JP2946833B2 - Quantization method in image coding - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quantization method in an image coding apparatus using an orthogonal transform.

[0002] At this time, there is a demand for a quantization method in image coding that improves the quality of a transmission image.

[0003]

2. Description of the Related Art FIG. 10 is a block diagram showing a configuration of an example of an image coding apparatus. FIG. 11 is a diagram illustrating a configuration of an example of a block.

FIG. 12 is a diagram showing an example of a zigzag scan. FIG. 13 is a diagram showing an example of the quantization step. FIG. 14 is a flowchart for explaining the operation of the second conventional example.

In FIG. 10, data obtained by digitizing analog image information by an analog / digital converter (not shown) is input to a blocking unit 1 and divided into a predetermined number of blocks for each frame. In this case, it is assumed that each block is composed of 4 × 4 pixels each in the vertical and horizontal directions as shown in FIG. The data of this block is applied to one input terminal of the subtraction unit 2, the image data of the same block position one frame before stored in the frame memory 9 is applied to the other input terminal, and the subtraction unit 2 subtracts the two input data. To obtain the difference between the two and output the orthogonal transform
Add to

The orthogonal transform unit 3 performs an orthogonal transform by a discrete cosine transform (DCT) on a block composed of 4 × 4 pixels as shown in FIG.
The pixel data is converted into frequency component data and a conversion coefficient is output. This transform coefficient is added to the quantization unit 4, and the quantization unit 4 performs a zigzag scan on each transform coefficient of the block as shown in FIG. The conversion coefficient after the zigzag scan is denoted by Ci.

As a first conventional example, a quantization unit 4 quantizes a transform coefficient Ci determined to be insignificant so as to assign a large amount of information to important information to zero. Then, this output is output to the variable length coding unit 5.
And the inverse orthogonal transform unit 7. Here, an example is shown in which interframe coding and orthogonal transform are combined as orthogonal transform coding, but the coding method is not limited to this.

Next, a second conventional example will be described below. Table 1 is a table showing the relationship between the input transform coefficient Ci of the quantizer 4 and the quantized output Qi.

[0009]

[Table 1]

In Table 1, the quantization step g shown in FIG. In addition, the width of the area where the quantized output Qi is 0, that is, the width of the dead zone T is changed from T = g to the maximum T (= TMA) by the number of continuous quantized outputs Qi = 0.
X) = 1.5 × g (see FIG. 14). And
A zigzag scan is performed in the block as shown in FIG.

(1) When i = 0, the input conversion coefficient Co = 4
9. Since the dead zone T = 32, Co = 49> T = 3
2, and 49 <2T = 64, and therefore, as the quantized output Qi, an intermediate value Qo = 48 between 32 and 2 × 32 = 64 is output (FIG.
In FIG. 14, ABS (Ci) is Ci.
Represents the absolute value of).

(2) When i = 1, C₁= 0, T = 32
From C₁= 0 <T = 32, thus the quantized output Q₁=
Outputs 0. (3) When i = 2, C_Two= 0. The previous quantized output, i.
Q when = 1₁= 0, the width of dead zone T is increased by one
Add. Therefore, T = 32 + 1 = 33, and C _Two= 0 <
T = 33 and therefore the quantized output Q_Two= 0 is output (Fig.
14, see).

(4) When i = 3, C ₃ = 0. Since the immediately preceding quantized output Q ₂ = 0, the width of T is further increased by one. Therefore, T = 33 + 1 = 34, and C ₃ = 0 <T = 34, and therefore a quantized output Q ₃ = 0 is output (see FIG. 14).

(5) When i = 4, C ₄ = 34. (4) mentioned above
In the same manner as in the above case, since the immediately preceding quantized output Q ₃ = 0, T
Is further increased by one. Therefore, T = 34 + 1 = 35, and C ₄ = 34 <T = 35, so that the quantized output Q ₄ =
0 is output (see FIG. 14).

(6) When i = 5, C ₅ = 35. Since the immediately preceding quantized output Q ₄ = 0, the width of T is further increased by one. Therefore, T = 35 + 1 = 36, and C ₅ = 35 <T = 36, and the quantized output Q ₅ = 0 is output (see FIG. 14).

(7) When i = 6, C ₆ = 0. Since the immediately preceding quantized output Q ₅ = 0, the width of T is further increased by one. Therefore, T = 36 + 1 = 37, and C ₆ = 0 <T = 37, and thus a quantized output Q ₆ = 0 is output (see FIG. 14).

(8) When i = 7, C₇= 46. Last quantization
Output Q₆Since = 0, the width of T is further increased by one. others
Therefore, T = 37 + 1 = 38, and C₇= 46> T = 38 and C
₇= 46 <2T = 76 and therefore the quantized output Q₇as
Q of the intermediate value between 32 and 2 × 32 = 64 ₇= 48 is output (Fig. 14
, See).

(9) When i = 8, C ₈ = 34. For quantized output Q ₇ of the immediately preceding non-zero, the width of the dead zone T is returned to the initial g = 32. Therefore, C ₈ = 34> T = 32 and C ₈
= 34 <2T = 64, therefore as the quantization output Q ₈ 32
And an intermediate value Q ₈ = 48 of 2 × 32 = 64 (see FIG. 14).

(10) When i = 9, C ₉ = 0. For quantized output Q ₈ of the immediately preceding non-0, T = 32. C ₉ = 0 <T = 32,
Therefore, a quantized output Q ₉ = 0 is output (FIG. 14,
reference).

In this way, the width of the dead zone T is increased by one each time the quantized output Qi outputs 0. In this case, the maximum value TMAX of T is T
MAX = 1.5 × g (see FIG. 14).

[0021]

In consideration of highly efficient encoding of image information and transmission at a low transmission rate (for example, 64 Kb / s), a large amount of information is allocated to blocks containing important information. Transmission leads to an improvement in the quality of the transmitted image.

However, in the above-described quantization method of the first conventional example, since a fixed step size g is used as the width of the dead zone T, even information such as noise in the high frequency region of the transform coefficient Ci remains unchanged. There is a problem in that the image is transmitted and the quality of the transmitted image is reduced.

In the second conventional example, the maximum value of the dead zone T is set to a constant value (TMAX = 1.5 × g), so that the high frequency of the conversion coefficient Ci is the same as in the first conventional example. There is a problem that information such as noise is transmitted as it is in the area, and the quality of the transmitted image is degraded.

Accordingly, it is an object of the present invention to provide a quantization method in image coding that improves the quality of a transmission image.

[0025]

The above problem is solved by the circuit configuration shown in FIG. 1, FIG. 2, or FIG.

In FIG. 1 showing the configuration of the first invention, 1
An orthogonal transform encoder (300) that receives image data for each frame, divides the data into blocks, performs encoding using orthogonal transform, and outputs a transform coefficient represented by a frequency component; and The quantization unit (40) receives the transform coefficient of the output of (300), performs quantization at a predetermined quantization step width, and outputs the result.
0), 450 is provided in the quantization unit 400, and the quantization output of the quantization unit 400 is 0.
Is a first weighting control unit that weights the quantization step width of the area and changes the weighting coefficient according to the predetermined frequency area of the transform coefficient.

In FIG. 2 showing the configuration of the second invention, 1
An orthogonal transform encoder (300) that receives image data for each frame, divides the data into blocks, performs encoding using orthogonal transform, and outputs a transform coefficient represented by a frequency component; and The quantization unit (40) receives the transform coefficient of the output of (300), performs quantization at a predetermined quantization step width, and outputs the result.
0), 460 is provided in the quantization unit 400, compares the transform coefficient in a predetermined frequency domain with a reference value, and when there is a transform coefficient larger than the reference value, The quantization step width of the region where the quantized output of the unit 400 is 0 is weighted by a first weighting coefficient, and when all the transform coefficients in a predetermined frequency region are smaller than the reference value, the quantization unit 400 quantization output is 0
Is a second weighting control unit that weights the quantization step width of the region with a second weighting coefficient.

In FIG. 3 showing the configuration of the third invention, 1
An orthogonal transform encoder (300) that receives image data for each frame, divides the data into blocks, performs encoding using orthogonal transform, and outputs a transform coefficient represented by a frequency component; and The quantization unit (40) receives the transform coefficient of the output of (300), performs quantization at a predetermined quantization step width, and outputs the result.
0), a third weight control unit 470 is provided in the quantization unit 400 and weights the maximum value of the quantization step width in a region where the quantization output is 0. is there.

[0029]

In FIG. 1 showing the structure of the first invention, the first weighting control section 450 sets the quantization output of the quantization section 400 to a quantization step width corresponding to a region of 0 (so-called dead zone). Weighting is applied to this. Then, the weighting coefficient for quantizing the transform coefficient on the high frequency side is set to a large value to weight the dend zone.

As a result, by quantizing the transform coefficient on the high frequency side to zero, it becomes possible to allocate a large amount of information to important information existing on the relatively low frequency side.
In FIG. 2 showing the configuration of the second invention, a second weighting control section 460 compares a transform coefficient in a predetermined frequency region (for example, a low frequency region) with a reference value (for example, a first quantization step as a reference). I do. When at least one transform coefficient larger than the reference value exists in the low frequency region, it is determined that important information exists, and the quantization output of the quantizing unit 400 is in a region where the quantization output is 0 (so-called dead zone). The quantization step width is weighted by a first weighting coefficient.

When all the transform coefficients in the low frequency area are smaller than the reference value, it is determined that there is not very important information, and the quantization step width of the dead zone is determined by the first weighting coefficient. Weighting is performed using a large second weighting coefficient.

As a result, by quantizing the transform coefficient on the high frequency side to 0, it becomes possible to allocate a large amount of information to important information existing on the relatively low frequency side.
In FIG. 3 showing the configuration of the third invention, the third weighting control section 470 weights the maximum value of the quantization step width in a region where the quantization output is 0 (dead zone). Then, the weighting coefficient is set to a value larger than the transform coefficient on the high frequency side.

As a result, by quantizing the transform coefficient on the high frequency side to 0, it becomes possible to allocate a large amount of information to important information existing on the relatively low frequency side.

[0034]

FIG. 4 is a diagram for explaining the operation of the embodiment of the first invention. FIG. 5 is a flowchart for explaining the operation of the embodiment of the first invention.

FIG. 6 is a diagram for explaining the operation of the embodiment of the second invention. FIG. 7 is a flowchart (part 1) for explaining the operation of the embodiment of the second invention.

FIG. 8 is a flowchart (part 2) for explaining the operation of the embodiment of the second invention. FIG. 9 is a flowchart for explaining the operation of the embodiment of the third invention.

The same reference numerals indicate the same objects throughout the drawings. First, an embodiment of the first invention will be described. FIG.
, The transform coefficient Ci (for example, i
= 0 to 15), it has been empirically found that the conversion coefficient on the low frequency side where the i value is small generally contains important information in many cases. Therefore, for the purpose of quantizing the transform coefficient on the high frequency side where the proportion of important information is small (i.e., the i value is large) to 0 as much as possible, the transform coefficient corresponding to the position (or frequency band) of the transform coefficient The so-called dead zone in the region where the quantized output Qi is 0 shown in FIG. 13 is weighted.

For example, in a 4 × 4 block as shown in FIG. 12, a weight matrix as shown in FIG. 4A is defined for the quantization step g. That is, when i = 0 to 2 for g, the weighting coefficient =
1. When i = 3 to 5, weighting factor = W1, and when i = 6 to 15, weighting factor = W2. Then, the width of the dead zone where the quantized output Qi becomes 0 is as shown in FIG. In this case, 1 ≦ W1 <W2, for example, W1 = 1.
4, W2 = 2.0.

In the quantization unit 4 shown in FIG. 10, the transform coefficients Ci (i = 0 to 1) in one block input from the orthogonal transform unit 3
(1) When i = 0 to 2 (see FIG. 5),
The absolute value of i, ABS (Ci), is compared with the quantization step g (see FIG. 5). When ABS (Ci) <g, the quantized output Qi is set to 0 (see FIG. 5). or,
When ABS (Ci) <g, the software
Alternatively, a value such as Qi = 1.5 g or 2.5 g as shown in FIG. 13 is output from a read-only memory (hereinafter referred to as a ROM, not shown) or the like in accordance with the input Ci value (see FIG. 5). ).

(2) When i = 3 to 5 (see FIG. 5), a value obtained by adding the weight coefficient W1 to the quantization step g, that is, W1 ×
g = 1.4 × g is compared with ABS (Ci) (see FIG. 5), and when ABS (Ci) <W1 × g = 1.4 × g, the quantized output Qi = 0 (see FIG. 5). . Also, A
When BS (Ci) <1.4 × g is not satisfied, a value such as Qi = 1.5 g or 2.5 g as shown in FIG. 13 is output as shown in FIG. (See FIG. 5).

(3) When i = 6 to 15, a value obtained by adding a weight coefficient W2 to the quantization step g, that is, W2 × g = 2.0 × g and A
Compare with BS (Ci) (see FIG. 5). And A
When BS (Ci) <W2 × g = 2.0 × g, the quantized output Q
Set i = 0 (see FIG. 5). ABS (Ci) <2.
When it is not 0 × g, Qi = 1.5 as shown in FIG. 13 corresponding to the input value of Ci as in the case of (1) described above.
g or 2.5 g is output (see FIG. 5).

In this way, by assigning weights to the dead zone T corresponding to the position (or frequency band) of the transform coefficient Ci, a large amount of information can be allocated to important information on the low frequency side. As a result, the quality of the transmitted image can be improved.

Next, an embodiment of the second invention will be described. As in the case of the above-described first embodiment of the present invention, the transform coefficient for the purpose of quantizing the transform coefficient on the high-frequency side, which contains a small percentage of important information (i.e., the i-value is large), to 0 as much as possible is used. The so-called dead zone in the region where the quantized output Qi is 0 shown in FIG. 13 is weighted in accordance with the position of the coefficient (or the frequency band). However, when there are significant coefficients in the conversion coefficients on the low frequency side (for example, Co, C ₁ , C ₂ ), that is, when Ci (i = 0, 1, 2)> g, the weight is not so much applied, When there is no significant coefficient in, the weight is controlled to be tight.

That is, when any _{one of} Co, C ₁ and C ₂ is significant, a weight of weight coefficient = W1 is assigned to g of i = 3 to 15. When Co, C ₁ and C ₂ are not significant, i = 3
A weight of weight coefficient = W2 is assigned to g of 1515. Then, the width of the dead zone where the quantized output Qi becomes 0 is as shown in FIGS. 6A and 6B, respectively. In this case, 1 <
W1 <W2, for example, W1 = 1.4 and W2 = 2.0.

In the quantization unit 4 shown in FIG. 10, the transform coefficient Ci (i = 0 to 1) in one block input from the orthogonal transform unit 3
(1) When i = 0 to 2, the absolute value of i, AB
Compare S (Ci) with the quantization step g (see FIG. 7). When ABS (Ci) <g, the quantized output Q
Set i = 0 (see FIG. 7). Also, ABS (Ci) <g
Otherwise, that is, when any of Co, C ₁ and C ₂ is significant (see FIG. 7), by software or RO
M (not shown), etc., corresponding to the input Ci value
A value such as Qi = 1.5 g or 2.5 g as shown in FIG. 13 is output (see FIG. 7).

(2) When i = 3 to 15, and when Co, C ₁ ,
When any of C ₂ is significant, as described above, ABS
(Ci) is compared with the value of W1 × g = 1.4 × g (see FIG. 8). Then, ABS (Ci) <W1 × g = 1.4
In the case of xg, the quantized output Qi is set to 0 (see FIG. 8). When ABS (Ci) <1.4 × g is not satisfied, Qi = 1.5 as shown in FIG. 13 corresponding to the input Ci value by software or ROM (not shown).
g or 2.5 g is output (see FIG. 7).

(3) When i = 3 to 15, Co, C ₁ , C ₂
When both are not significant, ABS (Ci) is compared with the value of W2 × g = 2.0 × g as described above (see FIG. 8). Then, when ABS (Ci) <W2 × g = 2.0 × g, the quantized output Qi = 0 (see FIG. 8). or,
When ABS (Ci) <2.0 × g is not satisfied, Qi = 1.5 g as shown in FIG. 13 corresponding to the input Ci value by software or ROM (not shown), or
A value such as 2.5 g is output (see FIG. 7).

In this way, the dead zone T is weighted in accordance with the position (or frequency band) of the transform coefficient Ci. If there is a significant coefficient on the low frequency side, the weight is not much applied, When there is no significant coefficient on the frequency side, a large amount of information can be allocated to important information on the low frequency side by controlling the weight to be tight. As a result, the quality of the transmitted image can be improved.

Next, an embodiment of the third invention will be described. The third invention is made to solve the problem in the second conventional example, and is characterized in that the quantization unit 4 weights the maximum value TMAX of the dead zone T.

That is, TMAX = WM for TMAX
(I) x g. As the weight matrix WM (i), a value similar to that of the above-described embodiment of the first invention (see FIG. 4A) is given. That is, WM (0) to WM (2) = 1, WM
(3) -WM (5) = 1.4, WM (6) -WM (15) =
2.0.

(1) The absolute value of Ci, ABS (Ci), and T are compared (see FIG. 9). (2) When ABS (Ci) <T, the quantized output Qi = 0, and the width of the dead zone T is T = MIN (T + 1,
TMAX), that is, T + 1 and TMAX increased by one.
= WM (i) × g The smaller value is adopted as the value of T (see FIG. 9).

(3) When ABS (Ci) <T is not satisfied,
Qi = 1.1 as shown in FIG. 13 corresponding to the input value of Ci by software or ROM (not shown).
Output a value such as 5 g or 2.5 g. Then, the value of T is returned to T = g (see FIG. 9).

In this way, TMAX is increased with respect to Ci on the high frequency side, and the transform coefficient on the high frequency side is quantized to zero. As a result, a large amount of information can be allocated to important information on the low frequency side, and as a result, the quality of a transmitted image can be improved.

[0054]

As described above, according to the present invention, the conversion coefficient on the high frequency side determined to be visually insignificant is set to 0.
Quantization allows a large amount of information to be allocated to important information. As a result, the quality of the transmitted image can be improved.

[Brief description of the drawings]

FIG. 1 is a principle diagram of the first invention,

FIG. 2 is a principle diagram of the second invention,

FIG. 3 is a principle diagram of the third invention,

FIG. 4 is a diagram for explaining the operation of the embodiment of the first invention;

FIG. 5 is a flowchart for explaining the operation of the embodiment of the first invention;

FIG. 6 is a diagram for explaining the operation of the embodiment of the second invention;

FIG. 7 is a flowchart (part 1) for explaining the operation of the embodiment of the second invention;

FIG. 8 is a flowchart (part 2) for explaining the operation of the embodiment of the second invention;

FIG. 9 is a flowchart for explaining the operation of the embodiment of the third invention;

FIG. 10 is a block diagram illustrating a configuration of an example of an image encoding device.

FIG. 11 is a diagram illustrating a configuration of an example of a block;

FIG. 12 is a diagram showing an example zigzag scan;

FIG. 13 is a diagram showing an example of a quantization step;

FIG. 14 is a flowchart for explaining the operation of the second conventional example.

[Explanation of symbols]

450 denotes a first weight control unit, 460 denotes a second weight control unit, and 470 denotes a third weight control unit.

────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Kiichi Matsuda 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Prefecture Inside Fujitsu Limited (56) References Proceedings of the 70th anniversary of IEICE, 1137 Determination of the range of coefficients to be quantized in hybrid coding ”(58) Fields investigated (Int. Cl. ⁶ , DB name) H04N 7/24-7/68 H04N 1/41-1/419

Claims (3)

(57) [Claims] An orthogonal transform encoding unit (30) that inputs image data for each frame, divides the data into blocks, encodes using orthogonal transform, and outputs a transform coefficient represented by a frequency component.
0), and a quantization unit (400) that receives the transform coefficient output from the orthogonal transform encoding unit (300), performs quantization at a predetermined quantization step width, and outputs the result. ,
A first weighting control unit (450) that weights a quantization step width in a region where the quantization output of the quantization unit (400) is 0 and changes a weighting coefficient according to a predetermined frequency region of the transform coefficient. Is provided in the quantization unit (400). 2. An orthogonal transform encoding unit (30) that inputs image data for each frame, divides the image data into blocks, performs encoding using orthogonal transform, and outputs transform coefficients represented by frequency components.
0), and a quantization unit (400) that receives the transform coefficient output from the orthogonal transform encoding unit (300), performs quantization at a predetermined quantization step width, and outputs the result. ,
A transform coefficient in a predetermined frequency domain is compared with a reference value, and when there is a transform coefficient larger than the reference value, the quantization unit (4
00) is weighted by a first weighting coefficient with respect to a quantization step width of a region where the quantization output is 0, and when all transform coefficients in the predetermined frequency region are smaller than the reference value, the quantization is performed. A second weighting control unit (460) for weighting the quantization step width of the region where the quantization output of the unit (400) is 0 with a second weighting factor larger than the first weighting factor. A quantization method in the image coding, which is provided in the coding unit (400). 3. An orthogonal transform encoding unit (30) that inputs image data for each frame, divides the image data into blocks, performs encoding using orthogonal transform, and outputs transform coefficients represented by frequency components.
0), and a quantization unit (400) that receives the transform coefficient output from the orthogonal transform encoding unit (300), performs quantization at a predetermined quantization step width, and outputs the result. ,
Image coding characterized in that a third weighting control unit (470) for weighting the maximum value of the quantization step width in a region where the quantization output is 0 is provided in the quantization unit (400). Quantization method.