JPH11243544A - Image data processing method and image data processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To compress and expand the image data with high compression efficiency by eliminating the high frequency components of a low luminance area, according to its visual characteristics. SOLUTION: A DC component data under attention 501, i.e., the data on an arbitrary part of a layer 2DC wavelet space 1610, i.e., a DC space that is generated by the wavelet conversion is noted. If the data 501 has a low value, the insignificant decision threshold is increased for the corresponding area (layer 1HL estimation reference data 502) included in an HL space 1609 of a layer 2. In the same way, the insignificant decision threshold is increased for the corresponding areas (layer 1LH estimation reference data 507 and layer 1HH estimation reference data 508) in a layer 2LH space 1608 and a layer 2HH space 1607, in terms of both LH and HH spaces respectively. Thus, it is possible to reduce the quantity of data by eliminating the high frequency components of a low luminance area in accordance with its visual characteristics.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an image data processing method for image compression processing and the like and an image data processing apparatus used for image compression processing and the like.

[0002]

2. Description of the Related Art DCT (Discrete Cosine Transform) technology has been actively applied as a high-efficiency image coding technology on the premise of a communication medium or a recording medium. However, as a fundamental problem inherent in the compression method using DCT, there is a problem that when the compression ratio is increased, block distortion, mosquito noise, and the like are visually recognized, and the compression ratio is limited.

Therefore, in recent years, a new compression method has been proposed in order to improve the compression ratio, and in particular, a compression technique using a wavelet transform, which is one of the sub-band codings, has attracted attention. When this wavelet transform is used, since there is no concept of a block, distortion between blocks generated by DCT is eliminated, and a considerable improvement in image quality can be visually expected.

[0004] One of the features of the compressed image obtained by the DCT and the wavelet is that the DCT is a digital image, and an image having discontinuous block-like distortion becomes conspicuous by increasing the compression ratio. On the other hand, the wavelet becomes an analog image, and the high frequency component is naturally lost by increasing the compression rate. In other words, it is gradually cut from the high frequency part of the signal band. As a result, the resolution is reduced as a whole, so that the visual quality degradation is reduced as compared with DCT at the same compression ratio.

A conventional wavelet compression / expansion method will be described below. FIG. 9 is a block diagram showing the overall configuration of a conventional wavelet image compression apparatus. In FIG. 9, reference numeral 1001 denotes an original image output device; 1002, a layer 0 subband decomposition device; 1003, a layer 1 subband decomposition device;
Is a layer 2 sub-band decomposition device, 1005 is a meaningless space estimation and deletion device, 1051 is layer 2 DC data, 1052 is layer 2 HL data, 1053 is layer 2 LH data, 1054 is layer 2 HH data, 1055 is layer 1 HL data, and 1056 is layer 1 LH. Data, 1057 is layer 1 HH data, 1058 is layer 0 HL data, 1059 is layer 0 LH data, 1060 is layer 0 HH data, 1061 is original image data, 1062 is layer 0 DC data, 10
63 is layer 0 HL data, 1064 is layer 0 LH data, 1065
Is layer 0HH data, 1066 is layer 1 DC data, 1067 is layer 1 HL data, 1068 is layer 1 LH data, and 1069 is layer 1 HH data.

FIG. 10 is a block diagram showing the overall configuration of a conventional wavelet image decompression device. In FIG. 10, reference numeral 1101 denotes an insignificant space estimation / expansion device, 1102 denotes a layer 0 sub-band synthesis device, 1103 denotes a layer 1 sub-band synthesis device,
1104 is a layer 2 subband synthesizing device, 1105 is an expanded image output device, 1151 is expanded image data, 1152 is layer 1 HL expanded data, 1153 is layer 1 LH expanded data, 1154 is layer 1
HH extension data, 1155 is layer 0HL extension data, 1156 is layer 0 LH extension data, 1157 is layer 0HH extension data, 11
58 is layer 1 DC extension data, and 1159 is layer 0 DC extension data.

FIG. 11 is a block diagram showing a conventional meaningless space estimation and deletion device. In FIG. 11, reference numeral 1201 denotes a layer 2
Insignificant space estimator from HL space, 1202 is insignificant space estimator from layer 2 LH space, 1203 is insignificant space estimator from layer 2 HH space, 1204 is layer 1 HL deleter, 1205 is layer 1
LH deletion unit, 1206 is layer 1HH deletion unit, 1207 is layer 1HL
Insignificant space estimating unit from space, 1208 is insignificant space estimating unit from layer 1 LH space, 1209 is insignificant space estimating unit from layer 1HH space, 1210 is layer 0HL deleting unit, 1211 is layer 0LH
Deletion unit, 1212 is a layer 0HH deletion unit, 1251 is layer 1HL insignificant space estimation data, 1252 is layer 1LH insignificant space estimation data, 1253 is layer 1HH insignificant space estimation data, 1254 is layer 0HL insignificant space estimation data, 1255 is layer 0LH meaningless space estimation data, 1256 is layer 0HH meaningless space estimation data.

FIG. 12 is a block diagram showing a conventional insignificant space estimation and development device. In FIG. 12, the insignificant space estimation unit
Since 1201, 1202, 1203, 1207, 1208, and 1209 are the same as those in FIG. 11, the same reference numerals are given and the description is omitted. 1301 is a layer 1 HL development unit, 1302 is a layer 1 LH development unit, 1303 is a layer 1 HH development unit, 1304 is a layer 0 HL development unit, 1305 is a layer 0 LH development unit, 1306 is a layer 0 HH development unit, and 1351 is a layer 1 HL involuntary space development estimation Data, 1352 is layer 1 LH insignificant space expansion estimation data, 1353 is layer 1HH insignificant space expansion estimation data, 1354 is layer 0HL insignificant space expansion estimation data, 13
Reference numeral 55 denotes layer 0 LH meaningless space expansion estimation data, and reference numeral 1356 denotes layer 0 HH meaningless space expansion estimation data.

FIG. 13 is a block diagram showing a sub-band decomposition apparatus. In FIG. 13, 1401 is a horizontal low-pass filter, 1402 is a horizontal high-pass filter, 1403 is a horizontal down-sampler, 1404 is a vertical low-pass filter, 1405 is a vertical high-pass filter, 1406 is a vertical down-sampler, and 1451 is DC
Separated data, 1452 is LH separated data, 1453 is HL separated data, 1454 is HH separated data, 1455 is input data, 1456 is horizontal low frequency data, 1457 is horizontal high frequency data, 1458 is horizontal
DC separation data, 1459 is horizontal H separation data, 1460 is horizontal DC
Vertical low frequency data, 1461 is horizontal DC vertical high frequency data, 14
62 is horizontal H vertical low frequency data, and 1463 is horizontal H vertical high frequency data.

FIG. 14 is a block diagram showing a subband synthesizing apparatus. 14, 1501 is a vertical upsampler, 1502 is a vertical lowpass filter, 1503 is a vertical highpass filter, 1504 is a horizontal upsampler, 1505 is a horizontal lowpass filter, 1506 is a horizontal highpass filter, 1551 is synthetic data, and 1552 is DC input data. , 1553 is LH input data,
1554 is HL input data, 1555 is HH input data, 1556 is DC upsample data, 1557 is LH upsample data,
1558 is HL upsampled data, 1559 is HH upsampled data, 1560 is horizontal DC combined data, 1561 is horizontal H combined data, 1562 is horizontal DC upsampled data, and 1563 is horizontal H upsampled data.

FIG. 15 is an explanatory diagram of a wavelet space. In FIG. 15, 1601, 1602, and 1603 are wavelet spaces of layer 0, 1601 is HH, 1602 is LH, and 1603 is
HL wavelet space. 1604, 1605, and 1606 are wavelet spaces of layer 1, 1604 is HH, and 1605 is L
H and 1606 are HL wavelet spaces. 1607, 1608, 1
609 is a layer 2 wavelet space, and 1607 is H
H and 1608 are LH and 1609 are HL wavelet spaces. 161
0 is a layer 2 DC wavelet space. 1651 is layer 0 HL significant data, and 1652 is layer 0 HL estimation reference data.

Next, the operation of the conventional image data processing apparatus will be described with reference to the drawings. First, the operation of the conventional wavelet image compression device will be described with reference to FIG. In FIG. 9, assuming that the original image data 1061 is provided from the original image output device 1001, the layer 0 subband decomposition device 1002 receives the original image data 1061, performs the first wavelet transform, and obtains the wavelet data 1062・ 1063 ・ 1064 ・
Generate 1065. The generated wavelet data is hereinafter referred to as layer 0 wavelet data.

Next, the layer 1 sub-band decomposition apparatus 1003 receives the wavelet DC data 1062 of layer 1 and performs a second wavelet transform to generate wavelet data 1066/1067/1068/1069. . The generated wavelet data is hereinafter referred to as layer 1 wavelet data. Next, the layer 2 sub-band decomposition device 1004 receives the wavelet DC data 1066 of the layer 2 and performs the third wavelet transform to obtain the wavelet data 1051, 1052, 1053, and 1010.
Generate 54. The generated wavelet data is hereinafter referred to as layer 2 wavelet data.

The wavelet data generated as described above has a strong correlation between adjacent layers. The meaningless space estimation and deletion device 1005 performs data compression based on this correlation, so that the wavelet data 1052, 1053,
In response to 54 · 1067 · 1068 · 1069 · 1063 · 1064 · 1065, data compression is performed by deleting the insignificant space.

Next, the operation of the subband decomposition devices 1002, 1003, and 1004 will be described with reference to FIG. In FIG. 13, receiving the two-dimensional input data 1455, 1401 performs horizontal low frequency filtering to generate 1456, 1402 performs horizontal high frequency filtering to generate 1457, and 1456,
1457 is horizontally downsampled by 1403 each 1458,
1459 is generated. 1458 is vertically filtered by 1404 and 1405 to generate 1460 and 1461. Likewise
1459 is vertically filtered by 1404 and 1405 to generate 1462 and 1463. 1460,1461,1462,1463
Are vertically downsampled by 1406, respectively, 1451, 145
Generate 2,1453,1454 and wavelet transform.

Next, the operation of the insignificant space estimation and deletion device 1005 will be described with reference to FIG. In FIG. 11, 1201 is
Receiving the wavelet data 1052, a minimum value area estimation of 1067 is performed by threshold judgment to generate 1251. Similarly, 12
02 receives 1053 to produce 1252, 1203 receives 1054 to 12
Generate 53. Receiving 1251 and 1067, 1204 deletes the data of the insignificant estimation area to generate compressed data 1055. Similarly, 1205 receives 1252 ・ 1068, 1206 becomes 1253, 10
Upon receiving 69, 1056 and 1057 are generated, respectively. Then, 12
07 receives 1055, performs 1063 minimum value area estimation by threshold value determination, and generates 1254. Similarly, 1208 receives 1056 and 1209 receives 1057 to generate 1255 and 1256, respectively. Next, 1210 receives 1254.1063, deletes the data of the insignificant estimation area, and generates compressed data 1058. Similarly, 1211 receives 1255 and 1065, and 1212 receives 1256 and 1065, generates 1059 and 1060, respectively, and deletes data in the insignificant estimation area. With the above operation, the image data is compressed.

Next, with reference to FIG. 15, the principle of data compression by deleting an insignificant space and decoding of the compressed data will be described. In FIG. 15, 1601, 1602, 1603,
Reference numerals 1604, 1605, 1606, 1607, 1608, 1609, and 1610 denote wavelet spaces, and each wavelet data has a strong correlation between adjacent layers. For example, when the value of the arbitrary region 1652 included in the region 1606 which is the wavelet space region of the layer 1 is small, the region 1652 included in the wavelet space region 1603 of the stage 0 corresponding to the wavelet space region 1606 of the layer 1 There is a high probability that the value of the region 1651 which is a similar region to is also small. Therefore, area 16
If the data value of 52 is smaller than a certain threshold value, assuming that the similarity area 1651 has a predetermined minimum value,
If there is only the data in the area 1652, the data in the area 1652 is not necessary for decoding. Therefore, at the time of encoding, if the data value of the area 1652 is equal to or smaller than an arbitrary threshold, the compression effect can be obtained by deleting the data of the area 1651 from the data of the area 1603. At the time of decoding, if the data value of the area 1652 is equal to or less than the threshold value used at the time of encoding, the area 1651
By filling the data with the minimum value, an approximate space of the original wavelet space can be reconstructed.

Next, the operation of the conventional wavelet image decompression apparatus will be described with reference to FIG. In FIG.
1102 is the layer 2 wavelet data 1051, 1052, 1053 compressed by the wavelet image compression apparatus.
Upon receiving 1054, inverse wavelet transform is performed to generate 1158. Next, the insignificant space estimation and development device 1101
Upon receiving 53.1054, 1055, 1056, 1057, 1058, 1059, and 1060, the data compressed by the insignificant space estimation and deletion device 1005 is decoded, and 1152, 1153, 1154, 1155, 1156, 11
Generate 57. Next, 1103 receives the layer 1 wavelet data 1158, 1152, 1153, and 1154 and performs inverse wavelet transform to generate 1159. Finally, 1104 is the wavelet data 1159-1155 / 1156/1157 of layer 0
Then, inverse wavelet transform is performed to generate 1151, and decoding is completed. By the above operation, the image data is expanded.

Next, the operation of the subband synthesizing apparatus will be described with reference to FIG. In FIG. 14, a vertical upsampler 1501 receives 1552, performs upsampling in the vertical direction, and generates 1556. Similarly, 1501 is 1153
In response to 1154 and 1155, 1557, 1558 and 1559 are generated, respectively. Next, the vertical low-pass filter 1502 receives 1556, and the vertical high-pass filter 1503 receives 1557, and performs filtering inverse mapping transform in the vertical direction to generate 1560.
Similarly, 1502 receives 1558 and 1503 receives 1559 to generate 1561. Next, the horizontal upsampler 1504
Received a 1560 and performed a horizontal upsample, 15
Generate 62. Similarly, 1504 receives 1561 and receives 15
Generate 63. Next, the horizontal low-pass filter 1505 is 1562
The horizontal high-pass filter 1506 receives 1563
Horizontal filtering inverse mapping transformation is performed to generate 1551.

Next, the operation of the insignificant space estimation / expansion apparatus 1101 will be described with reference to FIG. In FIG. 12, the insignificant space estimating unit 1201 is the data compressed by deleting the insignificant space.
Receiving 1052, the minimum value area estimation of 1055 is performed by threshold judgment, and 1351 is generated. Similarly, the insignificant space estimation unit 1202
Receiving 1053, insignificant space estimation section 1203 receives 1054 and generates 1352 and 1353, respectively. Next, the layer 1 development unit
1301 receives 1351 and 1055, and fills the minimum value in the meaningless estimation area to generate decompressed data 1152. Similarly, the layer 1 expanding unit 1302 receives 1352 · 1056, and
1303 receives 1353 · 1057 and generates 1153 and 1154, respectively. Next, the insignificant space estimation unit 1207 receives 1152,
The minimum value area estimation of 1058 is performed by the threshold value judgment, and 1354 is generated. Similarly, the insignificant space estimation unit 1208 receives 1153,
The insignificant space estimation unit 1209 receives 1154, and receives 1355,
Generate 1356. Next, the layer 0 decompression unit 1304 calculates 1354
Upon receiving 1058, the minimum value is filled in the meaningless estimation area to generate decompressed data 1155. Similarly, a layer 0 expanding unit 1305
Receives 1355 · 1059, the layer 0 decompression unit 1306 receives 1356 · 1059
Upon receiving 60, 1156 and 1157 are generated, respectively. With the above operation, the decoding of the wavelet data is performed.

[0021]

As described above, in the above-described conventional image data processing apparatus and image data processing method, image compression / expansion is performed using wavelet transform. Since the encoding process is performed without distinguishing between the high brightness area and the high brightness area, data is encoded with the same compression efficiency as the high brightness area where the visual characteristics are sensitive even in the low brightness area where the visual characteristics are insensitive. Therefore, if the insignificance determination threshold is set in the high luminance area in order to ensure high image quality, the encoded data has a problem that the high frequency components in the low luminance area have redundancy.

According to the present invention, there is provided an image data processing method for compressing / expanding image data with higher compression efficiency by removing high-frequency components in a low-luminance region in accordance with visual characteristics. Accordingly, it is an object of the present invention to provide an image data processing device capable of compressing / expanding image data with a higher compression efficiency than conventional by removing high frequency components in a low luminance region.

[0023]

In order to solve this problem, an image data processing method according to the present invention decomposes original image data into wavelet data by wavelet transform.
From the DC subband component of the wavelet data, it is determined whether or not the region is a compressible region based on a threshold value. A structure is provided in which an insignificant area to be code-deleted is estimated and deleted using the determination threshold.

Thus, there is provided an image data processing method for compressing image data with a higher compression efficiency than the conventional one by removing high-frequency components in a low-luminance region according to visual characteristics.

An image data processing apparatus according to the present invention for solving this problem decomposes original image data into wavelet data by wavelet transform, and determines whether or not the region is a compressible region based on a DC subband component of the wavelet data based on a threshold value. And when it is determined that the region is a compressible region, the insignificance determination threshold value of the relative square position of the wavelet data is set high, and the insignificant region to be subjected to code deletion is estimated and deleted from the wavelet data using the insignificance determination threshold value.

As a result, it is possible to obtain an image data processing apparatus capable of compressing image data with higher compression efficiency than the conventional one by removing high-frequency components in a low-luminance region according to visual characteristics.

[0027]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention is an image data processing method for compressing an original image, comprising a decomposition step of decomposing original image data into wavelet data by wavelet transform. When the value of the arbitrary square area of the DC sub-band component of the data is less than the threshold value, the setting process of setting the insignificance determination threshold value of the relative square area of the other sub-band components higher than when the value of the DC sub-band component is equal to or greater than the threshold value And an estimating step of estimating an insignificant area by determining an insignificant determination area of the wavelet data with an insignificance determination threshold, and a deleting step of deleting the insignificant area from the wavelet data. This has an effect that redundant data based on visual characteristics is deleted at the time of encoding, and the code amount is suppressed at the time of encoding.

According to a second aspect of the present invention, there is provided an image data processing method for decompressing compressed image data and restoring an original image, wherein the direct current of wavelet data in which an insignificant area, which is compressed image data, is deleted. When the value of the sub-band component is less than the threshold, the insignificance determination of the relative square position of the other sub-band components is set higher than when the value of the DC sub-band component is greater than or equal to the threshold; An estimation process of estimating an insignificant region by determining with a threshold, a restoration process of generating restoration wavelet data by filling insignificant data in an insignificant region of wavelet data, and an inverse wavelet transform of the restoration wavelet data to convert the original image data. And a reconstruction process for performing reconstruction. Redundant data deleted has an effect that the reconstructed original image data reconstructed in the decoding based on.

According to a third aspect of the present invention, there is provided an image data processing method for compressing an original image, comprising a decomposition step of decomposing the original image data into wavelet data by a wavelet transform. When the value of the arbitrary square area is less than the threshold, the insignificance determination threshold of the relative square area of the other subband components is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the horizontal component of the wavelet data is included. For the subband component, the insignificance determination threshold of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the vertical component of the wavelet data is set. For the subband component that has the DC subband, the insignificance determination threshold value of the relative rectangular area adjacent to both ends of the relative square area in the vertical direction is A setting step of setting the value of the component to be higher than a threshold or more, an estimation step of estimating an insignificant area by determining the wavelet data with an insignificance determination threshold, and a deleting step of deleting the insignificant area from the wavelet data, It is characterized by having
It has an effect that redundant data is deleted based on the visual characteristics and the characteristics of the wavelet function at the time of encoding, and the code amount is suppressed at the time of encoding.

According to a fourth aspect of the present invention, there is provided an image data processing method for decompressing compressed image data and restoring an original image, comprising the steps of: When the value of any square position of the subband component is less than the threshold, the insignificance determination threshold of the relative square area of the other subband components is set higher than when the value of the DC subband component is greater than or equal to the threshold, and the wavelet data For the subband component having the horizontal component of, the insignificance determination threshold of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area is set higher than when the value of the DC subband component is greater than or equal to the threshold, and the wavelet For the subband component having the vertical component of the data, the insignificance judgment threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area is set to the DC subband component. A setting process of the value is set higher than that in the above the threshold, the estimation process for estimating the meaningless area by determining the wavelet data in insignificant determination threshold,
It is characterized by comprising a restoring process of generating restoring wavelet data by embedding meaningless data in an insignificant region of wavelet data, and a reconstructing process of reconstructing original image data by inverse wavelet transform of the restored wavelet data. There is an effect that redundant data deleted based on visual characteristics and characteristics of a wavelet function at the time of encoding is restored in decoding and original image data is reconstructed.

According to a fifth aspect of the present invention, there is provided an image data processing apparatus for compressing an original image, comprising a sub-band decomposing device for decomposing original image data into wavelet data by wavelet transform. From the components, a determination is made as to whether or not the region is a compressible region based on a threshold value. If the region is determined to be a compressible region, the insignificance determination threshold value of the relative square position of the other subband components is set to a high value. It is characterized by comprising an insignificant space estimation and deletion device for estimating and deleting the insignificant region of the code deletion target, wherein redundant data based on visual characteristics is deleted at the time of encoding, and the code amount is reduced at the time of encoding. It has the effect of being suppressed.

According to a sixth aspect of the present invention, there is provided an image data processing apparatus for decompressing compressed image data and restoring an original image, wherein the direct current of the wavelet data in which the insignificant area which is the compressed image data is deleted. The subband component is used to determine whether or not it is a compressible region based on a threshold, and when it is determined that the region is a compressible region, the insignificant determination threshold of the relative square position of the other subband components is compared to when it is determined that the region is not a compressible region. Is also set high, estimating the insignificant region of the code removal target with the wavelet data and the insignificance determination threshold, filling the insignificant data in the insignificant region of the wavelet data and generating a meaningless space estimating and developing device,
And a sub-band synthesizer for performing inverse wavelet transform of the restored wavelet data, wherein redundant data deleted based on visual characteristics at the time of encoding is restored in decoding, and original image data is restored. It has the effect of being reconfigured.

According to a seventh aspect of the present invention, in the invention of the fifth aspect, when the insignificant space estimation and deletion device determines that the region is a compressible region, the device removes the subband component having the horizontal component of the wavelet data. In this case, the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area is set higher than when it is determined that the area is not a compressible area, and the subband component having the vertical component of the wavelet data is set. It is characterized in that the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area is set higher than when it is determined that the area is not a compressible area. There is an effect that redundant data is deleted based on the characteristics of the wavelet function, and the code amount is suppressed during encoding.

According to an eighth aspect of the present invention, in the invention of the sixth aspect, the insignificant space estimating and expanding device determines that the subband component having the horizontal component of the wavelet data is determined when the region is a compressible region. In this case, the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area is set higher than when it is determined that the area is not a compressible area, and the subband component having the vertical component of the wavelet data is set. It is characterized in that the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area is set higher than when it is determined that the area is not a compressible area. There is an effect that the redundant data deleted based on the characteristics of the wavelet function is expanded in decoding and the original image data is reconstructed.

Hereinafter, an embodiment of the present invention will be described with reference to FIG.
This will be described with reference to FIG. (Embodiment 1) FIG. 1 is a block diagram showing a wavelet image compression apparatus constituting an image data processing apparatus according to Embodiment 1 of the present invention.

In FIG. 1, reference numeral 1001 denotes an original image output device;
02 is a layer 0 sub-band decomposition device, 1003 is a layer 1 sub-band decomposition device, and 1004 is a layer 2 sub-band decomposition device. Reference numeral 1061 denotes original image data.

1062 is layer 0 DC data, 1063 is layer 0
HL data, 1064 is layer 0LH data, 1065 is layer 0HH
Data, 1066 is layer 1 DC data, 1067 is layer 1 HL data, 1068 is layer 1 LH data, and 1069 is layer 1 HH data. 1051 is layer 2 DC data, 1052 is layer 2HL
Data, 1053 is layer 2 LH data, and 1054 is layer 2 HH data. 151 is layer 1 HL compressed data, 152 is layer 1 LH compressed data, 153 is layer 1 HH compressed data, 154 is layer 0 HL compressed data, 155 is layer 0 LH compressed data, 156
Is compressed data of layer 0HH.

Reference numeral 101 denotes an insignificant space estimation and deletion device, which performs a masking process on each high-frequency component of layer 1 and layer 0 based on each data (1051, 1052, 1053, 1054) of layer 2. FIG. 2 is an explanatory diagram for explaining the effect of the mask processing of the image data processing apparatus according to the present invention. As shown in FIG. 2, the value will be zero in each space of the high frequency components of layer 1 and layer 0. By predicting the area and reducing the predicted area from the compressed data, the amount of data in the insignificant area can be reduced.

FIG. 5 is a block diagram showing the insignificant space estimation and deletion device 101 of FIG. In FIG. 5, reference numeral 301 denotes a layer 1H
L insignificant space estimating unit, 302 is a layer 1 LH insignificant space estimating unit, 303 is a layer 1HH insignificant space estimating unit, 304 is layer 0
The HL insignificant space estimation unit, 305 is a layer 0LH insignificant space estimation unit, and 306 is a layer 0HH insignificant space estimation unit. Reference numeral 1204 denotes a layer 1HL deletion unit, 1205 denotes a layer 1 LH deletion unit, and 1206 denotes a layer 1HH deletion unit. 1210 is a layer 0HL deletion unit,
Reference numeral 1211 denotes a layer 0 LH deletion unit, and reference numeral 1212 denotes a layer 0 HH deletion unit.

Reference numeral 351 denotes layer 1HL meaningless space estimation data, and 352
Is the layer 1 LH meaningless space estimation data, 353 is the layer 1HH meaningless space estimation data, 354 is the layer 0HL meaningless space estimation data, 355 is the layer 0LH meaningless space estimation data, and 356 is the layer 0HH meaningless space estimation data.

FIG. 3 is an explanatory diagram for explaining the mask processing in the insignificant space estimation and deletion device 101. The mask processing of the insignificant space estimation and deletion device 101 predicts, for example, the position of an element in each high-frequency component region of layer 0 based on the significant space data of layer 1 and forms a mask plane space 406 as shown in FIG. I do. The determination as to whether or not the element exists here is performed by a threshold determination as to whether or not there is an element that substantially exceeds a certain threshold (insignificant determination threshold). Therefore, the region not exceeding the threshold value is estimated to be an insignificant space as shown by the oblique lines in the mask plane space in FIG. Then, for example, the LH data (1064) of layer 0 is masked by this mask plane space, and only the remaining data becomes the target of compressed image data in the subsequent stage. Data in other space of layer 0 (HL: 1063, HH: 1065)
Is similarly masked in the mask plane space, and only the remaining data becomes the target of compressed image data in the subsequent stage.

In the present invention, the insignificance determination threshold is changed according to the value of the layer 2 DC data 1051. That is, if the value of the layer 2 DC data 1051 is lower than an arbitrary value, the insignificance determination threshold when creating the mask plane space is set higher than the data of the layer 2; otherwise, a lower value is set.

FIG. 4 is a block diagram showing a structure of a wavelet image decompression device constituting the image data processing device according to the first embodiment of the present invention. In FIG. 4, reference numeral 201 denotes an insignificant space estimation / expansion device, 202 denotes an expanded image output device, 1102 denotes a layer 0 sub-band synthesis device, 1103 denotes a layer 1 sub-band synthesis device, 1104 denotes a layer 2 sub-band synthesis device, and 251
Is expanded image data, 252 is layer 1 HL expanded data, 253 is layer 1 LH expanded data, 254 is layer 1 HH expanded data, 2
55 is layer 0HL expanded data, 256 is layer 0LH expanded data, 257 is layer 0HH expanded data, 258 is layer 0 DC expanded data, and 1158 is layer 1 DC expanded data.

FIG. 6 is a block diagram showing the insignificant space estimation / expansion device 201 of FIG. In FIG. 6, reference numeral 451 denotes layer 1H
L meaningless space development estimation data, 452 is layer 1 LH meaningless space development estimation data, and 453 is layer 1HH meaningless space development estimation data. 454 is layer 0HL insignificant space expansion estimation data, 4
Reference numeral 55 denotes layer 0 LH meaningless space expansion estimation data, and reference numeral 456 denotes layer 0 HH meaningless space expansion estimation data. 1301 is Layer 1HL
An expansion unit, 1302 is a layer 1 LH expansion unit, and 1303 is a layer 1 HH expansion unit. 1304 is a layer 0 HL development unit, 1305 is a layer 0
An LH expansion unit 1306 is a layer 0HH expansion unit.

FIG. 7 is an explanatory diagram of the meaningless space determination method according to the first embodiment of the present invention. In FIG. 7, reference numeral 1606 denotes a space where the layer 1 HL data 1067 exists, that is, layer 1 HL
Wavelet space. 1609 is layer 2 HL data 10
52 is the space in which it exists, that is, the layer 2 HL wavelet space. Reference numeral 1610 denotes a space where the layer 2 DC data 1051 exists, that is, a layer 2 DC wavelet space.

In FIG. 7, the layer 2 DC wavelet space
A description will be given focusing on a part of the layer 2 DC data 1051 of 1610. Reference numeral 501 denotes DC component attention data in the layer 2 DC wavelet space 1610. Reference numeral 502 denotes layer 1 HL estimation reference data. 503 is layer 0HL estimation reference data (0,0),
504 is layer 0HL estimation reference data (1, 0), 505 is layer 0H
L estimation reference data (0,1) and 506 are layer 0HL estimation reference data (1,1). 507 is Layer 1 LH estimation reference data, 509 is Layer 0 LH estimation reference data (0, 0), 510 is Layer 0 LH estimation reference data (1, 0), 511 is Layer 0 LH estimation reference data (0, 1), 512 Is the layer 0 LH estimation reference data (1, 1).

Next, the operation of the wavelet image compression apparatus of FIG. 1 will be described. In FIG. 1, assuming that original image data 1061 is given, a layer 0 sub-band decomposition device 1002
Performs the wavelet transform, and the wavelet data (subband data) 1062, 1063, 1064, and 10 of layer 0
Generate 65 (decomposition process). Next, the layer 1 sub-band decomposition apparatus 1003 receives the signal 1062 and performs wavelet transform to generate layer 1 wavelet data (sub-band data) 1066, 1067, 1068, and 1069 (decomposition process).
Next, the layer 2 sub-band decomposition device 1004 receives the 1066 and performs a wavelet transform, and the layer 2 wavelet data (sub-band data) 1051, 1052, 1053, 1054
(Decomposition process).

Next, the details of the operation (setting process, estimating process, deleting process) of the insignificant space estimation deleting device 101 in FIG. 5 will be described. In FIG. 5, the insignificant space estimation unit 301 of the layer 1HL
Receives the layer 2 DC data 1051, and if it is lower than an arbitrary value, sets a higher value for the insignificance determination threshold of the layer 2 HL data 1052 used for the insignificance determination of the layer 1 HL data 1067, and otherwise sets a lower value. I do. If the value of the layer 2 HL data 1052 is lower than this insignificance determination threshold, the layer 1
The HL data 1067 is determined to be insignificant, otherwise it is determined to be significant, and the layer 1 HL insignificant space estimation data 351 is generated. Similarly, the insignificant space estimation unit 302 of the layer 1 LH
Receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and determines whether the value of the layer 2 LH data 1053 is lower than the insignificance determination threshold.
The significance is determined, and the layer 1 LH meaningless space estimation data 352 is generated. Similarly, the insignificant space estimation unit 30 of the layer 1HH
3 receives the layer 2 DC data 1051, sets the insignificance determination threshold based on whether it is lower than an arbitrary value,
Insignificant by whether the value of data 1054 is lower than the insignificance determination threshold
Judgment of significance is performed, and layer 1HH insignificant space estimation data 353
Generate

Next, the layer 1 deleting unit 1204
Upon receiving the HL insignificant space estimation data 351 and the HL data 1067, if the estimation data 351 indicates insignificance, the data is deleted from the data 1067 to generate the layer 1 HL compressed data 151. Similarly, layer 1 deletion section 1205 receives layer 1 LH insignificant space estimation data 352 and LH data 1068, and if estimation data 352 indicates insignificance, deletes from data 1068 to remove layer 1 LH.
The compressed data 152 is generated. Similarly, the layer 1 deleting unit 1206 receives the layer 1 HH insignificant space estimation data 353 and the HH data 1069, and if the estimation data 353 indicates insignificance, deletes it from the data 1069 and deletes the layer 1 HH compressed data. Generate 153.

Next, layer 0HL insignificant space estimating section 304 receives layer 2 DC data 1051, and if it is lower than an arbitrary value, determines insignificance of layer 1HL compressed data 151 used for insignificant estimation of layer 0HL data 1063. Set the threshold high, otherwise set it low. If the value of the layer 1 HL compressed data 151 is lower than this insignificance determination threshold or if the layer 1 HL compressed data 151 has been deleted by the layer 1 HL deletion unit 1204, it is determined that the layer 0 HL data 1063 is insignificant, otherwise significant. And the layer 0HL insignificant space estimation data 354 is generated.

Similarly, the insignificant space estimation unit of layer 0LH
305 receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and determines whether the value of the layer 1 LH compressed data 152 is lower or the layer 1 LH compressed data 152 Insignificant / significant judgment is made based on whether or not it has been deleted by 1LH deletion unit 1205,
The layer 0 LH meaningless space estimation data 355 is generated.

Similarly, the insignificant space estimation unit of layer 0HH
306 receives the layer 2 DC data 1051 and sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and the value of the layer 1 HH compressed data 153 is lower than this insignificance determination threshold or the layer 1 HH compressed data 153 is Layer 1HH deletion unit 1206
The meaning of insignificance / significance is determined based on whether or not it has been deleted, and layer 0HH insignificant space estimation data 356 is generated.

Next, the layer 0 deletion unit 1210 outputs the
In response to the insignificant space estimation data 354 and the layer 0HL data 1063, if the layer 0HL insignificant space estimation data 354 indicates insignificance, it is deleted from the data of the layer 0HL data 1063,
The layer 0HL compressed data 154 is generated. Similarly, the layer 0 deletion unit 1211 outputs the layer 0 LH insignificant space estimation data 35
5 and layer 0 LH data 1064, if layer 0 LH insignificant space estimation data 355 indicates insignificance, layer 0 LH
Deleted from the data of data 1064, layer 0 LH compressed data
Generate 155. Similarly, the layer 0 deletion unit 1212
In response to the layer 0HH insignificant space estimation data 356 and 1065, if the layer 0HH insignificant space estimation data 356 indicates insignificance, the layer 0HH data 1065 is deleted from the data, and the layer 0HH compressed data 156 is generated. Since the insignificant data of each of the compressed data 151, 152, 153, 154, 155, 156 has been deleted by the above operation, it means that the data has been compressed.

Here, a supplementary description of the operation of the meaningless space estimation and deletion device 101 will be given with reference to FIG. Attention is paid to the data of an arbitrary part of the DC space generated by the wavelet transform, that is, the layer 2 DC wavelet space 1610, that is, the DC component attention data 501. If the value of the data is low, the insignificance determination threshold for the corresponding area (layer 1 HL estimation reference data 502) in the HL space 1609 of layer 2 is increased. Similarly, for the LH space, the layer 2 LH space
The corresponding area in 1608 (layer 1 LH estimation reference data 50
By increasing the insignificance determination threshold for 7), and similarly for the HH space, by increasing the insignificance determination threshold of the corresponding area (layer 1 HH estimation reference data 508) in the layer 2 HH space 1607, the visual characteristics are matched. The amount of data is reduced by removing high-frequency components in low-luminance portions. That is, a low data value in the partial area (501) of the DC space area (layer 2 DC wavelet space 1610) means low luminance, and in this case, the level of the high-frequency component to be deleted is set high. .

Next, the operation of the wavelet image decompression device shown in FIG. 4 will be described. In FIG. 4, the insignificant space estimation / expansion / decompression device 201 includes subband data 1051, 1052, and 10 of layer 2 compressed by the wavelet image compression device in FIG.
In response to 53.1054.151.152.153.154.155.156, layer 1 subband data 252.253.254 and layer 0
The subband data 255, 256, and 257 are decoded. Next, the layer 0 sub-band combining apparatus 1102 receives the layer 2 sub-band data 1051, 1052, 1053, and 1054, performs sub-band combining, and performs layer 1D as layer 1 sub-band data.
The C expansion data 1158 is generated (reconstruction process). Next, the layer 1 sub-band synthesizing device 1103 receives the layer 1 sub-band data 1158 and each of the expanded data 252, 253, and 254 of the layer and performs sub-band synthesis to generate layer 0 sub-band data 258 (reconstruction process). ). Next, the layer 0 subband synthesizing device 1104 receives the DC expansion data 258 of the layer 0 and each of the expansion data 255, 256, and 257 to perform subband synthesis to generate expanded image data 251 (reconstruction process). By the above operation, the image data is expanded.

Next, the operation (setting process, restoring process) of the insignificant space estimation / expansion device 201 of FIG. 6 will be described.

In FIG. 6, layer 1HL insignificant space estimating section 301 receives layer 2 DC data 1051, and if it is lower than an arbitrary value, 1052 insignificance determination threshold used for insignificant estimation of layer 1HL compressed data 151. Set high, otherwise set a low value. If the value of the layer 2 HL data 1052 is lower than the insignificance determination threshold, the layer 1 HL compressed data 151 is determined to be ineffective data, otherwise it is determined to be significant, and the layer 1 HL insignificant space expansion estimation data 45
Generates 1. Similarly, the insignificant space estimation unit 302 of the layer 1 LH receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether or not it is lower than an arbitrary value, and
Insignificance / significance is determined based on whether the value of the LH data 1053 is lower than the insignificance determination threshold, and layer 1 LH insignificant space development estimation data 452 is generated. Similarly, the insignificant space estimation unit 303 of the layer 1HH receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and sets the value of the layer 2HH data 1054 to be less than the insignificance determination threshold. Insignificant / significant judgment is made depending on whether it is low, and layer 1HH insignificant space development estimation data 453 is generated.

Next, the layer 1 HL developing unit 1301
HL insignificant space expansion estimation data 451 and HL compressed data 151, and if the HL insignificant space expansion estimation data 451 indicates insignificant data, an insignificant value is given to the layer 1 HL expansion data 252 to indicate significance. Then, the value of the HL compressed data 151 is given to the layer 1 HL expanded data 252, and the restored layer 1H
The L expansion data 252 is generated.

Similarly, the layer 1 LH decompression unit 1302 transmits the LH insignificant space decompression estimation data 452 and the LH compressed data 152
In response, the restored layer 1 LH expanded data 253 is generated. Similarly, the layer 1 HH developing unit 1303
Receiving the HH insignificant space expansion estimation data 453 and the HH compression data 153, it generates restored layer 1 expansion HH data 254.

Next, the insignificant space estimation unit 304 of the layer 0HL
Receives the layer 2 DC data 1051, and if it is lower than an arbitrary value, sets the insignificance determination threshold of the layer 1 HL decompressed data 252 used for the innocent estimation of the layer 0 HL compressed data 154 to be high; Set. If the value of the layer 1 HL expanded data 252 is lower than this insignificance determination threshold, the layer 0 HL compressed data 154 is determined to be ineffective data, otherwise it is determined to be significant and the layer 0 HL insignificant space expanded estimation data 454 is determined. Generate.

Similarly, the insignificant space estimation unit of layer 0LH
A layer 305 receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and
Depending on whether the value of the expansion data 253 is lower than the insignificance determination threshold,
Insignificance / significance is determined, and layer 0 LH insignificant space development estimation data 455 is generated. Similarly, the insignificant space estimation unit 306 of the layer 0HH receives the layer 2 DC data 1051, sets an insignificance determination threshold based on whether it is lower than an arbitrary value, and determines whether the value of the layer 1HH expanded data 254 is insignificant. Whether it is insignificant or significant is determined based on whether it is lower than the threshold value, and layer 0HH insignificant space development estimation data 456 is generated.

Next, the layer 0 HL development unit 1304
Receiving the HL insignificant space expansion estimation data 454 and the HL compressed data 154, if the HL insignificant space expansion estimation data 454 indicates insignificant data, gives an insignificant value to the layer 0 HL expansion data 255, indicating significance. Then layer 0 HL compressed data 15
The value of 4 is given to the layer 0 HL expansion data 255, and the restored layer 0 HL data 255 is generated. Similarly, layer 0L
The H expansion unit 1305 generates LH insignificant space expansion estimation data 4 of layer 0.
55 and the LH compressed data 155, and the restored layer 0LH
Generate data 256. Similarly, layer 0HH development section 1
306 receives the HH insignificant space expansion estimation data 456 of the layer 0 and the HH compressed data 156, and generates the restored layer 0 HH data 257.

As described above, according to the present embodiment, the subband decomposing devices 1002, 1003, and 1004 that decompose image data into subbands by wavelet transform can be compressed by a threshold from the DC space of the decomposed subbands. A determination is made as to whether or not the region is a compressible region, and if it is determined that the region is a compressible region, the insignificance determination threshold of the relative square position of the decomposed wavelet space is set high, and the insignificance of the code deletion target is determined by the insignificance determination threshold from the insignificance determination space. By providing the insignificant space estimation deletion device 101 that estimates and deletes a space, redundant data based on visual characteristics can be deleted at the time of encoding, so that the amount of code can be suppressed at the time of encoding. it can.

Also, it is determined from the DC sub-band component of the wavelet data from which the insignificant space has been deleted whether or not it is a compressible area by using a threshold. If it is determined that the area is compressible, the insignificant determination threshold of the relative square position is set. Set higher than when it is determined that it is not a compressible area, estimate the insignificant space to be code-deleted with the insignificant decision space and the insignificant decision threshold, and fill in the insignificant space with the insignificant data and estimate other than the insignificant space estimated The significant space of the insignificant space is directly expanded wavelet data is deleted, the insignificant space estimation and expansion device 201 that generates the restored wavelet data, and the sub-band synthesizer 1102, 1103, and 1104 that performs the inverse wavelet transform of the restored wavelet data. , The redundant data deleted based on the visual characteristics at the time of encoding It is possible to reconstruct the original by the original image data.

(Embodiment 2) FIGS. 8A and 8B are explanatory diagrams of a function (setting process) of adding a significant attribute in an image data processing apparatus according to Embodiment 2 of the present invention.

In FIG. 8, reference numeral 601 denotes layer 0HL vertical additional estimation reference data (0, -1), 602 denotes layer 0HL vertical additional estimation reference data (1, -1), and 603 denotes layer 0HL vertical additional estimation reference data (0). , 2) and 604 are reference data for layer 0HL vertical addition estimation
(1, 2), 605 are layer 0HL horizontal estimation reference data (-1, 0), 60
Reference numeral 6 denotes layer 0HL horizontal estimation reference data (2, 0), reference numeral 607 denotes layer 0HL horizontal estimation reference data (-1, 1), and reference numeral 608 denotes layer 0HL horizontal estimation reference data (2, 1).

Vertical component data 16 after wavelet transform
The pattern shown in FIG. 8A having the long side in the vertical direction is applied to 06, and the pattern shown in FIG. 8B having the long side in the horizontal direction is applied to the horizontal component data 1605 after the wavelet transform. The compression rate is further increased because the area to be compressed is increased.

As described above, according to the present embodiment, when the insignificant space estimation and deletion device 101 determines that the region is a compressible region, the insignificant space having the decomposed horizontal component has the horizontal position of the relative square position. The insignificance determination threshold of the relative rectangular position adjacent to both ends of the direction is set higher than when it is determined that the region is not a compressible area, and the insignificance determination space having the decomposed vertical component is located at both ends in the vertical direction of the relative square position. By setting the insignificance determination threshold value of the adjacent relative rectangular position higher than when it is determined that the region is not a compressible region, redundant data can be deleted based on the visual characteristics and the characteristics of the wavelet function during encoding. Since it is possible, the amount of code can be further reduced at the time of encoding.

When the insignificant space estimating and expanding device 201 determines that the region is a compressible region, the insignificant determination space having the decomposed horizontal component has a relative rectangular position adjacent to both ends in the horizontal direction of the relative square position. Is set higher than when it is determined that the region is not the compressible area, and the insignificance determination threshold of the relative rectangular position adjacent to both ends in the vertical direction of the relative square position is set in the insignificance determination space having the decomposed vertical component. Is set higher than when it is determined that the region is not a compressible region, the redundant data deleted at the time of encoding is restored based on the visual characteristics and the characteristics of the wavelet function in decoding, and the original image data is reconstructed. can do.

[0070]

As described above, according to the image data processing method according to the first aspect of the present invention, the decomposition process of decomposing the original image data into wavelet data by the wavelet transform, and the direct current of the decomposed wavelet data When the arbitrary square area value of the sub-band component is less than the threshold, the setting process of setting the insignificance determination threshold of the relative square position of the other sub-band components higher than when the value of the DC sub-band component is equal to or more than the threshold, An estimation process of estimating the insignificant region by determining the insignificant region of the insignificance threshold,
And deleting the insignificant area from the wavelet data, thereby obtaining an advantageous effect that redundant data based on visual characteristics is deleted at the time of encoding, and the code amount is suppressed at the time of encoding. .

According to the second aspect of the present invention, when the value of the DC subband component of the wavelet data from which the insignificant region has been deleted is less than the threshold value, the insignificance determination threshold value of the relative square position of the other subband components is set to DC. A setting step of setting the value of the sub-band component to be higher than the threshold value or more, an estimation step of estimating the insignificant area by determining the wavelet data with the insignificant determination threshold, and embedding the insignificant data in the insignificant area of the wavelet data. And a reconstruction step of reconstructing image data by performing an inverse wavelet transform on the restored wavelet space, thereby eliminating the restoration wavelet data based on visual characteristics at the time of encoding. The advantageous effect is obtained that the redundant data is restored in the decoding and the original image data is reproduced.

According to the third aspect of the present invention, when the decomposition process of decomposing the original image data into sub-bands by the wavelet transform, and when the value of the arbitrary square area of the DC sub-band component of the wavelet data is smaller than the threshold value, The insignificance determination threshold of the relative square position of the other subband components is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the horizontal band of the relative square region is set for the subband component having the horizontal component of the wavelet data. The insignificance determination threshold of the relative rectangular area adjacent to both ends in the direction is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the relative square area is used for the subband component having the vertical component of the wavelet data. To set the insignificant judgment threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the vertical direction higher than when the value of the DC subband component is equal to or greater than the threshold value And an estimating step of estimating an insignificant region of the wavelet data by using an insignificant determination threshold, and a deleting process of deleting the insignificant region from the wavelet data. Based on the above characteristic, redundant data is deleted, and an advantageous effect that the code amount is further suppressed at the time of encoding is obtained.

According to the fourth aspect of the present invention, when the value of the DC subband component of the wavelet data from which the insignificant space has been deleted is less than the threshold value, the insignificance determination threshold value of the relative square area of the other subband components is set. The value of the DC subband component is set higher than the threshold value. For the subband component having the horizontal component of the wavelet data, the insignificance judgment threshold value of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area Is set higher than when the value of the DC subband component is greater than or equal to the threshold, and for the subband component having the vertical component of the wavelet data, the insignificant determination of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area Setting the threshold higher than when the value of the DC subband component is equal to or greater than the threshold, and determining the insignificant region of the wavelet data with the insignificant determination threshold An estimating process for estimating the reconstructed wavelet data, a restoring process of generating the restored wavelet data by filling the insignificant data in the insignificant region of the wavelet data, a reconstruction process of reconstructing the original image data by performing an inverse wavelet transform on the restored wavelet space, Has an advantageous effect that the redundant data deleted based on the visual characteristics and the characteristics of the wavelet function at the time of encoding is restored in decoding and the original image data is reconstructed.

According to the image data processing apparatus of the fifth aspect of the present invention, a sub-band decomposition apparatus for decomposing image data into wavelet data by wavelet transform, and a compressible area based on a DC sub-band component of the wavelet data by a threshold value When it is determined that the region is a compressible region, the insignificant determination threshold of the relative square position of the other subband components is set high, and the insignificant region of the code deletion target is determined from the wavelet data by the insignificant determination threshold. With the provision of the insignificant space estimation and deletion device that estimates and deletes, redundant data based on visual characteristics can be deleted at the time of encoding, so that the amount of code can be suppressed at the time of encoding. Effects can be obtained.

According to the sixth aspect of the present invention, it is determined from the DC subband component of the wavelet data from which the insignificant space has been deleted whether or not the region is a compressible region by using a threshold value. Is set higher than when the insignificance determination threshold of the relative square position of the other sub-band component is determined not to be a compressible region, and estimates the insignificant region of the code removal target with the wavelet data and the insignificance determination threshold,
By having an insignificant space estimation and expansion device that buries the insignificant data in the insignificant region of the wavelet data to generate the restored wavelet data, and a subband synthesizer that performs the inverse wavelet transform of the restored wavelet data,
The advantage is obtained that the redundant data deleted based on the visual characteristics at the time of encoding can be restored in decoding and the original image data can be reproduced.

According to the invention of claim 7, according to claim 5,
In the invention described in the above, the insignificant space estimation deletion device determines that the subband component having the decomposed horizontal component is a relative length adjacent to both ends in the horizontal direction of the relative square region when it is determined that the region is a compressible region. The insignificance determination threshold of the square area is set higher than when it is determined that the area is not a compressible area, and the relative rectangular area adjacent to both ends in the vertical direction of the relative square area for the subband component having the decomposed vertical component By setting the insignificance determination threshold of と き higher than when it is determined that the region is not a compressible region, redundant data can be deleted based on the visual characteristics and the characteristics of the wavelet function at the time of encoding. There is an advantageous effect that the code amount can sometimes be further suppressed.

According to the invention of claim 8, according to claim 6,
In the invention described in the above, the insignificant space estimation and expansion device, when it is determined that it is a compressible region, the relative rectangular region adjacent to both ends in the horizontal direction of the relative square region with respect to the subband component having the decomposed horizontal component The insignificance determination threshold is set higher than when it is determined that the region is not a compressible region, and the insignificance determination region for the subband component having the decomposed vertical component has a relative rectangular region adjacent to both ends in the vertical direction of the relative square region. By setting the insignificance determination threshold higher than when it is determined that the region is not the compressible region, the redundant data deleted based on the visual characteristics and the characteristics of the wavelet function at the time of encoding is expanded in decoding to obtain the original image data. Can be reproduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a wavelet image compression device included in an image data processing device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram for explaining the effect of the mask processing of the image data processing apparatus according to the present invention;

FIG. 3 is an explanatory diagram for explaining mask processing of the image data processing apparatus according to the present invention;

FIG. 4 is a block diagram showing a structure of a wavelet image decompression device constituting the image data processing device according to the first embodiment of the present invention;

FIG. 5 is a block diagram showing the insignificant space estimation and deletion device of FIG. 1;

FIG. 6 is a block diagram showing an insignificant space estimation and development device in FIG. 4;

FIG. 7 is an explanatory diagram of a meaningless space determination method according to the first embodiment of the present invention.

8A is a diagram illustrating a function of adding a significant attribute in an image data processing device according to a second embodiment of the present invention; FIG. 8B is a diagram illustrating a function of adding a significant attribute in an image data processing device according to a second embodiment of the present invention; Illustration

FIG. 9 is a block diagram showing the overall configuration of a conventional wavelet image compression apparatus.

FIG. 10 is a block diagram showing an overall configuration of a conventional wavelet image decompression device.

FIG. 11 is a block diagram showing a conventional meaningless space estimation and deletion device.

FIG. 12 is a block diagram showing a conventional insignificant space estimation and development device.

FIG. 13 is a block diagram showing a sub-band decomposition device.

FIG. 14 is a block diagram showing a sub-band combining device.

FIG. 15 is an explanatory diagram of a wavelet space.

[Explanation of symbols]

 101 Insignificant Space Estimation Deletion Device 201 Insignificant Space Estimation Expansion Device 202 Decompressed Image Input Device 301 Layer 1HL Insignificant Space Estimator 302 Layer 1LH Insignificant Space Estimator 303 Layer 1HH Insignificant Space Estimator 304 Layer 0HL Insignificant Space Estimator 305 Layer 0 LH meaningless space estimation unit 306 Layer 0 HH meaningless space estimation unit 1001 Original image output device 1002 Layer 0 subband decomposition device 1003 Layer 1 subband decomposition device 1004 Layer 2 subband decomposition device 1102 Layer 0 subband synthesis device 1103 layer 1 subband combining device 1104 Layer 2 subband combining device 1201 Involuntary space estimating unit from Layer 2 HL space 1202 Insignificant space estimating unit from Layer 2 LH space 1203 Insignificant space estimating unit from Layer 2 HH space 1204 Layer 1HL deleting unit 1205 Layer 1 LH Deletion unit 1206 Layer 1HH deletion unit 1207 Insignificant space estimation unit from Layer 1 HL space 1208 Layer 1HH deletion unit from Layer 1 LH space Unintended space estimation unit 1209 Unintended space estimation unit from layer 1 HH space 1210 Layer 0HL deletion unit 1211 Layer 0 LH deletion unit 1212 Layer 0HH deletion unit 1301 Layer 1HL expansion unit 1302 Layer 1 LH expansion unit 1303 Layer 1HH expansion unit 1304 Layer 0HL expansion unit 1305 Layer 0 LH expansion section 1306 Layer 0 HH expansion section 1401 Horizontal low pass filter 1402 Horizontal high pass filter 1403 Horizontal down sampler 1404 Vertical low pass filter 1405 Vertical high pass filter 1406 Vertical down sampler 1501 Vertical up sampler 1502 Vertical low pass filter 1503 Vertical high pass filter 1504 Horizontal up sampler 1505 Horizontal low-pass filter 1506 Horizontal high-pass filter

Claims (8)

[Claims] 1. An image data processing method for compressing an original image, comprising a decomposition step of decomposing original image data into wavelet data by wavelet transform, wherein a value of an arbitrary square area of a DC subband component of the wavelet data is A setting step of setting the insignificance determination threshold of the relative square area of the other subband components to be higher than when the value of the DC subband component is equal to or greater than the threshold, and setting the insignificance determination area of the wavelet data to An image data processing method, comprising: an estimation step of estimating an insignificant area by making a determination using a determination threshold; and a deleting step of deleting the insignificant area from the wavelet data. 2. An image data processing method for decompressing compressed image data and restoring an original image, wherein a value of a DC sub-band component of wavelet data from which insignificant regions, which are compressed image data, are deleted, is set to a threshold value. If it is less than the setting step of setting the insignificance determination threshold value of the relative square position of the other subband components higher than when the value of the DC subband component is equal to or greater than the threshold value, the wavelet data is determined by the insignificance determination threshold value The estimation process of estimating the insignificant region by the above process, the restoration process of generating restoration wavelet data by filling the insignificant region in the insignificant region of the wavelet data, and the inverse wavelet transform of the restoration wavelet data restores the original image data. And a reconstructing step. 3. An image data processing method for compressing an original image, comprising a decomposition step of decomposing original image data into wavelet data by wavelet transform, wherein a value of an arbitrary square area of a DC subband component of the wavelet data is When the value is less than the threshold value, the insignificance determination threshold value of the relative square area of the other subband components is set higher than when the value of the DC subband component is equal to or greater than the threshold value, and the subband component having a horizontal component of the wavelet data is set. For the relative square area adjacent to both ends in the horizontal direction of the relative square area, the insignificance determination threshold is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the vertical component of the wavelet data is included. For the sub-band component, the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area is set to the DC A setting step of setting the value of the band component higher than a threshold value or more; an estimation step of estimating an insignificant area by determining the wavelet data with the insignificance determination threshold; and deleting the insignificant area from the wavelet data. An image data processing method, comprising: a deleting step. 4. An image data processing method for decompressing compressed image data and restoring an original image, wherein an arbitrary square position of a DC subband component of wavelet data from which an insignificant region, which is compressed image data, is deleted. Is smaller than the threshold value, the insignificance determination threshold value of the relative square area of the other subband components is set higher than when the value of the DC subband component is equal to or greater than the threshold value, and the sub-component having the horizontal component of the wavelet data is set. For the band component, the insignificance determination threshold of the relative rectangular area adjacent to both ends in the horizontal direction of the relative square area is set higher than when the value of the DC subband component is equal to or greater than the threshold, and the vertical of the wavelet data is set. For the sub-band component having the component, the insignificance determination threshold value of the relative rectangular area adjacent to both ends in the vertical direction of the relative square area is set to the DC sub-band component. A setting step of setting a value higher than when the value is equal to or greater than a threshold, an estimation step of estimating an insignificant area by determining the wavelet data with the insignificance determination threshold, and embedding and restoring insignificant data in the insignificant area of the wavelet data. An image data processing method comprising: a restoration step of generating wavelet data; and a reconstruction step of reconstructing original image data by performing an inverse wavelet transform of the restored wavelet data. 5. An image data processing apparatus for compressing an original image, comprising: a sub-band decomposer for decomposing original image data into wavelet data by wavelet transform;
It is determined whether or not a compressible region from the DC subband component of the wavelet data by a threshold, and when it is determined that the compressible region is set, the insignificant determination threshold of the relative square position of the other subband components is set high, An image data processing device, comprising: an insignificant space estimation and deletion device for estimating and deleting an insignificant region to be deleted from the wavelet data using the insignificance determination threshold. 6. An image data processing apparatus for decompressing compressed image data and restoring an original image, comprising compressing a DC sub-band component of wavelet data from which insignificant regions, which are compressed image data, have been deleted, using a threshold value. A determination is made as to whether or not the region is a compressible region, and when it is determined that the region is a compressible region, the insignificance determination threshold value of the relative square position of the other subband components is set higher than when it is determined that the region is not a compressible region. An insignificant space estimation and expansion device that estimates an insignificant region of a code to be deleted with the data and the insignificance determination threshold, fills the insignificant data in the insignificant region of the wavelet data to generate restored wavelet data, and converts the restored wavelet data into an inverse wavelet. An image data processing device comprising: a sub-band synthesizing device for performing conversion. 7. The insignificant space estimation and deletion device, when determining that the region is a compressible region, adjoins both ends in the horizontal direction of the relative square region with respect to a subband component having a horizontal component of the wavelet data. The insignificance determination threshold of the relative rectangular region is set higher than when it is determined that the region is not a compressible region, and the subband component having the vertical component of the wavelet data is adjacent to both ends in the vertical direction of the relative square region. The image data processing apparatus according to claim 5, wherein the insignificance determination threshold value of the relative rectangular area is set higher than when it is determined that the area is not a compressible area. 8. The insignificant space estimating / decompressing device, when judging that the region is a compressible region, adjoins both ends in the horizontal direction of the relative square region with respect to a subband component having a horizontal component of the wavelet data. The insignificance determination threshold of the relative rectangular region is set higher than when it is determined that the region is not a compressible region, and the subband component having the vertical component of the wavelet data is adjacent to both ends in the vertical direction of the relative square region. The image data processing apparatus according to claim 6, wherein the insignificance determination threshold value of the relative rectangular area is set higher than when it is determined that the area is not a compressible area.