JP2001045484A - Image coder, image decoder, image coding method and image decoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To apply coding to a target area with precedence over other areas without causing increase of a memory capacity storing wavelet transform coefficients by applying entropy coding to the bit plane of the target area denoted by a ROI mask and the bit plane of a non-target area preceding the target area by the number of priorities altogether. SOLUTION: A wavelet transform section 21 of an image coder divides a digital image into subbands through wavelet transform and outputs a wavelet transform coefficient of each subband. An ROI selection section 22 selects a target area on an image. A quantization section 25 quantizes the wavelet transform coefficient outputted from the transform section 21, and an entropy coding section 26 selects a bit plane of a coding object on the basis of the number (k) of coded bit planes or the like from a plurality of bit planes configuring the wavelet transform coefficient in each subband and applies entropy coding to the selected bit plane.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a function (Regio) for preferentially retaining and encoding the image quality of a selected area of an image.
TECHNICAL FIELD The present invention relates to an image encoding device and an image encoding method having no of interest (ROI), and an image decoding device and an image decoding method for decoding encoded data generated by the image encoding device.

[0002]

2. Description of the Related Art In recent years, a study has been made on a coding method having an ROI which is a function of coding while preserving the image quality of a selected area of an image preferentially. (Reference: “Trends in JPEG2000”, Journal of the Institute of Image Information and Television Engineers, Vol. 52, No.
12, 1998, “Lossy / Lossless R
egion-of-Interest coding
based on Set Partitioning
Hierarchical Trees ”, IEEE
Int. Conf. Image Processin
g, 1998).

FIG. 13 is a block diagram showing a conventional image encoding apparatus. In the figure, reference numeral 1 denotes a wavelet transform unit for converting a digital image into a wavelet and outputting the wavelet transform coefficient, and 2 denotes a wavelet transform unit. ROI selection unit for selecting a region of interest with RO, and RO for setting a priority value S to be given to the region of interest selected by ROI selection unit 2
The I priority setting unit 4 generates an ROI mask that specifies an ROI transform coefficient corresponding to the attention area selected by the ROI selection unit 2 (a transform coefficient of a portion corresponding to the attention area in the wavelet transform coefficients). The mask generation unit 5 is a quantization unit that quantizes the wavelet transform coefficients output from the wavelet transformation unit 1.

Reference numeral 6 denotes a quantization unit 5 with reference to the ROI mask.
An ROI scale-up unit that scales up the ROI transform coefficient quantized according to the priority value S,
Reference numeral 7 denotes an entropy coder for entropy-encoding the wavelet transform coefficients including the ROI transform coefficients scaled up by the ROI scale-up unit 6 in order from the upper bit plane, and 8 denotes entropy-encoded by the entropy coder 7. It is an encoded data generation unit that combines encoding parameters such as encoded data, ROI position information, a priority value S, and a quantization width into one encoded stream.

Next, the operation will be described. First, the wavelet transform unit 1 performs a wavelet transform on a digital image in L stages and outputs the wavelet transform coefficients. For example, as shown in FIG. 14B, in the first stage, the digital image is represented by LL ₁ (vertical low frequency component, horizontal low frequency component) and LH ₁ (vertical high frequency component, horizontal low frequency component). , HL ₁ (vertical low frequency component, horizontal high frequency component), HH ₁ (vertical high frequency component, horizontal high frequency component)
Is divided into sub-bands and wavelet transform is performed. The second stage, by applying the same subband division and the first stage to the lowest frequency component LL _1, the digital image LL _2, LH
₂ , HL ₂ , HH ₂ , LH ₁ , HL ₁ , and HH ₁ are divided into sub-bands and subjected to wavelet transform. Also 3 subsequent stages, the second stage as well as the lowest frequency component LL ₂ by sub-band division, the digital image is converted wavelet repeatedly executed until wavelet transform of L stage is completed.

On the other hand, the ROI priority setting unit 3
When the ROI selection unit 2 selects a region of interest on the image, a priority value S to be assigned to the region of interest is set,
The I mask generation unit 4 generates an ROI mask that specifies an ROI conversion coefficient corresponding to the region of interest.

When the wavelet transform unit 1 outputs a wavelet transform coefficient including an ROI transform coefficient, the quantization unit 5 quantizes the wavelet transform coefficient. RO
When the quantization unit 5 quantizes the wavelet transform coefficient, the I scale-up unit 6 refers to the ROI mask and
The OI conversion coefficient is scaled up by the priority value S. That is, the ROI conversion coefficient is scaled up by shifting the value of the ROI conversion coefficient to the left by S bits. FIG. 14A shows a case where the priority value is “2”.

When the ROI scale-up unit 6 scales up the ROI transform coefficients, the entropy coding unit 7 entropy-codes the wavelet transform coefficients including the ROI transform coefficients in order from the upper bit plane. Finally, the coded data generation unit 8 generates the entropy coded data entropy-coded by the entropy coding unit 7, the position information of the ROI, and the priority value S.
And encoding parameters such as the quantization width into one encoded stream, and output the encoded stream to the image decoding device.

In the above-described image coding apparatus, if the entropy coding process is performed from the most significant bit plane to the least significant bit plane, the image quality of the attention area is higher than that of the other area (non-interest area). Alternatively, encoded data capable of decoding the entire image to the highest image quality under the condition of quantization is generated. On the other hand, if the entropy encoding process is stopped before encoding the least significant bit plane, encoded data capable of decoding and reproducing an image in which the image quality of the attention area is superior to other areas is generated.

FIG. 15 is a block diagram showing a conventional image decoding apparatus. In the figure, 11 extracts entropy coded data, position information of ROI, and priority value S from a coded stream generated by the image coding apparatus. The coded data extraction unit 12 performs a ROI, which is a wavelet transform coefficient corresponding to a region of interest, based on the position information of the ROI.
An ROI mask generator for generating an ROI mask for specifying a transform coefficient decodes entropy-encoded data extracted by the encoded data extractor for each bit plane, and stores entropy for storing wavelet transform coefficients in a memory. It is a decoding unit.

Reference numeral 14 denotes an ROI scale-down unit that scales down the ROI transform coefficient among the wavelet transform coefficients output from the entropy decoding unit 13 by the priority value S with reference to the ROI mask. An inverse quantization unit 16 for inversely quantizing a wavelet transform coefficient including an ROI transform coefficient;
5 is a wavelet inverse transform unit that performs a wavelet inverse transform on the wavelet transform coefficient inversely quantized by 5.

Next, the operation will be described. First, when receiving the encoded stream generated by the image encoding device, the encoded data extracting unit 11 extracts the entropy encoded data, the position information of the ROI, and the priority value S from the encoded stream. ROI mask generator 1
When the encoded data extraction unit 11 extracts the ROI position information, the coded data extraction unit 11 generates a wavelet transform coefficient corresponding to the attention area from the ROI position information, that is, generates an ROI mask that specifies the ROI transform coefficient. On the other hand, the entropy decoding unit 13 decodes the entropy coded data extracted by the coded data extraction unit 11 for each bit plane, and stores the wavelet transform coefficients in the memory.

When the entropy decoding unit 13 stores the wavelet transform coefficient in the memory, the ROI scale-down unit 14 refers to the ROI mask and converts the ROI transform coefficient in the wavelet transform coefficient into the priority value S.
Scale down by minutes. That is, the ROI conversion coefficient is scaled down by shifting the value of the ROI conversion coefficient to the right by S bits. FIG. 16A shows a case where the priority value is “2”.

When the ROI scale-down unit 14 scales down the ROI transform coefficient, the inverse quantization unit 15 inversely quantizes the wavelet transform coefficient including the ROI transform coefficient. The inverse wavelet transform unit 16 includes the inverse quantization unit 1
5 dequantizes the wavelet transform coefficients.
As shown in (b), the wavelet transform coefficient is L
Perform wavelet inverse transformation by combining into stages.

In the above-described image decoding apparatus, when coded data having been subjected to entropy coding up to the least significant bit plane is received, if all of the coded data are decoded, the entire image is subjected to the quantization condition. The reproduction result with the highest image quality is obtained below. On the other hand, if a part of the encoded data is decoded, a reproduced image in which the image quality of the attention area is superior to the image quality of the other areas can be obtained.

In the above-mentioned conventional example, the entropy-encoded data is subjected to a decoding process collectively, but as shown in FIG.
Alternatively, a progressive decoding control unit 17 for repeatedly executing the decoding process for each fixed bit plane number may be provided.

The control process of the progressive decoding control unit 17 is as follows. First, the progressive decoding control unit 1
7 causes the entropy decoding unit 13 to entropy-decode the entropy-encoded data extracted by the encoded data extraction unit 11 by a fixed number of bytes or a fixed number of bitplanes, and display the entropy-decoded result in the previous decoding stage. Are superimposed on the values of the wavelet transform coefficients that have been entropy-decoded up to this point.

Then, the progressive decoding control unit 17 causes the ROI scale-down unit 14 to scale down the value of the ROI transform coefficient in the wavelet transform coefficient, which is the processing result of the entropy decoding unit 13, by the priority value S. After that, the inverse quantization unit 15 outputs RO
The wavelet transform coefficients including the I transform coefficients are dequantized. Then, the progressive decoding control unit 17 causes the inverse wavelet transform unit 16 to perform a process of performing an inverse wavelet transform on the inversely quantized wavelet transform coefficients.

As described above, in the progressive decoding type image decoding apparatus, the attention area is reproduced with higher image quality than the other areas in the stage of sequentially updating the image from the initial stage in which the image is progressively decoded and reproduced. If the encoding has been performed up to the least significant bit plane when the encoded data is created, the entire image is reproduced to the highest quality when the final stage of progressive decoding is reached.

The ROI mask generation unit, which is a component of the above-described image coding apparatus and image decoding apparatus, is described in the reference "Lossless Region of Inter."
est with a Naturally Prog
resistive StillImage Coding
Algorithm ", IEEE Int. Con.
f. As shown in FIG. 18 shown in Image Processing, 1998, an operation for specifying a wavelet transform coefficient which is indispensable for completely reconstructing a region of interest selected on an image is performed.

That is, in order to specify a wavelet transform coefficient required for reproducing each pixel value in the attention area, a wavelet which is an input of the inverse transformation processing is input from a pixel in the attention area which is an output of the inverse transformation processing. A process of tracing the relational expression of the inverse wavelet transform to the area of the transform coefficient is performed. First, the sub-bands LL ₁ , LH ₁ , HL ₁ , and HH ₁ divided by one level from the attention area in the image area (FIG. 18C) are traced, and LL ₁ , LH ₁ , HL ₁ , and HH ₁ are used. Is determined.

[0022] Subsequently, considers ROI transform coefficients in LL ₁ band shown in FIG. 18 (b) and region of interest in the image area, it is the conversion factor required to reproduce the same area to the next division level sub Specified within bands LL ₂ , LH ₂ , HL ₂ , HH ₂ . The result of repeatedly applying the same process to LL _k until the number of division levels L set at the time of encoding is reached is information indicating a wavelet transform coefficient indispensable for completely reconstructing the attention area, that is, FIG.
The ROI mask shown in FIG.

In the operation of tracing the relational expression of each wavelet inverse transform, as shown in FIG. 19, the filter range (filter length) of the wavelet filter used for the inverse transform,
A set of transform coefficients corresponding to each pixel in the region of interest is derived from the rules of downsampling (decimation) or upsampling. When the shape of the attention area is complicated, it is necessary to hold a map indicating whether or not an ROI conversion coefficient is present in a 1-bit / pixel memory of the same number as the number of wavelet conversion coefficients. When the shape of the attention area is a polygon such as a rectangle, holding the position of each outermost conversion coefficient among the conversion coefficients corresponding to each vertex of the polygon, thereby holding the determination result of the ROI conversion coefficient. There is also.

[0024]

Since the conventional image encoding apparatus and image decoding apparatus are configured as described above,
By scaling up the ROI transform coefficient of the region of interest, encoding and decoding can be performed on the region of interest with higher priority than other regions. The number of bit planes of the let transform coefficient increases. As a result, the capacity of the memory for storing the wavelet transform coefficients increases by the increase in the number of bit planes, and the number of bit plane encoding and decoding processes increases, resulting in an increase in the amount of computation. .

In particular, when the encoded data is progressively decoded, the number of bit plane entropy decoding processes is increased due to the increase in the number of bit planes, and the wavelet inverse transform of the decoding result of each plane is performed. The number of times increases by the number of times of the decoding stage as compared with the case of performing non-progressive decoding processing. For this reason, an increase in the number of decoding stages that results in an increase in the number of bit planes due to the scale-up of the ROI transform coefficient has a problem that the amount of computation increases as compared with the non-progressive decoding process.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and encodes a target area with priority over other areas without increasing the capacity of a memory for storing wavelet transform coefficients. It is an object of the present invention to obtain an image encoding device and an image encoding method that can perform processing. Further, the present invention provides an image in which an attention area can be preferentially decoded over other areas without increasing the capacity of a memory for storing wavelet transform coefficients and an increase in the amount of calculation for decoding processing. It is an object to obtain a decoding device and an image decoding method.

[0027]

According to the present invention, there is provided an image coding apparatus comprising: a plurality of bit planes constituting a wavelet transform coefficient of each sub-band outputted from a transforming unit; If the number of bit planes of the non-attention area is less than the priority number, the wavelet transform coefficient of the non-interest area is regarded as 0, and the bit plane of the non-interest area and the bit plane of the attention area indicated by the ROI mask are entropy together. If the number of coded bit planes in each subband is equal to or greater than the priority number, the bit plane of the attention area indicated by the ROI mask and the bit plane of the non-interest area preceding the attention area by the priority number And encoding means for performing entropy encoding together.

According to the image encoding apparatus of the present invention, even when the number of coded bit planes in each subband is equal to or higher than the priority number, the number of coded bit planes in each subband is If the total number of bit planes is equal to or more than the total number of bit planes, the wavelet transform coefficient of the attention area is regarded as 0, and the bit planes of the attention area and the bit planes of the non-interest area that are prior to the attention area by the priority number are combined. Encoding means for performing entropy encoding.

According to the image decoding apparatus of the present invention, when the number of decoded bit planes among the plurality of bit planes constituting the wavelet transform coefficient of each subband output from the decoding means is less than the priority number , Only the bit plane of the region of interest indicated by the ROI mask is subjected to the inverse wavelet transform, and if the number of decoded bit planes is equal to or greater than the priority number, the bit plane of the region of interest indicated by the ROI mask and the bit of the non-interest region An inverse conversion means for performing an inverse wavelet conversion on the planes together is provided.

According to the image decoding apparatus of the present invention, even when the number of decoded bit planes of each sub-band is equal to or more than the priority number, the number of decoded bit planes is calculated from the total number of bit planes of each sub-band. When the number of priority levels is equal to or greater than the number obtained by subtracting the priority number, an inverse transforming means is provided for inversely transforming the wavelet of only the bit plane of the non-interest area.

The image decoding apparatus according to the present invention, when generating an ROI mask for specifying a wavelet transform coefficient corresponding to a region of interest, specifies a wavelet transform coefficient at a final division level in subband division by the wavelet transform. In addition to the ROI mask, an ROI for specifying a wavelet transform coefficient in the subband at the half-split level
A mask generating means for generating a mask is provided.

In the image decoding apparatus according to the present invention, when performing the inverse wavelet transform at each division level, the non-interest area in the low-frequency subband at the next division level is excluded from the range of the inverse wavelet transform. A conversion means is provided.

In the image decoding apparatus according to the present invention, the ROI mask is specified when the wavelet transform coefficients in each subband output for each bit plane by entropy decoding are stored in the inverse wavelet transform memory. Decoding means is provided for reducing the bit plane corresponding to the wavelet transform coefficient of the region of interest and then storing it in a memory.

The image decoding apparatus according to the present invention is provided with control means for causing the decoding processing of the decoding means and the wavelet inverse transformation processing of the inverse transformation means to be executed stepwise for each predetermined processing unit.

According to the image encoding method of the present invention, when the number of encoded bit planes among the plurality of bit planes constituting the wavelet transform coefficient is less than the priority number, the wavelet transform of the non-interest area is performed. Assuming that the coefficient has a value of 0, the bit plane and the ROI of the non-interest area
If the bit planes of the region of interest indicated by the mask are entropy-encoded together and the number of encoded bit planes is equal to or greater than the priority number, the bit planes of the region of interest indicated by the ROI mask and the priority number In this case, entropy coding is performed for the bit plane of the non-interest area just before.

According to the image encoding method of the present invention, even when the number of coded bit planes in each sub-band is equal to or more than the priority number, the number of coded bit planes is reduced in each sub-band. If the total number of bit planes is equal to or more than the total number of bit planes, the wavelet transform coefficient of the attention area is regarded as 0, and the bit planes of the attention area and the bit planes of the non-interest area that are prior to the attention area by the priority number are combined. In this case, entropy coding is performed.

In the image decoding method according to the present invention, when the number of decoded bit planes among the plurality of bit planes constituting the wavelet transform coefficient in each subband is less than the priority number, the ROI mask indicates the ROI mask. The wavelet inverse transform is performed only on the bit plane of the attention area, and when the number of decoded bit planes is equal to or more than the priority number, the bit plane of the attention area indicated by the ROI mask and the bit plane of the non-interest area are combined. The wavelet is inversely transformed.

[0038] In the image decoding method according to the present invention, even when the number of decoded bit planes in each sub-band is equal to or more than the priority number, the number of decoded bit planes is equal to the total number of bit planes in each sub-band. When the number is equal to or greater than the number obtained by subtracting the priority number from the number, only the bit plane in the non-interest area is subjected to the inverse wavelet transform.

According to the image decoding method of the present invention, when generating an ROI mask for specifying a wavelet transform coefficient corresponding to a region of interest, the wavelet transform coefficient at the final division level in subband division by the wavelet transform is specified. In addition to the ROI mask, an ROI for specifying a wavelet transform coefficient in the subband at the half-split level
A mask is also generated.

In the image decoding method according to the present invention, when performing the inverse wavelet transform at each division level, the non-interest area in the low-frequency subband at the next division level is excluded from the range of the inverse wavelet transform. It was made.

In the image decoding method according to the present invention, the ROI mask is specified when the wavelet transform coefficients in each subband output for each bit plane by entropy decoding are stored in the inverse wavelet transform memory. The bit plane corresponding to the wavelet transform coefficient of the attention area is scaled down and then stored in the memory.

In the image decoding method according to the present invention, entropy decoding of entropy-encoded data and inverse wavelet transform of bit planes are executed stepwise for each predetermined processing unit.

[0043]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. FIG. 1 is a block diagram showing an image coding apparatus according to Embodiment 1 of the present invention. In the figure, reference numeral 21 denotes a sub-band division of a digital image by a wavelet transform, and outputs a wavelet transform coefficient of each sub-band. A wavelet conversion unit (conversion means), a ROI selection unit for selecting a region of interest on the image, and an ROI for setting a priority value S (priority number) to be assigned to the region of interest selected by the ROI selection unit The priority setting unit (setting unit) 24 sets an ROI mask that specifies an ROI transform coefficient corresponding to the attention area selected by the ROI selection unit 22 (a transform coefficient of a portion corresponding to the attention area among the wavelet transform coefficients). This is the ROI mask generation unit to generate. The ROI selection unit 22 and the ROI mask generation unit 24 constitute a mask generation unit.

Numeral 25 denotes a quantizer for quantizing the wavelet transform coefficients output from the wavelet transformer 21.
26 selects a bit plane to be encoded from a plurality of bit planes constituting the wavelet transform coefficient in each subband quantized by the quantization unit 25 based on the number k of encoded bit planes and the like. And an entropy encoding unit for entropy encoding. It should be noted that an encoding unit is constituted by the quantization unit 25 and the entropy encoding unit 26.

Reference numeral 27 denotes coded data generation for combining the coded data entropy coded by the entropy coding unit 26, the position information of the ROI, the coding value such as the priority value S and the quantization width into one coded stream. Department. FIG. 3 is a flowchart showing an image encoding method according to Embodiment 1 of the present invention.

Next, the operation will be described. The image encoding apparatus according to the first embodiment generates encoded data similar to the case where the ROI transform coefficient is scaled up, without scaling up the ROI transform coefficient as in the conventional example.

First, the wavelet transform unit 21 performs a wavelet transform on the digital image in L stages, and outputs the wavelet transform coefficients. For example, FIG.
In the first stage, as shown in FIG.
₁ (vertical low frequency component, horizontal low frequency component), LH
₁ (vertical high frequency component, horizontal low frequency component), HL
₁ (vertical low frequency component, horizontal high frequency component), HH
Divide into sub-bands into ₁ (longitudinal high-frequency component, horizontal high-frequency component) and perform wavelet transform. The second stage, by applying the same subband division and the first stage to the lowest frequency component LL _1, the digital image LL _2, LH _2, H
The sub-band is divided into L ₂ , HH ₂ , LH ₁ , HL ₁ , and HH ₁ for wavelet transform. Also 3 subsequent stages, the second stage as well as the lowest frequency component LL ₂ by sub-band division, the digital image is converted wavelet repeatedly executed until wavelet transform of L stage is completed.

On the other hand, ROI priority setting section 23
Sets the priority value S to be given to the region of interest when the ROI selection unit 22 selects the region of interest on the image, and the ROI mask generation unit 24 specifies the ROI mask that specifies the ROI conversion coefficient corresponding to the region of interest. Generate
When the wavelet transform unit 21 outputs a wavelet transform coefficient including the ROI transform coefficient, the quantization unit 25 quantizes the wavelet transform coefficient.

The entropy encoding unit 26 includes a quantization unit 2
5 quantizes the wavelet transform coefficients, and among the plurality of bit planes forming the wavelet transform coefficients of each subband, the number k of already encoded bit planes
Is compared with the priority value S (step ST1).
Here, the total number of bit planes in each subband is N,
The most significant bit plane is called "1" and the least significant bit plane is called "N". In the example of FIG. 2, the total number N of bit planes is six (see FIG. 2A).

When the number k of encoded bit planes is less than the priority value S (k <S), the entropy encoding unit 26 encodes the (k + 1) bit plane of the attention area with reference to the ROI mask. However, since this stage is the initial stage of encoding, and it is necessary to give priority to the encoding process of the attention area, (k) of the non-interest area
The bit plane of +1) is not selected for encoding. Therefore, as shown in FIG. 2B, the wavelet transform coefficient of the non-interest area is regarded as 0 value, and the 0-value bit plane and the (k + 1) bit plane of the attention area are collectively entropy-coded. (Steps ST2 and S
T3).

When the priority value S is not larger than the number k of encoded bit planes (k ≧ S), the entropy encoding unit 26 determines that the initial stage of encoding has already been completed. The number k of encoded bit planes is compared with the total number N of bit planes (step S
T4).

If the number k of coded bit planes in each subband is less than the total number N of bitplanes in each subband (k <N), the entropy coding unit 26 at this stage Since the encoding process for all the bit planes has not been completed, the (k + 1) bit plane of the attention area is selected as an encoding target with reference to the ROI mask, and the (k + 1−S) of the non-interest area is selected.
Are selected as encoding targets (step S
T5).

Then, as shown in FIGS. 2 (c) and 2 (d), the entropy coding section 26 sets the (k + 1)
And the (k + 1-S) bit plane of the non-interest area are entropy-encoded (step ST6). In the example of FIG. 2, since the priority value S is set to “2”, the bit plane in the non-interest area is a bit plane two bits before the bit plane in the area of interest.

When the number k of coded bit planes is not smaller than the total number N of bit planes (k ≧ N), the entropy coding unit 26 refers to the ROI mask to determine the (k + 1−S) Are selected as encoding targets, but at this stage, the encoding processing of all the bit planes in the attention area has already been completed (the encoding processing of the bit plane in the non-interest area has not been completed). ), The (k + 1) bit plane of the attention area is not selected as an encoding target. Therefore, as shown in FIG. 2E, the wavelet transform coefficient of the attention area is regarded as 0 value, and the 0-value bit plane and (k +
The entropy encoding is performed on the bit planes 1-S) (steps ST7 and ST8).

As is apparent from the above, the first embodiment
According to the first S stage, the wavelet transform coefficient of the non-interest area is regarded as a 0-value sub-bit plane,
In the last S stage, encoding is performed by regarding the ROI transform coefficient of the attention area as a sub-bit plane of 0 value, so that the number of encoding stages increases from N times to N + S times as in the conventional example. However, it is not necessary to store the conversion coefficient of the sub-bit plane regarded as the 0 value in the memory, which has the effect of reducing the memory size as compared with the conventional example.

Embodiment 2 FIG. 4 is a block diagram showing an image decoding apparatus according to Embodiment 2 of the present invention. In the figure, reference numeral 31 denotes an entropy coded data, position information of a ROI, and a priority value from an encoded stream generated by the image encoding apparatus. An encoded data extracting unit (extracting means) 32 for extracting S generates an ROI mask generating unit (mask generating unit) for generating an ROI mask for specifying an ROI transform coefficient which is a wavelet transform coefficient corresponding to the attention area from the position information of the ROI. Means), 33 is an encoded data extraction unit 31
Is an entropy decoding unit (decoding means) that decodes the entropy-encoded data extracted by the above for each bit plane and stores the wavelet transform coefficients in a memory.

Reference numeral 34 denotes an ROI scale-down unit that scales down the ROI transform coefficient among the wavelet transform coefficients output from the entropy decoding unit 33 by the priority value S with reference to the ROI mask.
RO scaled down by scale down unit 34
An inverse quantization unit 36 for inversely quantizing the wavelet transform coefficients including the I-transformation coefficient, and a number of decoded planes among a plurality of bit planes constituting the wavelet transform coefficients inversely quantized by the inverse quantization unit 35 A selective wavelet inverse transform unit that selects a bit plane to be decoded based on k and the like and performs wavelet inverse transform. Note that RO
The I scale down unit 34, the inverse quantization unit 35, and the selective wavelet inverse transform unit 36 constitute an inverse transform unit.

Reference numeral 37 denotes progressive decoding in which entropy decoding of entropy-encoded data and inverse wavelet conversion of bit planes are performed stepwise for each predetermined processing unit (for example, a fixed number of bytes or a fixed number of bit planes). It is a control unit (control means). FIG.
9 is a flowchart showing an image decoding method according to Embodiment 2 of the present invention.

Next, the operation will be described. First, when receiving the encoded stream generated by the image encoding device, the encoded data extracting unit 31 extracts the entropy encoded data, the position information of the ROI, and the priority value S from the encoded stream.

When the encoded data extraction unit 31 extracts the position information of the ROI, the ROI mask generation unit 32
Based on the position information of I, a wavelet transform coefficient corresponding to the attention area, that is, an ROI mask for specifying the ROI transform coefficient is generated. On the other hand, the entropy decoding unit 33 decodes the entropy coded data extracted by the coded data extraction unit 31 for each bit plane, and stores the wavelet transform coefficients in the memory.

When the entropy decoding unit 33 stores the wavelet transform coefficient in the memory, the ROI scale-down unit 34 refers to the ROI mask and converts the ROI transform coefficient in the wavelet transform coefficient into the priority value S.
Scale down by minutes. That is, the ROI transform coefficient is scaled down by right shifting the value of the ROI transform coefficient by S bits (equivalent to dividing the wavelet transform coefficient by 2 ^S ).

When the ROI scale-down unit 34 scales down the ROI transform coefficient, the inverse quantization unit 35 inversely quantizes the wavelet transform coefficient including the ROI transform coefficient. When the inverse quantization unit 35 inversely quantizes the wavelet transform coefficient, the selective wavelet inverse transform unit 36
A bit plane to be decoded is selected from the plurality of bit planes constituting the wavelet transform coefficient based on the number k of decoded planes and the like, and wavelet inverse transform is performed.

More specifically, of the plurality of bit planes constituting the wavelet transform coefficient of each subband, the number k of bit planes already decoded and the priority value S
Are compared (step ST11). Here, the total number of bit planes in each subband is N, the most significant bit plane is “1”, and the least significant bit plane is “N”.

When the number k of decoded bit planes is less than the priority value S (k <S), the selective wavelet inverse transform unit 36 is an initial stage in which it is necessary to give priority to the decoding process of the attention area. , With reference to the ROI mask (step ST12), as shown in FIG. 5, the transform coefficients to be subjected to the inverse wavelet transform are limited to only the ROI transform coefficients (step ST12).

Then, the selective wavelet inverse transform unit 3
6 shows the ROI transform coefficient and the ROI in the low band sub-band generated stepwise during the wavelet inverse transform process.
The inverse wavelet transform process is performed only for the transform coefficient reflecting the value of the transform coefficient to obtain a decoded image (step ST13).

When the priority value S is not larger than the number k of decoded bit planes (k ≧ S), the selective wavelet inverse transform unit 36 determines that the decoded bit A value (NS) obtained by subtracting the priority value S from the number k of planes and the total number N of bit planes is compared (step ST14).

If the number k of decoded bit planes is smaller than the value (N−S) obtained by subtracting the priority value S from the total number N of bit planes (k <N−S), Since decoding of all bit planes in the region of interest has not been completed, all wavelet transform coefficients in each subband are subjected to inverse wavelet transform to obtain a decoded image (step ST15).

The selective wavelet inverse transform unit 36 determines that the number k of decoded bit planes is not smaller than a value (N−S) obtained by subtracting the priority value S from the total number N of bit planes (k ≧ N−S). Since decoding of all bit planes in the region of interest has already been completed (decoding of bit planes in the region of non-interest has not been completed), the inverse wavelet transform is performed with reference to the ROI mask. The transform coefficients to be applied are limited to those other than the ROI transform coefficients (step ST16).

Then, the selective wavelet inverse transform unit 3
Reference numeral 6 denotes a non-ROI transform coefficient (a wavelet transform coefficient other than the ROI transform coefficient) and a transform that reflects the value of the non-ROI transform coefficient in the low-frequency sub-band generated stepwise during the wavelet inverse transform processing. The inverse wavelet transform process is performed only for the coefficients to obtain a decoded image (step ST17).

As is apparent from the above, the second embodiment
According to the above, in step ST13, the processing amount is suppressed by performing the inverse wavelet transform only when the ROI transform coefficient is included in the filtering range, and the processing amount is reduced in step ST17.
Since the amount of processing is suppressed by performing the inverse wavelet transform only when the non-ROI transform coefficient is included in the filtering range, the amount of processing required for progressive decoding of the entire image is reduced by emphasizing the image quality in the region of interest. There is an effect that the amount of calculation can be suppressed to the same amount as in the progressive decoding process when not performed.

Embodiment 3 In the second embodiment, the ROI mask that specifies the wavelet transform coefficient of the final division level in the subband division by the wavelet transform has been described. However, the wave of the final division level in the subband division by the wavelet transform is shown. In addition to the ROI mask that specifies the let transform coefficient,
R that specifies the wavelet transform coefficient at the halfway split level
An OI mask may be generated.

That is, as shown in FIG. 7, the ROI mask generating section 32 of the third embodiment includes, in addition to the ROI mask of FIG. Low frequency subbands LL ₁ , L
The transformation coefficients corresponding to the region of interest in L ₂ ,..., LL _L-1 are also held as ROI transformation coefficients.

When the wavelet inverse transform of the first stage is performed by referring to the ROI mask holding all the results of specifying the ROI transform coefficients at each division level (the inverse transform from L level division to L-1 level division) ), When performing the wavelet inverse transformation at any stage (inverse transformation from k-level division to k-1 level division), the next division level is determined without judging from the inverse transformation result at each stage. It is possible to determine a conversion coefficient to be used at the time of the reverse conversion. For this reason, there is an effect that the processing amount required for the selection control of the wavelet inverse transformation processing can be further reduced.

Embodiment 4 In the third embodiment, the RO that specifies the wavelet transform coefficient at the halfway split level
Although the generation of the I mask has been described, as shown in FIG. 8, when performing the inverse wavelet transform at each division level, the non-interest area in the low frequency subband at the next division level is subjected to the inverse wavelet transform. May be excluded from the range.

More specifically, the selective wavelet inverse transform unit 36 calculates k levels from the sub-band divided k levels.
In the case where inverse conversion is performed on subbands in a state of -1 level division (k = 0, 1,..., L: k = 0, division is performed 0 times, that is, an image is shown), The range in which the wavelet inverse transform is performed only for the inverse transform that gives each ROI transform coefficient in the frequency sub-band LL _k-1 as an output value is changed. In the wavelet inverse transform process, it is necessary to perform a filter process on a plurality of transform coefficients in order to obtain one output value of the inverse transform. Here, if it is clear which output value is to be subjected to the inverse transform, the selective wavelet can be selected without performing the process of determining whether the ROI transform coefficient is included in the filtering range. Inverse conversion control becomes possible, and processing can be made more efficient.

Embodiment 5 FIG. 9 is a block diagram showing an image decoding apparatus according to Embodiment 5 of the present invention. In the figure, the same reference numerals as those in FIG. 4 denote the same or corresponding parts, and a description thereof will be omitted. Reference numeral 38 denotes an ROI scale-down unit that scales down the ROI transform coefficients in the wavelet transform coefficients output from the entropy decoding unit 33 and stores the scale-down coefficients in a memory.

Next, the operation will be described. In the second to fourth embodiments, the bit planes constituting the wavelet transform coefficients output from the entropy decoding unit 33 are sequentially stored in the memory, and then the ROI scale-down unit 3
8 shows a case in which the ROI transform coefficient among the wavelet transform coefficients is scaled down, but the ROI scale-down unit 38 outputs the ROI in each bit plane constituting the wavelet transform coefficient output from the entropy decoding unit 33. A sub-bit plane serving as information of a transform coefficient may be stored in a memory after being scaled down.

That is, every time the entropy decoding unit 33 outputs each bit plane of the wavelet transform coefficient, R
By referring to the ROI mask, the OI scale-down unit 38 scales down the sub-bit plane portion of the bit plane corresponding to the ROI conversion coefficient by the priority value S, and stores the same data in the conversion coefficient buffer as a memory. It is stored at the plane position scaled down by the priority value S. The sub-bit planes corresponding to the non-ROI conversion coefficients in the S bit planes including the most significant bit plane to the most significant bit plane are stored in the transformation coefficient buffer. Since this information is information of a value “0” generated by scaling up the ROI conversion coefficient by the priority value S, discarding this does not affect the image quality. The information of the ROI transform coefficient in the bit plane of the level lower than the priority value S counted from the least significant bit plane is lost by the same scale down, but this information is always "0" generated by the scale up at the time of encoding. Information,
Without losing information necessary for progressive reproduction of an image, the size of a memory for storing wavelet transform coefficients can be suppressed to the same size as when ROI processing is not performed (see FIG. 10).

Embodiment 6 FIG. In the second to fifth embodiments, the progressive decoding control unit 37 is added to execute the entropy decoding process and the inverse wavelet transform process step by step for each predetermined processing unit.
As shown in FIGS. 11 and 12, the progressive decoding control unit 37 may be deleted and the entropy-encoded data may be collectively subjected to a decoding process. Can play.

[0080]

As described above, according to the present invention, among the plurality of bit planes constituting the wavelet transform coefficient of each sub-band outputted from the transforming means, the coded bit plane of each sub-band has been encoded. Is smaller than the priority number, the wavelet transform coefficient of the non-interest area is set to 0.
Value and the bit plane and RO of the non-interest area
If the bit planes of the region of interest indicated by the I mask are entropy-encoded together, and the number of encoded bit planes in each subband is equal to or greater than the priority number, R
Since the encoding means for entropy encoding the bit plane of the attention area indicated by the OI mask and the bit plane of the non-interest area preceding the attention area by the number of priorities together is provided, the wavelet transform coefficient is stored. This has the effect that the attention area can be preferentially encoded over the other areas without increasing the capacity of the memory to be used.

According to the present invention, even when the number of coded bit planes in each sub-band is equal to or more than the priority number, the number of coded bit planes is equal to the total number of bit planes in each sub-band. In the above case, the wavelet transform coefficient of the attention area is regarded as 0 value, and the bit plane of the attention area and the bit plane of the non-interest area before the attention area by the priority number are entropy-encoded together. Since the encoding means is provided to perform the encoding, the capacity of the memory for storing the wavelet transform coefficients can be further reduced.

According to the present invention, when the number of decoded bit planes among the plurality of bit planes constituting the wavelet transform coefficient of each subband output from the decoding means is less than the priority number, the ROI mask When the number of decoded bit planes is equal to or greater than the priority number, the bit plane of the ROI mask and the bit plane of the non-ROI area are compared with each other. Since the inverse transform means for collectively performing the inverse transform of the wavelet is provided, it is possible to increase the memory capacity for storing the wavelet transform coefficients and increase the calculation amount of the decoding process without changing the region of interest. There is an effect that the decoding process can be performed with priority over the region.

According to the present invention, even when the number of decoded bit planes of each sub-band is equal to or larger than the priority number, the number of decoded bit planes can be calculated based on the total number of bit planes of each sub-band. When the number is smaller than the subtracted number, the inverse wavelet transform means for inversely transforming only the bit plane of the non-interest area is provided, so that the capacity of the memory for storing the wavelet transform coefficients can be further reduced. In addition to this, there is an effect that the calculation amount of the decoding process can be further reduced.

According to the present invention, when the ROI mask for specifying the wavelet transform coefficient corresponding to the attention area is generated, the R for specifying the wavelet transform coefficient at the final division level in the sub-band division by the wavelet transform.
In addition to the OI mask, a mask generating means for generating an ROI mask for specifying a wavelet transform coefficient in a subband at a half-way split level is provided. There is an effect that can be reduced.

According to the present invention, when performing the inverse wavelet transform at each division level, the inverse transform means for excluding the non-interest area in the low-frequency subband at the next division level from the range of the inverse wavelet transform is provided. Since the configuration is provided, the processing amount required for the selection control of the wavelet inverse transformation processing can be reduced.

According to the present invention, when the wavelet transform coefficients in each sub-band output for each bit plane by entropy decoding are stored in the wavelet inverse transform memory, the ROI mask specifies the region of interest. Since the decoding means for reducing the bit plane corresponding to the wavelet transform coefficient and storing it in the memory is provided, there is an effect that the capacity of the memory for storing the wavelet transform coefficient can be further reduced.

According to the present invention, since the control means for executing the decoding processing of the decoding means and the wavelet inverse transformation processing of the inverse transformation means in a stepwise manner for each predetermined processing unit is provided, the decoded image This has the effect that the image can be reproduced with the same processing amount as the processing for reproducing the image without using the stepwise processing.

According to the present invention, when the number of coded bit planes among the plurality of bit planes constituting the wavelet transform coefficient is less than the priority number, the wavelet transform coefficient of the non-interest area is set to 0 value. And entropy-encoding the bit planes of the non-interest area and the bit planes of the attention area indicated by the ROI mask together, and if the number of encoded bit planes is equal to or greater than the priority number, the ROI mask is The bit plane of the attention area shown and the bit plane of the non-interest area that is prior to the attention area by the priority number are configured to be entropy-encoded, thereby increasing the capacity of the memory for storing the wavelet transform coefficients. Thus, there is an effect that the attention area can be subjected to the encoding process with priority over the other areas.

According to the present invention, even when the number of coded bit planes in each subband is equal to or more than the priority number, the number of coded bit planes is equal to the total number of bit planes in each subband. In the above case, the wavelet transform coefficient of the attention area is regarded as 0 value, and the bit plane of the attention area and the bit plane of the non-interest area before the attention area by the priority number are entropy-encoded together. Therefore, the memory capacity for storing the wavelet transform coefficients can be further reduced.

According to the present invention, when the number of decoded bit planes among the plurality of bit planes constituting the wavelet transform coefficient in each subband is less than the priority number, the ROI mask indicates the region of interest indicated by the ROI mask. If only the bit plane is subjected to the wavelet inverse transform and the number of decoded bit planes is equal to or more than the priority number, the ROI
Since the bit plane of the attention area and the bit plane of the non-interest area indicated by the mask are configured to be subjected to the inverse wavelet transform, the capacity of the memory for storing the wavelet transform coefficients is increased, and the calculation amount of the decoding process is reduced. There is an effect that the attention area can be preferentially decoded over other areas without increasing.

According to the present invention, even when the number of decoded bitplanes in each subband is equal to or higher than the priority number, the number of decoded bitplanes is higher than the total number of bitplanes in each subband. When the number is equal to or more than the number obtained by subtracting the frequency, the configuration is such that only the bit planes in the non-interest area are subjected to the wavelet inverse transform, so that the capacity of the memory for storing the wavelet transform coefficients can be further reduced, There is an effect that the amount of calculation in the decoding process can be further reduced.

According to the present invention, when the ROI mask for specifying the wavelet transform coefficient corresponding to the attention area is generated, the ROI mask for specifying the wavelet transform coefficient at the final division level in the sub-band division by the wavelet transform is used. In addition, since the ROI mask for specifying the wavelet transform coefficient in the subband at the halfway division level is also generated, the processing amount required for the selection control of the wavelet inverse transform process can be reduced. .

According to the present invention, when the inverse wavelet transform is performed at each division level, the non-interest area in the low-frequency subband at the next division level is excluded from the range of the inverse wavelet transform. Therefore, there is an effect that the processing amount required for the selection control of the wavelet inverse transformation processing can be reduced.

According to the present invention, when the wavelet transform coefficients in each subband output for each bit plane by the entropy decoding are stored in the wavelet inverse transform memory, the ROI mask specifies the region of interest. Since the bit plane corresponding to the wavelet transform coefficient is configured to be scaled down and then stored in the memory, there is an effect that the capacity of the memory for storing the wavelet transform coefficient can be further reduced.

According to the present invention, the entropy decoding process of the entropy coded data and the wavelet inverse transform process of the bit plane are configured to be executed step by step for each predetermined processing unit. There is an effect that an image can be reproduced with the same processing amount as the processing of reproducing an image without using the stepwise processing.

[Brief description of the drawings]

FIG. 1 is a configuration diagram illustrating an image encoding device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram illustrating a flow of a process of the image encoding device.

FIG. 3 is a flowchart showing an image encoding method according to the first embodiment of the present invention.

FIG. 4 is a configuration diagram showing an image decoding apparatus according to Embodiment 2 of the present invention.

FIG. 5 is an explanatory diagram illustrating a selective inverse wavelet transform.

FIG. 6 is a flowchart showing an image decoding method according to Embodiment 2 of the present invention.

FIG. 7 is an explanatory diagram illustrating generation of an ROI mask.

FIG. 8 is an explanatory diagram illustrating a selective inverse wavelet transform.

FIG. 9 is a configuration diagram showing an image decoding apparatus according to Embodiment 5 of the present invention.

FIG. 10 is an explanatory diagram illustrating a flow of a process of the image decoding device.

FIG. 11 is a configuration diagram showing an image decoding apparatus according to Embodiment 6 of the present invention.

FIG. 12 is an explanatory diagram illustrating a processing flow of the image decoding device.

FIG. 13 is a configuration diagram illustrating a conventional image encoding device.

FIG. 14 is an explanatory diagram illustrating a flow of a process of the image encoding device.

FIG. 15 is a configuration diagram illustrating a conventional image decoding device.

FIG. 16 is an explanatory diagram illustrating a flow of a process of the image decoding device.

FIG. 17 is a configuration diagram illustrating a conventional image decoding device.

FIG. 18 is an explanatory diagram illustrating generation of an ROI mask.

FIG. 19 is an explanatory diagram illustrating identification of an ROI conversion coefficient.

[Explanation of symbols]

21 Wavelet conversion unit (conversion means), 22 RO
I selection unit (mask generation unit), 23 ROI priority setting unit (setting unit), 24 ROI mask generation unit (mask generation unit), 25 quantization unit (encoding unit),
26 Entropy encoding unit (encoding unit), 27 encoded data generation unit, 31 encoded data extraction unit (extraction unit), 32 ROI mask generation unit (mask generation unit),
33 entropy decoding unit (decoding means), 34, 38
ROI scale-down section (inverse transforming means), 35 inverse quantizing section (inverse transforming means), 36 selective wavelet inverse transforming section (inverse transforming means), 37 progressive decoding control section (control means).

Claims (16)

[Claims] A converting unit that divides an image into sub-bands by a wavelet transform and outputs a wavelet transform coefficient of each sub-band; a setting unit that sets a priority number of a region of interest on the image; Mask generating means for generating an ROI mask for specifying a wavelet transform coefficient corresponding to the sub-band, and a code of each sub-band among a plurality of bit planes constituting the wavelet transform coefficient of each sub-band output from the transform means. If the number of converted bit planes is less than the priority number, the wavelet transform coefficient of the non-interest area is regarded as 0, and the bit plane of the non-interest area and the bit plane of the attention area indicated by the ROI mask are compared with each other. Are entropy coded together and the number of coded bit planes in each subband is equal to or greater than the priority number. If this is the case, an image coding apparatus comprising coding means for entropy coding the bit planes of the region of interest indicated by the ROI mask and the bit planes of the non-target region preceding the region of interest by the number of priorities. 2. The coding means according to claim 1, wherein, even when the number of coded bit planes of each subband is equal to or more than the priority number, the number of coded bit planes is equal to or more than the total number of bit planes of each subband. In this case, the wavelet transform coefficient of the region of interest is regarded as 0, and the bit planes of the region of interest and the bit planes of the non-region of interest that are prior to the region of interest by the number of priorities are entropy-encoded together. The image encoding device according to claim 1, wherein: 3. Extraction means for extracting entropy-encoded data, position information of a region of interest, and priority number of the region of interest from the image-encoded data, and a wavelet corresponding to the region of interest from the position information extracted by the extraction means. Mask generating means for generating an ROI mask for specifying a transform coefficient; decoding means for entropy decoding the entropy-encoded data extracted by the extracting means to output wavelet transform coefficients for each sub-band; If the number of decoded bit planes among the plurality of bit planes constituting the output wavelet transform coefficient of each subband is less than the priority number, the above R
When only the bit plane of the attention area indicated by the OI mask is subjected to the inverse wavelet transform, and the number of decoded bit planes is equal to or more than the priority number, the bit plane of the attention area indicated by the ROI mask and the bit of the non-interest area An image decoding apparatus comprising: an inverse transform unit that performs a wavelet inverse transform by combining planes. 4. The inverse transform means, even when the number of decoded bitplanes of each subband is equal to or more than the priority number,
When the number of decoded bit planes is equal to or greater than the number obtained by subtracting the number of priorities from the total number of bit planes of each subband, only the bit planes in the non-interest area are subjected to the inverse wavelet transform. Item 3. The image decoding device according to Item 3. 5. An ROI mask for specifying a wavelet transform coefficient at a final division level in subband division by wavelet transform when generating an ROI mask for specifying a wavelet transform coefficient corresponding to a region of interest. 5. The image decoding apparatus according to claim 3, further comprising generating an ROI mask for specifying a wavelet transform coefficient in the subband at the halfway division level. 6. The inverse transform means, when performing the inverse wavelet transform at each division level, excludes the non-interest area in the low-frequency subband at the next division level from the range of the inverse wavelet transform. Claim 5
The image decoding device according to any one of the preceding claims. 7. When the wavelet transform coefficients in each sub-band output for each bit plane by entropy decoding are stored in a wavelet inverse transform memory, the decoding means determines a region of interest specified by the ROI mask. 7. The image decoding apparatus according to claim 3, wherein the bit plane corresponding to the wavelet transform coefficient is scaled down and then stored in a memory. 8. A control means for performing stepwise execution of the decoding processing of the decoding means and the wavelet inverse transformation processing of the inverse transformation means for each predetermined processing unit.
The image decoding device according to any one of claims 1 to 7. 9. An image is divided into sub-bands by a wavelet transform, a wavelet transform coefficient of each sub-band is output, and a priority number of an attention area on the image is set, and a wave corresponding to the attention area is set. When the ROI mask specifying the wavelet transform coefficient is generated, if the number of coded bit planes among the plurality of bit planes constituting the wavelet transform coefficient is less than the priority number, the wavelet of the non-interest area is generated. Assuming that the transform coefficient has a value of 0, the bit planes of the non-interest area and the bit planes of the attention area indicated by the ROI mask are collectively entropy-encoded, and the number of encoded bit planes is greater than or equal to the priority number. If the above RO
An image coding method for collectively entropy coding a bit plane of a region of interest indicated by an I mask and a bit plane of a non-region of interest preceding the region of interest by the number of priorities. 10. Even when the number of coded bit planes in each sub-band is equal to or more than the priority number, the number of coded bit planes is equal to or more than the total number of bit planes in each sub-band. Is characterized in that a wavelet transform coefficient of a region of interest is regarded as 0 value, and a bit plane of the region of interest and a bit plane of a non-region of interest that are prior to the region of interest by the number of priorities are collectively entropy-encoded. The image coding method according to claim 9, wherein 11. A wavelet transform that extracts entropy-encoded data, position information of a region of interest, and priority number of a region of interest from image-encoded data, and entropy-decodes the entropy-coded data to form each subband. When the coefficients are output in units of bit planes and an ROI mask that specifies a wavelet transform coefficient corresponding to the region of interest is generated from the position information, a plurality of bit planes constituting the wavelet transform coefficients in each subband are generated. When the number of decoded bit planes is less than the priority number, only the bit plane of the attention area indicated by the ROI mask is subjected to the inverse wavelet transform,
An image decoding method in which, when the number of decoded bit planes is equal to or larger than the priority number, the bit plane of the attention area and the bit plane of the non-interest area indicated by the ROI mask are collectively subjected to the inverse wavelet transform. 12. Even when the number of decoded bit planes in each sub-band is equal to or greater than the priority number, the number of decoded bit planes is obtained by subtracting the priority number from the total number of bit planes in each sub-band. 12. The image decoding method according to claim 11, wherein when the number is equal to or more than the number, only the bit plane of the non-interest area is subjected to the inverse wavelet transform. 13. When generating an ROI mask for specifying a wavelet transform coefficient corresponding to a region of interest, in addition to an ROI mask for specifying a wavelet transform coefficient at a final division level in subband division by wavelet transform, 13. The image decoding method according to claim 11, wherein an ROI mask for specifying a wavelet transform coefficient in a sub-band at a division level is also generated. 14. The method according to claim 13, wherein when the inverse wavelet transform is performed at each division level, a non-interest area in the low-frequency sub-band at the next division level is excluded from the range of the inverse wavelet transform. The image decoding method described in the above. 15. A wavelet transform coefficient of a region of interest specified by an ROI mask when storing a wavelet transform coefficient in each subband output for each bit plane by entropy decoding in a memory for inverse wavelet transform. The image decoding method according to any one of claims 11 to 14, wherein a bit plane corresponding to (i) is scaled down and stored in a memory. 16. The method according to claim 11, wherein the entropy decoding process of the entropy-encoded data and the wavelet inverse transform process of the bit plane are executed stepwise for each predetermined processing unit. 2. The image decoding method according to claim 1.