video encoding apparatus according to an embodiment of the present invention.

The image encoding apparatus 200 according to an embodiment of the present invention is an apparatus for encoding an image, and includes a predictor 210, a subtractor 220, a transformer 230, a quantizer 240, and an encoder 250. It may be configured to include).

The video encoding apparatus 200 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). And a communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for encoding an image, and a program. It means a variety of devices having a microprocessor, etc. for execution and operation and control.

The predictor 210 predicts a current block of an image and generates a predicted block. That is, the prediction unit 210 predicts the current block to be encoded in the image by using intra prediction or inter prediction, thereby predicting the predicted pixel value by each pixel. A predicted block having a pixel value of is generated.

The subtractor 220 subtracts the prediction block from the current block to generate a residual block. That is, the subtractor 220 generates a residual block having a residual signal by calculating a difference value between an original pixel value of each pixel of the current block and a predicted pixel value of each pixel of the prediction block. do.

The converter 230 converts the residual block into the frequency domain. That is, the transformer 130 converts the residual block generated by the subtractor 220 into the frequency domain to generate a residual block having a frequency coefficient. The transform unit 330 may convert the residual signal into a frequency using various transform techniques such as a Hadamard transform and a discrete cosine transform based transform. It can be converted into a domain, and the residual signal converted into the frequency domain becomes a frequency coefficient.

The quantization unit 240 quantizes the residual block having the residual signal converted into the frequency domain by the transform unit 230. Here, the quantization unit 240 quantizes the transformed residual block using Dead Zone Uniform Threshold Quantization (DZUTQ), quantized according to the position of the frequency coefficient of the residual block. Quantization is performed using a quantization table having coefficients (QPs). A process of quantizing the residual block by the quantization unit 240 will be described in detail with reference to FIG. 3 in a later process.

The encoder 250 generates a bitstream by encoding the frequency coefficients of the residual block quantized by the quantizer 240. As the encoding technique, an entropy encoding technique may be used, but various encoding techniques may be used without being limited thereto.

3 is a block diagram schematically illustrating an electronic configuration of a quantization device according to an embodiment of the present invention.

The quantization apparatus according to an embodiment of the present invention may be implemented as the quantization unit 240 in the image encoding apparatus 200 described above with reference to FIG. 2. Hereinafter, for convenience of description, the quantization apparatus according to an embodiment of the present invention is called a quantization unit 240.

As described above, the quantization unit 240 quantizes the frequency coefficients of the residual block using DZUTQ, but uses a quantization table having a quantization coefficient according to the position of the frequency coefficients of the residual block. Quantize.

In H.264, a quantization coefficient is set for each macroblock to control a quantization step (Qstep). In H.264, the quantization coefficients may have a value from 0 to 51, and a quantization step other than '0' according to each quantization coefficient is fixed to a constant value. For example, if the quantization coefficient is '0', the quantization step has a value of 0.625, and each time the quantization coefficient is increased by 6, the quantization step is doubled. When the quantization coefficient is 51, the quantization step is a value of 224. Has As such, by setting the quantization coefficients in various ways, the quantization step may be adjusted to control optimization of bit rate and image quality for encoding.

DZUTQ refers to a method in which the quantization step except for the dead zone in which the quantization level becomes '0' is quantized while keeping the other quantization step constant. The quantization coefficients can be changed and set for each macroblock. In the case of DZUTQ, the frequency coefficients are quantized using quantization coefficients set to the same value regardless of the position of the frequency coefficients.

For example, when the subtraction block 220 generates the residual block 'X' and transmits the residual block 'X' to the conversion unit 230, the conversion unit 230 transmits each residual signal of the input residual block 'X' as shown in Equation 1 below. A 'core' transform is used to transform the frequency domain to generate a frequency coefficient W _{i, j} and transmit the generated frequency coefficient to the quantizer 240.

W _{i, j} = C _f XC _f ^T

Here, C _f represents a matrix of forward integer transformations, and C _f ^T means that the matrix is transposed.

The quantization unit 240 post-scales and quantizes the frequency coefficient W _{i, j} input from the transform unit 230. The absolute value of the quantized frequency level generated through such quantization (| Z _{i, j} |) And the sign (sign (Z _{i, j} )) are calculated as in Equation 2.

| Z _{i, j} | = (W _{i, j} | × MF _{i, j} (QP% 6) + f) >> (15 + (QP / 6))

sign (Z _{i, j} ) = sign (W _{i, j} )

Here, W _{i, j} refers to a frequency coefficient according to each position (i, j) of the frequency coefficient matrix W, MF _{i, j} (QP% 6) refers to a multiplication factor, and post in frequency conversion A fixed value combined with a scaling factor and a quantization step is fixed as shown in Table 1 according to the value of QP% 6. In addition, f is a factor that determines the size of errors and dead zones due to rounding in the quantization process, and is 2 ^{(15+ (QP / 6)) / 3} when the prediction block is predicted by intra prediction. In case of prediction, it is fixed as 2 ^{15 + ((QP / 6) / 6)} . In addition, QP% 6 refers to the remainder obtained by dividing the quantization coefficient by 6, QP / 6 refers to the quotient obtained by dividing the quantization coefficient by 6, and '>>' denotes a binary shift right.

Meanwhile, the frequency coefficient of the residual block may be quantized by using a quantization method using a quantization weighted matrix instead of DZUTQ. In the quantization method using the quantization weight matrix, quantization is performed by applying a quantization step that is relatively larger than the frequency coefficient of the low frequency component to the frequency coefficient of the high frequency component in consideration of the visual characteristics of the human. That is, the human eye quantizes the high frequency component less delicately than the low frequency component by using the characteristic that the human eye is sensitive to the image quality degradation of the low frequency component but not the image quality degradation of the high frequency component.

The quantization weighting matrix can be represented as shown in Table 2. Table 2 shows that the quantization coefficients of the quantization weighting matrix are small in the low frequency components sensitive to image quality degradation, and that the quantization coefficients of the quantization weighting matrix are large in the high frequency components which are not sensitive to image quality degradation. It can be seen that.

The process of quantization using the quantization weight matrix is described later. For example, when the subtraction block 220 generates the residual block 'X' and transmits the residual block 'X' to the conversion unit 230, the conversion unit 230 transmits each residual signal of the input residual block 'X' as shown in Equation 1 below. The core is transformed into a frequency domain using a 'core' transform, and a frequency coefficient W _{i, j} is generated and transferred to the quantization unit 240.

The quantization unit 240 post-scales and quantizes the frequency coefficient W _{i, j} input from the conversion unit 230. The absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) is calculated as in Equation 3.

| Z _{i, j} | = (W _{i, j} | × MF _{i, j} (QP% 6) / P _{i, j} + f) >> (15 + (QP / 6))

sign (Z _{i, j} ) = sign (W _{i, j} )

Here, W _{i, j} , MF _{i, j} (QP% 6), f, QP% 6, QP / 6 and '>>' are as described above through Equation 2, and P _{i, j} is the quantization weight ( Quantization Weighted Value). The quantization weight refers to the ratio of each quantization coefficient to the basic quantization coefficient in the quantization weight matrix. In the quantization weight matrix shown in Table 2, the weight is 1 when the quantization coefficient is 16 and the weight is 1.5 when the quantization coefficient is 24.

Although the quantization weight matrix can be used to quantize the frequency coefficient to reflect the visual characteristics of a person, the operation of dividing the multiplication factor MF _{i, j} (QP% 6) by the quantization weight P _{i, j} as shown in Equation 3. Because of the additional computation, the computational complexity for quantization increases, which not only complicates hardware implementation but also reduces coding speed. The inefficiency of the quantization method using the quantization weight matrix is more prominent in the inverse quantization process which will be described later.

Using DZUTQ, frequency can be used to quickly perform quantization calculations similar to DZUTQ's calculation process while taking advantage of quantization by reflecting the visual characteristics of humans that can be achieved through quantization using quantization weight matrix. The coefficients may be quantized but quantized using a quantization table having quantization coefficients according to positions of frequency coefficients. Here, the quantization table having quantization coefficients according to the positions of the frequency coefficients may be generalized as shown in Table 3.

As shown in Table 3, different quantization coefficients are arranged in the quantization table according to the position of the frequency coefficients, and small quantization coefficients are arranged on the low frequency component where a human is sensitive to image degradation, and a human is not sensitive to image degradation. A large quantization coefficient is arranged on the side of the high frequency component that reacts. In Table 3, QP represents the quantization coefficient of the macroblock, and the number in () represents the order of the ZigZag Scan.

The quantization coefficients that scanned the quantization table in Table 3 by zigzag scan are "QP (1), QP (2), QP (3), QP + 1 (4), QP + 1 (5), QP + 1 (6) , QP + 1 (7), QP + 1 (8), QP + 1 (9), QP + 1 (10), QP + 2 (11), QP + 2 (12), QP + 2 (13), QP + 3 (14), QP + 3 (15), and QP + 3 (16) "are listed in this order.

When encoding such a quantization table, the QP of the DC component is encoded by only X of QP + X without encoding as the quantization coefficient of the macroblock, so that "0 (2), 0 (3), 1 (4), 1 (5)" is encoded. ), 1 (6), 1 (7) 1 (8), 1 (9), 1 (10), 2 (11), 2 (12), 2 (13), 3 (14), 3 (15) , 3 (16) ". In addition, when coded by differential pulse coded modulation (DPCM), "0 (2), 0 (3), 1 (4), 0 (5), 0 (6), 0 (7) 0. (8), 0 (9), 0 (10), 1 (11), 0 (12), 0 (13), 1 (14), 0 (15), 0 (16) ". . At this time, the order of () is not encoded. The quantization table is transmitted from the image encoding apparatus 200 to the image decoding apparatus to be described later in the form of a sequence parameter or a picture parameter, or fixedly used by the image encoding apparatus 200 and the image decoding apparatus. Can be.

If QP is 25 in the quantization table shown in Table 3, the quantization table may be represented as shown in Table 4.

Quantization using DZUTQ, but quantization using the above-described quantization table will be described later. For example, when the subtraction block 220 generates the residual block 'X' and transmits the residual block 'X' to the conversion unit 230, the conversion unit 230 transmits each residual signal of the input residual block 'X' as shown in Equation 1 below. The core is transformed into a frequency domain using a 'Core' transform, and a frequency coefficient W _{i, j} is generated and transmitted to the quantization unit 240.

The quantization unit 240 post-scales and quantizes the frequency coefficient W _{i, j} input from the conversion unit 230. The absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) is calculated as in Equation 4.

| Z _{i, j} | = (W _{i, j} | × MF _{i, j} (QP _{i, j} % 6) + f) >> (15 + (QP _{i, j} / 6))

sign (Z _{i, j} ) = sign (W _{i, j} )

Here, W _{i, j} , f and '>>' are as described above through Equation 2, and MF _{i, j} (QP _{i, j} % 6) is a multiplication factor determined according to each quantization coefficient of the quantization table. , Using the multiplication factor table shown in Table 1. QP _{i, j} % 6 is the remainder of dividing the quantization coefficients (QP) for each position in the quantization table by 6, and QP _{i, j} / 6 is the quotient of quantization coefficients for each position in the quantization table divided by 6. .

As described above, quantization using DZUTQ, but quantization using a quantization table having quantization coefficients according to the position of frequency coefficients, enables quantization while taking into account human visual characteristics.

To this end, the quantization unit 240 according to an embodiment of the present invention may include a quantization table input unit 310 and a quantization operator 320.

The quantization table input unit 310 receives the quantization table described above. The quantization operation unit 320 quantizes the frequency coefficients using DZUTQ, but obtains the multiplication coefficient value according to the quantization coefficients of the quantization table input from the multiplication coefficient table as shown in Table 1 to obtain an absolute value of the quantized frequency level as shown in Equation 4. Calculate the and sign. That is, the quantization operator 320 post-scales and quantizes the frequency coefficients W _{i, j} and quantizes them using Equation 4 using DZUTQ.

4 is a flowchart illustrating a quantization method according to an embodiment of the present invention.

The image encoding apparatus 200 according to an embodiment of the present invention described above with reference to FIG. 2 divides an image into macroblocks or subblocks to encode an image, and uses a prediction technique such as inter prediction or intra prediction. The current block of the image is predicted, and a residual block, which is a difference between the current block and the prediction block, is generated and converted into a frequency domain. When the residual block is transformed into the frequency domain, each residual signal of the residual block is output as a frequency coefficient and quantized.

The quantization unit 240 quantizes the frequency coefficients using DZUTQ, and quantizes them using a quantization table having quantization coefficients according to the positions of the frequency coefficients. To this end, the quantization unit 240 receives a quantization table having a quantization coefficient according to the position of the frequency coefficient (S410). At this time, the value QP of the quantization coefficient of the macroblock may be input together. That is, when a fixed QP value is not used, it can be input. Thereafter, when the frequency coefficient W _{i, j} is input from the converter 230, the quantization unit 240 quantizes the frequency coefficient as shown in Equation 4 to calculate the absolute value and sign of the quantized frequency level (S420).

In this case, a simple calculation such as DZUTQ is used to quickly quantize, while calculating the absolute value and sign of the quantized frequency level using a quantization table having a quantization coefficient according to the position of the frequency coefficient. Considered quantization can be performed.

As described above, the image encoded in the bitstream by the video encoding apparatus 200 is real-time or non-real-time through the wired or wireless communication network such as the Internet, local area wireless communication network, wireless LAN network, WiBro network, mobile communication network or the like, The image decoding apparatus may be transmitted to a video decoding apparatus through a communication interface such as a universal serial bus (USB), decoded by a video decoding apparatus to be described later, and restored and reproduced.

5 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus according to an embodiment of the present invention.

The image decoding apparatus 500 according to an embodiment of the present invention is an apparatus for decoding an image, and includes a decoder 510, an inverse quantizer 520, an inverse transformer 530, a predictor 540, and an adder. And 550.

The video decoding apparatus 500, like the video encoding apparatus 200 described above with reference to FIG. 2, may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player ( PMP: Portable Multimedia Player (PSP), PlayStation Portable (PSP: PlayStation Portable), Mobile Communication Terminal (Mobile Communication Terminal) and the like, communication devices, such as a communication modem for performing communication with various devices or wired and wireless communication network, decoding the image Means a variety of devices including a memory for storing various programs and data, a microprocessor for executing and controlling the program by executing.

The decoder 510 extracts a residual block by decoding the bitstream. That is, the decoder 510 decodes a bitstream, which is an image encoded by the image encoding apparatus 300, and extracts a residual block including pixel information of the current block of the image. Residual signals of the residual block at this time have an absolute value and a sign of the quantized frequency level.

The inverse quantizer 520 de-quantizes the residual block extracted from the bitstream by the decoder 510. In this case, the inverse quantizer 520 inverse quantizes the residual block extracted from the bitstream using DZUTQ, but inversely quantizes the quantization table having a quantization coefficient according to the position of the frequency coefficient of the residual block. A process of inverse quantization of the residual block by the inverse quantization unit 520 will be described in detail with reference to FIG. 6 in a later process.

The inverse transformer 530 inversely transforms the inverse quantized residual block into the time domain. That is, the inverse transformer 530 inversely transforms the frequency coefficient of the residual block inversely quantized by the inverse quantizer 520 into the time domain by using Hadamard transform or DCT based transform.

The prediction unit 540 generates a prediction block by predicting a current block to be decoded. In this case, inter prediction or intra prediction may be used as the prediction technique.

The adder 550 reconstructs the current block by adding the inverse transformed residual block to the prediction block, thereby ultimately reconstructing the original image.

6 is a block diagram schematically illustrating an electronic configuration of an inverse quantization apparatus according to an embodiment of the present invention.

The inverse quantization apparatus 600 according to an embodiment of the present invention may be implemented by the inverse quantization unit 520 described above with reference to FIG. 5. Therefore, hereinafter, referred to as inverse quantization unit 520 for convenience of explanation.

The inverse quantizer 520 according to an embodiment of the present invention inverse quantizes the frequency coefficients of the residual block quantized using DZUTQ, but uses an inverse quantization table having a quantization coefficient according to the position of the frequency coefficients of the residual block. Quantize.

The process of inverse quantization using DZUTQ is described later. For example, the residual block extracted from the bitstream by the decoder 510 may include an inverse quantizer (Z _{i, j} |) and a sign (sign (Z _{i, j} )) of the quantized frequency level. When inputted to 520, the inverse quantizer 520 prescales and dequantizes the absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) of the quantized frequency level. Calculate the quantized frequency coefficient Z _{i, j} using the absolute value (| Z _{i, j} |) and sign (sign (Z _{i, j} ) of the quantized frequency level, as shown in Equation 5 Z _{i, j} is calculated by using the de-quantized frequency coefficients W _{'i, j.}

Z _{i, j} = sign (Z _{i, j} ) × | Z _{i, j} |

W _{i, j} '= Z _{i, j} × V _{i, j} (QP% 6) << floor (QP / 6)

Here, V _{i, j} (QP% 6) refers to a scaling factor, and is a fixed value combining a quantization step according to a prescaling factor and a quantization coefficient in a frequency inverse transform, and is a table according to the value of QP% 6. It is fixed as 5. In addition, QP% 6 is the remainder of quantization coefficient divided by 6, QP / 6 is the quotient of quantization coefficient divided by 6, and '<<' means binary shift right, and floor is Round down operation.

When the inverse quantized frequency coefficient _{W'i, j} is calculated, the inverse quantizer 520 transfers the inverse quantized frequency coefficient _{W'i, j} to the inverse transformer 530, and the inverse transformer 530 inverse The quantized frequency coefficient W ' _{i, j} is transformed into a time domain by performing an inverse' core 'transform as shown in Equation 6, and as a result, the inverse quantized and inverse transformed residual block is inverse transform unit. It is output from 530.

W = C _i ^T X C _i

Here, C _i represents a matrix of inverse integer transformations, and C _i ^T means that the matrix is transposed.

Meanwhile, an inverse quantized frequency coefficient is obtained by inverse quantizing the absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) of the quantized frequency level using a quantization method using a quantization weight matrix rather than DZUTQ. We can generate W ' _{i, j} . In this case, as in the case of quantization, the quantization weight matrix shown in Table 2 may be used to inverse quantization by reflecting the visual characteristics of a person.

Inverse quantization using the quantization weight matrix is described below. For example, the residual block extracted from the bitstream by the decoder 510 may include an inverse quantizer (Z _{i, j} |) and a sign (sign (Z _{i, j} )) of the quantized frequency level. Inversely, the inverse quantizer 520 prescales and inversely quantizes the absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) of the quantized frequency level. the absolute value of the frequency level of quantization steps (| Z _{i, j} |) and the sign _{(sign (Z i, j)} ) to the calculated quantized frequency coefficient Z _{i, j} and the quantized frequency coefficients using the Z _{i, j} Calculate the inverse quantized frequency coefficient _{W'i, j} using

Z _{i, j} = sign (Z _{i, j} ) × | Z _{i, j} |

W _{i, j} '= Z _{i, j} x P _{i, j} x V _{i, j} (QP% 6) << (floor (QP / 6)-4). If QP / 6 ≥ 4

W _{i, j} '= (Z _{i, j} × P _{i, j} × V _{i, j} (QP% 6) + 1 << (3-QP / 6)) >> 4-QP / 6. If QP / 6 <4

Here, V _{i, j} (QP% 6), QP% 6, QP / 6, floor, '<<' are as described in Equation 5, P _{i, j} in the quantization weight matrix described above in Table 2 The weight refers to the position of each frequency coefficient, and '>>' means a binary right shift operation.

When the inverse quantized frequency coefficient _{W'i, j} is calculated, the inverse quantizer 520 transfers the inverse quantized frequency coefficient _{W'i, j} to the inverse transformer 530, and the inverse transformer 530 inverse As shown in Equation 6 _{, the} quantized frequency coefficient _{W'i, j} is transformed into a time domain by performing an inverse 'core' transform, and as a result, an inverse quantized and inverse transformed residual block is output from the inverse transform unit 530.

Comparing Equation 7, which is the equation for inverse quantization using the quantization weight metric, to Equation 5, which is the equation of inverse quantization using DZUTQ, in inverse quantization using the quantization weight metric, the inverse quantized frequency coefficient W _{i, j} In order to calculate ', not only does the computation multiply by the weights P _{i, j} , but also the complexity of the computation that requires checking whether the value of QP / 6 is greater than or equal to 4 and performing other calculations accordingly. Increases.

As described above, if the quantized frequency coefficient is inversely quantized using the quantization weight matrix, the quantized frequency coefficient may be inversely quantized in consideration of the visual characteristics of a person. This becomes quite complicated, ultimately increasing the complexity of the implementation of hardware for encoding or decoding and lowering the coding speed to reduce the efficiency of compression. Therefore, while it is possible to take into account the visual characteristics of a person such as a quantization weight matrix, a method of reducing the complexity of a calculation such as DZUTQ is efficient. Such a method may be quantized using DZUTQ as described below, but may be inverse quantized using the quantization table described above with reference to Table 3.

For example, the residual block extracted from the bitstream by the decoder 510 may include an inverse quantizer (Z _{i, j} |) and a sign (sign (Z _{i, j} )) of the quantized frequency level. Inversely, the inverse quantizer 520 prescales and inversely quantizes the absolute value (| Z _{i, j} |) and the sign (sign (Z _{i, j} )) of the quantized frequency level. the absolute value of the frequency level of quantization steps (| Z _{i, j} |) and the sign _{(sign (Z i, j)} ) to the calculated quantized frequency coefficient Z _{i, j} and the quantized frequency coefficients using the Z _{i, j} Calculate the inverse quantized frequency coefficient _{W'i, j} using

Z _{i, j} = sign (Z _{i, j} ) × | Z _{i, j} |

W _{i, j} '= Z _{i, j} × V _{i, j} (QP _{i, j} % 6) << floor (QP _{i, j} / 6)

Here, floor, '<<' are as described in Equation 5, V _{i, j} (QP _{i, j} % 6) is a scaling factor determined according to each quantization coefficient of the quantization table, the scaling factor shown in Table 5 You can get it using a table. QP _{i, j} % 6 denotes the remainder of dividing the quantization coefficients for each position in the quantization table by 6, and QP _{i, j} / 6 denotes the quotient of the quantization coefficient for each position in the quantization table divided by 6.

When the inverse quantized frequency coefficient _{W'i, j} is calculated, the inverse quantizer 520 transfers the inverse quantized frequency coefficient _{W'i, j} to the inverse transformer 530, and the inverse transformer 530 inverse As shown in Equation 6 _{, the} quantized frequency coefficient _{W'i, j} is transformed into a time domain by performing an inverse 'core' transform, and as a result, an inverse quantized and inverse transformed residual block is output from the inverse transform unit 530.

Equation 8 is different from Equation 5 in that it uses a quantization coefficient of the quantization table, that is, a quantization coefficient according to a frequency position. Accordingly, the method of inverse quantization using DZUTQ but inverse quantization using a quantization table having quantization coefficients according to the position of the frequency coefficient has almost no difference in computational complexity when compared to the inverse quantization method using DZUTQ. However, since a quantization table having a quantization coefficient according to the position of the frequency coefficient is used, a visual characteristic of a person can be considered.

To this end, the inverse quantization unit 520 according to an embodiment of the present invention may include a quantization table input unit 610 and an inverse quantization operation unit 620.

The quantization table input unit 610 receives the quantization table described above through Table 3. The inverse quantization calculator 620 inverse quantizes the frequency coefficients of the quantized residual block using DZUTQ, and calculates the inverse quantized frequency coefficient by inversely quantizing the absolute value and the sign of the quantized frequency level using a quantization table.

7 is a flowchart illustrating an inverse quantization method according to an embodiment of the present invention.

The image decoding apparatus 500 that receives and stores a bitstream of an image through a wired or wireless communication network or a cable, decodes the image, and restores the image to reproduce the image according to a user's selection or an algorithm of another program being executed.

To this end, the decoder 510 of the image decoding apparatus 500 decodes the bitstream and extracts a residual block representing information about pixel values of the current block of the image. When the residual block is extracted, the inverse quantizer 520 inverse quantizes the quantized frequency coefficients included in the residual block using DZUTQ, and inverse quantizes the quantization table having a quantization coefficient according to the position of the frequency coefficients. To this end, the inverse quantizer 520 receives a quantization table having a quantization coefficient according to the position of the frequency coefficient (S710). At this time, the value QP of the quantization coefficient of the macroblock may be input together. That is, when a fixed QP value is not used, it can be input. Subsequently, when the absolute value and the sign of the quantized frequency level, which are quantized frequency coefficients, are input from the decoder 510, the inverse quantizer 520 inversely quantizes the absolute value and the sign of the frequency level as shown in Equation (8). The quantized frequency coefficient is calculated (S720).

In this case, while inverse quantization is quickly performed using a simple calculation such as DZUTQ, the inverse quantization is calculated using a quantization table having quantization coefficients according to the position of the frequency coefficients. Can be performed.

As described above, the method of quantizing or inverse quantizing using DZUTQ, but quantizing or inverse quantizing using a quantization table having a quantization coefficient according to the position of the frequency coefficient is a method of quantizing or inverse quantizing using DZUTQ. Both the fastness of the calculation and the good quality quantization and inverse quantization of the method of quantizing or inverse quantizing using a quantization weight matrix are possible.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

As described above, the present invention is applied to the field of an apparatus for encoding or decoding an image, thereby generating an effect of improving encoding and decoding efficiency by simplifying calculation for quantization and inverse quantization while reflecting human visual characteristics. It is a very useful invention.

1 is a diagram illustrating frames constituting an image;

2 is a block diagram schematically illustrating an electronic configuration of an image encoding apparatus according to an embodiment of the present invention;

3 is a block diagram schematically illustrating an electronic configuration of a quantization device according to an embodiment of the present invention;

4 is a flowchart illustrating a quantization method according to an embodiment of the present invention;

5 is a block diagram schematically illustrating an electronic configuration of an image decoding apparatus according to an embodiment of the present invention;

6 is a block diagram schematically illustrating an electronic configuration of an inverse quantization apparatus according to an embodiment of the present invention;

7 is a flowchart illustrating an inverse quantization method according to an embodiment of the present invention.

<Description of Symbols for Main Parts of Drawings>

310 and 610: quantization table input unit 320: quantization operation unit

620: inverse quantization operation unit

Claims (8)

In the quantization device, Quantize the frequency coefficients of the residual block using dead zone uniform threshold quantization (DZUTQ), but quantize using a quantization table having quantization coefficients according to the position of the frequency coefficients of the residual block. Quantization device. The method of claim 1, wherein the quantization device, A quantization table input unit to receive the quantization table; A quantization calculator that quantizes the frequency coefficient using the dead zone uniform boundary quantization, and calculates an absolute value and a sign of the quantized frequency level using the quantization table. Quantization device comprising a. In the quantization method, A quantization table input step of receiving a quantization table having quantization coefficients according to positions of frequency coefficients; A quantization calculation step of quantizing the frequency coefficients using dead zone uniform threshold quantization (DZUTQ), and calculating the absolute value and sign of the quantized frequency level using the quantization table. Quantization method comprising a. In inverse quantization device, Inverse quantization of the frequency coefficients of the quantized residual block using Dead Zone Uniform Threshold Quantization (DZUTQ), but inverse quantization using a quantization table having quantization coefficients according to the positions of the frequency coefficients of the residual block. Inverse quantization device, characterized in that. The apparatus of claim 4, wherein the inverse quantization device comprises: A quantization table input unit to receive the quantization table; Inverse quantization of the frequency coefficients of the quantized residual block using the dead zone uniform boundary quantization, and inverse quantization of the absolute value and the sign of the quantized frequency level using the quantization table to calculate an inverse quantized frequency coefficient Quantization calculator Inverse quantization apparatus comprising a. In the inverse quantization method, A quantization table input step of receiving a quantization table having quantization coefficients according to positions of frequency coefficients; And Inverse quantization of frequency coefficients of the quantized residual block using Dead Zone Uniform Threshold Quantization (DZUTQ), and inverse quantization by inverse quantizing the absolute value and sign of the quantized frequency level using the quantization table Quantization calculation step to calculate the calculated frequency coefficients Inverse quantization method comprising a. In the apparatus for encoding a video, A prediction unit for predicting a current block of the image to generate a prediction block; A subtraction unit for generating a residual block by subtracting the prediction block from the current block; A transformer for converting the residual block into a frequency domain; A quantization unit for quantizing the transformed residual block using Dead Zone Uniform Threshold Quantization (DZUTQ) and quantizing the quantization table using a quantization table having a quantization coefficient according to a position of a frequency coefficient of the residual block; And An encoder that encodes the quantized residual block to generate a bitstream An image encoding apparatus comprising a. In the apparatus for decoding an image, A decoder which extracts a residual block by decoding the bitstream; Inverse quantization of the extracted residual block using Dead Zone Uniform Threshold Quantization (DZUTQ), and inverse quantization using a quantization table having a quantization coefficient according to the position of the frequency coefficient of the residual block. Quantization unit; An inverse transformer for inversely transforming the inverse quantized residual block into a time domain; A prediction unit for predicting a current block to generate a prediction block; And An adder configured to add the inverse transformed residual block to the prediction block to restore the current block; Video decoding apparatus comprising a.

Images