WO2003041394A1

WO2003041394A1 - Image compression apparatus, image compression program, and image compression method

Info

Publication number: WO2003041394A1
Application number: PCT/JP2002/011353
Authority: WO
Inventors: Hideyasu Kuniba
Original assignee: Nikon Corporation
Priority date: 2001-11-05
Filing date: 2002-10-31
Publication date: 2003-05-15
Also published as: JP4065978B2; JPWO2003041394A1

Abstract

An image compression apparatus for dividing image data into a plurality of tiles and performing image compression for each of the tiles. The apparatus includes a statistical processing unit, a parameter setting unit, and an image compression unit. The statistical processing unit processes image data to obtain a statistical feature correlating with “noticeability of a gradation shift when a gradation shift is caused in the image data”. According to this statistical feature, the parameter setting unit predicts noticeability of the tile boundary upon image decompression and adjusts “an image compression parameter affecting the amount of compression distortion”. The image compression unit performs image compression of image data by using the image compression parameter thus adjusted.

Description

Specification

FIELD OF THE INVENTION The present invention relates to an image compression apparatus, an image compression program, and an image compression method.

The present invention relates to an image compression device that divides image data into a plurality of tiles and performs image compression for each tile.

The present invention relates to an image compression program for causing a computer to function as an image compression device.

The present invention relates to an image compression method for compressing image data for each tile. Description of the background art

Conventionally, there has been known an image compression method in which image data is divided into small areas called tiles and image compression is performed for each of the tiles, as in JPEG200. This type of tile-based image compression has the advantage of saving work memory without having to deal with a large amount of image data at once.

Even if the screen size (the number of vertical and horizontal pixels) of the image data fluctuates, the screen size of each tile can be kept substantially constant by changing the number of tiles. In this case, there is an advantage that it is possible to flexibly cope with various screen sizes without significantly changing the image compression processing in units of tiles.

Furthermore, since image compression processing can be separated and executed for each tile, there is an advantage that parallel processing of image compression and multi-threading can be easily realized.

Due to the above advantages, tile-based image compression is an important technology in image compression technology.

It should be noted that, although different from image compression in tile units, Patent Document 1 below is known as a conventional technique for image compression in pixel block units, which are division ranges of orthogonal transform. In Patent Document 1, an activity indicating the degree to which a spatial frequency component is included is calculated for each pixel block, and the amount of compression code assigned to each pixel block is adjusted according to the activity.

(Patent Document 1) Japanese Unexamined Patent Publication No. Hei 2 _ 2 6 2 786 (US Pat. No. 5,085,184)

Meanwhile, in the above-described image compression on a tile-by-tile basis, image compression is individually performed on adjacent tiles. Therefore, the compression distortion also occurs independently for each tile, and the compression distortion is discontinuous at the tile boundary. As a result, the decompressed image had problems such as boundaries between tiles appearing as boundaries (hereinafter, these boundaries are referred to as “tile boundaries”).

In order to improve such a problem of the tile boundary, it is only necessary to perform image compression close to lossless compression to suppress the generation of compression distortion itself. However, if the compression distortion is suppressed more than necessary, the encoding efficiency of image compression will be reduced and the amount of compressed data will increase unnecessarily.

Further, in Patent Document 1 described above, the larger the activity of a pixel block, the greater the amount of compression code to be allocated. Therefore, in the flat part where the tile boundary is most conspicuous (the activity is almost zero), the amount of compression code allocated is small. As a result, in Patent Document 1, it is difficult to positively improve the tile boundary.

Further, in Patent Document 1, only the compression code amount is adjusted in units of pixel blocks, and the frequency distribution of the compression code amount (particularly, the code amount allocated to low-frequency components) is not changed. Therefore, at very high image compression ratios where tile boundaries are a real problem, it seems difficult to make fine adjustments such as suppressing image information degradation in the mid-high frequency range while improving tile boundaries. Disclosure of the invention

In view of the above-mentioned problems, an object of the present invention is to provide an image compression technique capable of appropriately suppressing tile boundaries from being inconspicuous at the time of image expansion.

Hereinafter, the present invention will be described.

An image compression apparatus according to the present invention is an image compression apparatus that divides image data into a plurality of tiles and performs image compression for each tile, and includes a statistical processing unit, a parameter setting unit, and an image compression unit.

The statistical processing unit processes the image data and displays a message “if a gradation shift occurs in the image data. And the statistical properties that correlate with the "conspicuousness of the gradation shift in the case."

The parameter setting unit predicts the conspicuousness of the tile boundary at the time of image expansion based on this statistical property, and sets the “image compression parameter that affects the appearance of compression distortion” according to the conspicuousness of the tile boundary. Adjust the settings.

The image compression unit compresses the image data using the image compression parameters adjusted and set as described above.

More preferably, in the present invention, the statistical property is a value indicating the flatness of the gradation of the image data. The parameter setting unit adjusts and sets the image compression parameter in such a direction as to reduce the compression distortion of the low-frequency spatial frequency component as the degree of flatness increases. Also preferably, in the present invention, the statistical property is a ratio of a gradation flat tile (tile that can be regarded as a flat gradation) to a plurality of tiles. The parameter setting unit adjusts and sets the image compression parameter in a direction to reduce the compression distortion of the low-frequency spatial frequency component as the ratio of the gradation flat tile increases.

More preferably, in the present invention, the statistical property is a variance obtained for a transform coefficient obtained by transforming image data into a spatial frequency domain. The parameter setting unit adjusts and sets the image compression parameter in such a direction as to reduce the compression distortion of the low-frequency spatial frequency component as the variance of the transform coefficient is smaller.

Also preferably, in the present invention, the statistical property is a variance obtained for a transform coefficient obtained by performing a wavelet transform on the image data.

More preferably, the parameter setting unit of the present invention uniformly sets the image compression parameters for all of the plurality of tiles.

Also preferably, the statistical processing unit of the present invention obtains a statistical property for each tile. The parameter setting unit predicts the conspicuousness of the tile boundaries in tile units based on the statistical properties obtained in tile units, and reduces the compression distortion of low-frequency spatial frequency components as the tiles are more conspicuous. Adjust and set the image compression parameters.

Further, preferably, the parameter setting unit also adjusts and sets the image compression parameter for the tile adjacent to the tile in which the tile boundary is conspicuous so as to reduce the compression distortion of the low-frequency spatial frequency component. On the other hand, an image compression program according to the present invention causes a computer to function as the above-described statistical processing unit, parameter setting unit, and image compression unit.

Further, the image compression method of the present invention is an image compression method in which image data is divided into a plurality of tiles, and image compression is performed for each tile. A statistical processing step for obtaining a statistical property that correlates with "conspicuousness in the event of a shift"; and predicting the conspicuousness of the tile boundary at the time of image expansion based on the statistical property. Parameter setting step to adjust and set the "image compression parameter that affects the amount of compression distortion" according to the conspicuousness of the image, and image compression to compress the image data using the adjusted image compression parameter And steps. BRIEF DESCRIPTION OF THE FIGURES

The above objects and other objects of the present invention can be easily confirmed by the following description and the accompanying drawings.

FIG. 1 is a diagram illustrating a configuration of an electronic camera 11 according to the present embodiment.

FIG. 2 is a flowchart illustrating an image compression process according to the first embodiment.

FIG. 3 is a diagram showing a state of subband division by wavelet transform. FIG. 4 is a flowchart illustrating an image compression process according to the second embodiment.

FIG. 5 is a flowchart illustrating an image compression process according to the third embodiment.

Fig. 6 is a diagram showing the standard deviation (square root of variance) of the transform coefficients for each subband, picking up the characteristic tiles of the test image.

FIG. 7 is a diagram illustrating an example of the weighting factor Ws for each subband used for calculating the evaluation parameter k.

FIG. 8 is a diagram showing characteristic parameters k of the test image picked up.

FIG. 9 is a map in which the evaluation parameters k are arranged according to the tile division. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings. << 1st Embodiment >>

[Description of electronic camera configuration]

In FIG. 1, a photographing lens 12 is attached to an electronic camera 11. The light receiving surface of the image sensor 13 is arranged in the image space of the photographing lens 12. The output of the image sensor 13 is processed through the AD conversion unit 14 and the signal processing unit 15 and then recorded in the buffer memory 16.

An image processing unit 17 and an image compression unit 18 are connected to the data bus of the buffer memory 16 respectively.

The image compression section 18 has the following configuration requirements 1 to 5.

①Tile processing unit 2 1

②Wavelet converter 2 2

③ Quantization unit 2 3

④Bit modeling part 2 4

⑤ Arithmetic encoder 2 5

⑥ Bit stream generator 2 6

⑦ ROI setting section 2 7

The compressed file generated in the image compression section 18 is provided to the recording section 32. The recording unit 32 records and stores the compressed file on the memory card 33.

Further, an operation member 38 including a release button, a setting button, and the like is provided on a housing of the electronic camera 11. Output signals of these operation members 38 are supplied to a microprocessor 39. The microprocessor 39 is a microprocessor for controlling the system of the electronic camera 11 and is connected to the image compression section 18 and the image processing section 17 via signal lines.

[Correspondence with invention]

Here, the correspondence between the items described in the claims and the above-described electronic camera 11 will be described. It should be noted that the correspondence relationship here is merely an example for reference, and does not limit the present invention. The statistical processing unit described in the claims corresponds to the image compression unit 18 and the microprocessor 39.

The parameter setting unit described in the claims corresponds to the image compression unit 18 and the microprocessor 39.

The image compression unit described in the claims corresponds to the image compression unit 18.

[Explanation of Operation of First Embodiment]

Hereinafter, the operation of the first embodiment will be described.

First, the analog image data generated by the image sensor 13 is digitized via the AD converter 14 and then supplied to the signal processor 15. The signal processing unit 15 performs signal processing such as black level correction and gamma correction on the image data. The image data after the signal processing is temporarily recorded in the buffer memory 16. The image processing unit 17 reads out the image data in the buffer memory 16 in units of processing and performs image processing such as color interpolation processing and color coordinate conversion.

The image compression section 18 reads out the image data after the image processing from the buffer memory 16 and performs the image compression processing. FIG. 2 is a flowchart illustrating this image compression processing.

Hereinafter, the image compression processing which is a feature of the present invention will be described along the step numbers shown in FIG.

Step S 1: The tile processing section 21 in the image compression section 18 divides the image data in the buffer memory 16 into tile units and sequentially reads the data.

Step S2: The wavelet transform unit 22 performs wavelet transform on each tile and divides the tile into a plurality of subbands as shown in FIG.

Step S3: The wavelet transformer 22 finds the standard deviation (square root of variance) of the wavelet transform coefficient for each subband for each tile. In general, the smaller these standard deviations are overall, the higher the flatness of the gradation (including the color tone) of the image data. Therefore, even if a slight gradation shift occurs in the image data, it can be determined that the gradation shift is conspicuous. The wavelet transform unit 22 transmits the standard deviation thus obtained to the microprocessor 39.

Step S4: The microprocessor 39 selects an image according to the standard deviation of the conversion coefficient. Estimate the conspicuousness of discontinuous compression distortion generated at tile boundaries during image expansion. The microprocessor 39 adjusts and sets the visual weight (visual importance) of each subband according to the prediction situation.

Hereinafter, a specific example of this adjustment setting will be described.

First, the microprocessor 39 obtains the ratio of the gradation flat tile whose standard deviation is equal to or smaller than the threshold value (here, equal to or smaller than 3) in the 3 LL band (lowest band) of each tile, and determines the ratio as r 1. I do. As the value of 1 is larger, the gradation of the image data is globally flat, and it can be determined that the gradation deviation is more conspicuous.

Further, the microprocessor 39 calculates the ratio of the flat tone tiles whose standard deviation is equal to or less than the threshold value (here, 3 or less) in the 3 LH, 3HL, and 3 HH bands of each tile, and calculates the ratio as r 2 And This is the ratio of tiles that can be regarded as gradation flat in the intermediate spatial frequency range. In addition to rl, it can be determined that the larger the value of r2, the flatter the gradation of the image data and the more noticeable the gradation deviation. Further, the microprocessor 39 obtains the ratio of the gradation flat tile whose standard deviation is equal to or smaller than the threshold value (here, equal to or smaller than 3) among the 2 LH, 2 HL, and 2HH bands of each tile. I do. This r 3 is also the ratio of tiles that can be regarded as having a flat gradation in the intermediate spatial frequency range. In addition to r 1 and r 2, it can be determined that the larger the value of 3 is, the flatter the gradation of the image data is, and the more the gradation deviation is more conspicuous.

Further, the microprocessor 39 calculates the ratio of the gradation flat tile whose standard deviation is equal to or less than the threshold value (here, 3 or less) in the 1 LH, 1 HL, and 1 HH bands of each tile, and calculates the ratio as r And 4. This r4 is the ratio of tiles that can be regarded as gradation flat in the high spatial frequency range. In addition to r l to r 3, it can be determined that as the value of r 4 is larger, the gradation of the image data is flatter like silk, and the gradation deviation is more conspicuous.

Using the ratio r 1 to r 4 of the gradation flat tile obtained in this way, the microprocessor 39

Parameter k = 8 (2Xr1 + 0.5Xr2 + 0.25Xr3 + 0.25Xr4)

Is calculated, and the parameter k is obtained. This parameter k is the same as r1 to r4 described above. These are integrated evaluation parameters. Therefore, it can be determined that the larger the evaluation parameter k is, the flatter the gradation of the image data is from the low frequency band to the high frequency band, and that the gradation deviation is conspicuous.

As such image data having a large evaluation parameter k, for example, a picture such as a blue sky in fine weather is assumed. Such image data is expected to be particularly conspicuous because the tile boundaries are raised because of the monotonous pattern.

On the other hand, image data with a small evaluation parameter k is assumed to be a complex picture rich in detail. In such image data, discontinuous compression distortions generated at tile boundaries are buried in complex patterns and are expected to be inconspicuous. The microprocessor 39 determines the visual weight of each subband by using the following equation in consideration of the prediction situation based on the evaluation parameter k.

y 3LL = 1.0

γ 3LH = y 3HL = exp (-l / 18 X k)

7 3HH = exp (-2/18 Xk)

y 2LH = y 2HL = exp (-4/18 X k)

y 2HH = exp (-8/18 Xk)

y 1LH = y lHL = exp (-9/18 X k)

y lHH = exp (-18/18 Xk)

In the above equation, as the evaluation parameter k increases, the visual weight γ 3Ι of the low band sub-band is set relatively large.

This is a measure based on the findings that the inventor clarified based on a number of image compression experiments that "the more low-frequency compression distortion occurs, the more noticeable the tile boundary becomes."

The microprocessor 39 transmits the selected visual weight to the quantization unit 23.

Step S5: The quantization unit 23 optimizes the quantization step width of the transform coefficient so as to minimize visual image quality deterioration due to quantization distortion under an appropriate coding rate constraint condition.

Hereinafter, the method of optimizing the quantization step width will be described in principle. Normally, the transform coefficients for each subband show a probability distribution very similar to the Laplace distribution. Therefore, the probability distribution of the conversion coefficient is approximated by the following Laplace distribution equation, and the variable α that specifies the Laplace distribution is obtained. f (y) = ^ -e- ^ · '· [1] When such a transform coefficient is quantized with the quantization step width Q, the quantized transform coefficient takes the k-th quantized value. The probability pk is

Becomes At this time, the coding rate R (Q) after quantization is ideally equal to the entropy, and

R (Q) = -∑P _k logP _k

• [3]

Becomes

On the other hand, the quantization distortion D (Q) is calculated using the square error

D (Q) -∑L_ ₂₎ (y-kQ † f (y) dy

1/2)

It is expressed as

Here, the quantization step width of the i-th subband is Q i, the coding rate at that time is R i (Q i), and the quantization distortion is D i (R i).

Here, the visual quality degradation due to quantization distortion is minimized under the constraint that the sum of the subband coding rates be the target rate. Therefore, the following function J is introduced using Lagrange's undetermined multiplier;

J = ∑rA (Ri + ^ ∑ Qi) '· [5] In the expression of this function J, γ i corresponds to the visual weight of the transform coefficient belonging to subband i, and is a group of values γ 3Ι to γ 1HH already obtained in step S4 described above. Next, this function J is partially differentiated by R i to obtain a conditional expression in which the first term of the function J is a stationary value. By transforming this equation,

… _M

Is obtained.

By solving this equation [7], giving the value of L, Q i for each sub-band is obtained. Based on these Q i,

To calculate the total coding rate R λ.

By fine-tuning the undetermined multiplier; L so that the coding rate R matches the target coding rate, the quantization step width Q i is determined.

(Note that the operation results of the above-described optimization may be summarized in advance as a data table. In this case, the quantization unit 23 refers to this data table based on the variable α and the visual weight, The optimal quantization step width can be determined immediately.)

Step S 6: The quantization unit 23 quantizes the transform coefficient of each subband according to the quantization step width Q i thus optimized.

By the way, in the above-described optimization of the quantization step width, the quantization step width of the lower sub-band is set to be narrower as the evaluation parameter k is larger. As a result, the lower the compression distortion (quantization distortion in this case), the lower the image data with a large evaluation parameter k.

Step S7: The bit modeling section 24 reads a mask image of the selected area preset in the R〇I setting section 27. The bit modeling section 24 It is determined whether or not the transform coefficient is included in the selected area for each subband based on the mask image of. The bit modeling unit 24 shifts the quantized transform coefficient located in the selected area by S bits. The S bits are the number of bits determined according to the known Max Shift method.

Step S8: The bit modeling unit 24 divides the quantized transform coefficient into bit planes. The bit modeling unit 24 executes three kinds of encoding passes specified in JPEG 2000 in order from the bit plane of the most significant digit.

Step S 9: The arithmetic encoding unit 25 acquires encoded data from the bit modeling unit 24 for each processing unit. Arithmetic coding section 25 performs arithmetic coding on the coded data by using an MQ coder 1 which is a binary arithmetic code.

Step S10: The bitstream generation unit 26 rearranges the encoded data according to the SNR progressive and other priorities, and generates a bitstream. Step S11: The bitstream generation unit 26 cuts off this bitstream with a predetermined code amount and outputs it as a compressed file. The recording unit 32 records and stores the compressed file on the memory card 33.

With the series of operations described above, the imaging operation of the electronic camera 11 and the image compression process are completed.

[Effects of the first embodiment, etc.]

As described above, the electronic camera 11 predicts the conspicuousness of the tile boundary using the standard deviation of the wavelet transform coefficient. At this time, the quantization step width is adjusted and set so that the image distortion of the tile boundary is less conspicuous and the compression distortion in the low frequency band is suppressed lower.

As a result, it is possible to appropriately select image data (for example, a picture such as a blue sky) in which the tile boundaries are visually conspicuous, and to suppress in advance the problematic low-frequency compression distortion.

Conversely, it is possible to appropriately select image data (for example, fine and complicated patterns) in which tile boundaries are not visually noticeable, and to allow moderately low-frequency compression distortion. Next, another embodiment will be described.

<< 2nd Embodiment >> Since the configuration of the second embodiment is the same as that of the first embodiment (FIG. 1), the description is omitted here.

FIG. 4 is a diagram illustrating an image compression process according to the second embodiment. Hereinafter, the operation of the second embodiment will be described along the step numbers shown in FIG.

Step S21: The microprocessor 39 extracts the image data to be compressed by dividing it into a plurality of divided areas (for example, tiles), and standard deviation of pixel values (values of luminance, color, color difference, etc.) for each divided area. Ask for.

Step S22: The ratio of the "segmented region where the standard deviation is equal to or less than a predetermined threshold" to the entire extracted divided region is determined, and the ratio is defined as the flatness R of the image data. It can be determined that the larger the flatness R is, the higher the flatness of the gradation (including the color tone) of the image data is and the more noticeable the gradation deviation is.

Step S23: It is predicted that the larger the flatness R is, the more noticeable the discontinuous compression distortion that occurs at the tile boundary during image expansion. Therefore, the microprocessor 39 relatively increases the visual weight of the low-frequency subband as the value of the flatness R increases. The microprocessor 39 transmits the visual weight adjusted and set as described above to the quantization unit 23.

Step S24: The tile processing section 21 divides the image data in the buffer memory 16 into tile units and sequentially reads them.

Step S25: The wavelet transform unit 22 independently performs wavelet transform on each tile and divides it into subbands as shown in FIG.

Note that the subsequent operations (steps S26 to S32) are the same as steps S5 to S11 described in the first embodiment, and a description thereof will be omitted.

[Effects of the second embodiment, etc.]

By the operation described above, the same effects as those of the first embodiment can be obtained in the second embodiment.

In particular, in the second embodiment, the conspicuousness of the tile boundary is predicted based on the standard deviation of the pixel values. As a result, unlike the first embodiment, the prediction operation can be performed without waiting for the wavelet change. As a result, the processing of predicting the conspicuousness of the tile boundaries and the wavelet transform can be processed in parallel. Suitable for improving efficiency.

Hereinafter, another embodiment will be described.

<< Third embodiment >>

Since the configuration of the third embodiment is the same as that of the first embodiment (FIG. 1), the description is omitted here.

FIG. 5 is a diagram illustrating an image compression process according to the third embodiment. Hereinafter, the operation of the third embodiment will be described along the step numbers shown in FIG.

Steps S41 to S42: Same as steps S1 to S2 described in the first embodiment.

Step S43: The wavelet transform unit 22 finds the variance of the wavelet transform coefficient for each subband for each tile. The following equation is used to calculate the variance (standard deviation squared) of the wavelet transform coefficient. And = ^ ∑ [^)-] ² · '· [9]

, Nu

. _s fucking one

Two

a _s nohan r variance of S

C _s (x, y): Conversion coefficient of position (JC,) in tile of subband S

Cj: Average value of conversion coefficient of subband S

N: Number of transform coefficients of subband S

Here, the variance is calculated for the entire area inside the tile. However, the variance may be calculated only for the inner circumference of the tile related to the tile boundary (that is, the central area of the tile not related to the tile boundary may be excluded from the variance calculation). In addition, in order to distinguish and predict the prominence of the tile boundary for each side of the tile, the variance may be calculated by dividing the tile into the vicinity of the top side, near the bottom side, near the left side, and near the right side. . FIG. 6 is a diagram in which characteristic image regions of a test image (person image) are picked up and the standard deviation of the エー -Apelet transform coefficient is shown for each sub-band.

In the case of this test image, the tile corresponding to the “background” has a uniformly low standard deviation from the high-frequency subband to the low-frequency subband because the entire image is blurred, and the tile boundary is It can be judged that it is very noticeable.

On the other hand, the tiles of “Knitted Clothes”, “Eyes”, and “Mouths” It can be determined that the standard deviation is relatively high over the area subband, and the tile boundaries are relatively inconspicuous.

The “hair” tile has a high standard deviation value from the low-frequency subband to the high-frequency subband. Therefore, it can be determined that the tile boundaries are hardly noticeable because they are buried in a complicated pattern.

The wavelet transform unit 22 transmits the standard deviation thus obtained to the microprocessor 39.

Step S44: The microprocessor 39 obtains an evaluation parameter k which indicates the conspicuousness of the tile boundary by a numerical value in order to perform the evaluation judgment as shown in FIG. 6 by calculation. The evaluation parameter k here is a value calculated for each tile, using the following equation, different from the first embodiment. …]

The weight coefficient W s in the expression [10] is a value indicating the ratio of each subband S contributing to the conspicuousness of the tile boundary, and may be determined based on a subjective evaluation experiment of an image or the like. FIG. 7 is a table showing an example of the weight coefficient Ws determined in this way. A in [10] is a proportionality constant. '

When the variance is calculated by dividing the tile into the upper side, the lower side, the right side, and the left side in step S43, it is preferable to obtain the evaluation parameter k for each of these sections.

Incidentally, here, the evaluation parameter k is determined using the [10] expression, but the present invention is not limited to this evaluation expression. An evaluation formula that calculates an evaluation parameter that makes the tile boundary less noticeable as the value of the variance is larger may be used.

Fig. 8 is a table showing the values of the evaluation parameter k by picking up characteristic image areas of the test image (person image). It can be predicted that the larger the value of the evaluation parameter k is, the more noticeable the tile boundary is.

Step S45: Subsequently, the microprocessor 39 adjusts the value of the value parameter k between adjacent tiles. Here, a local area including the upper, lower, left, and right neighboring tiles is first set for each tile to be processed by the microprocessor. Next In addition, the microprocessor 39 smoothes the evaluation parameter k (such as a simple average and a weighted average) in the local region.

Figure 9 [A] is a map in which the evaluation parameters k of the human image are arranged according to the tile division. Fig. 9 [B] is a map showing the evaluation parameter k after the smoothing process.

Note that instead of the smoothing operation here, a calculation operation such as a MAX value or a median value may be used.

When the standard deviation is calculated by dividing the tile into the upper side, the lower side, the right side, and the left side in step S43, the evaluation parameter k may be adjusted between actually adjacent sides. . (For example, the adjustment of the evaluation parameter k may be performed between the left side of the tile located on the right and the right side of the tile located on the left.)

By adjusting the evaluation parameter k as described above, if there is a tile or a side where the tile boundary is conspicuous, it is determined that the tile boundary is also conspicuous for the tile adjacent to the tile or the side.

Step S46: The microprocessor 39 determines the visual weight of each sub-band using the following equation, taking into account the prediction situation based on such evaluation parameters. y3LL = 1.0

y 3LH = y3HL = exp (-l / 18Xk)

y 3HH = exp (-2 / 18Xk)

y 2LH = γ 2HL = exp (-4/18 X k)

y2HH = exp (-8 / 18Xk)

γ 1LH = y lHL = exp (-9/18 X k)

γ lHH = exp (-18 / 18Xk)

In the above equation, as the evaluation parameter k increases, the visual weight T 3LL of the low-frequency subband is set to be relatively large.

The microprocessor 39 transmits the visual weight thus selected to the quantization unit 23.

Step S47: The quantization unit 23 performs a quantization step of the transform coefficient according to the visual weight set for each tile so as to minimize the quantization distortion and the visual quality degradation. Is optimized for each tile. The procedure of this optimization method is the same as that in step S5 of the first embodiment, and thus the duplicate description is omitted here.

Step S48: The quantizing section 23 quantizes the transform coefficient for each tile using the quantization step width of each subband optimized for each tile.

By the way, in the above-described optimization of the quantization step width, the quantization step width of the lower sub-band is set to be narrower as the evaluation parameter k is larger. As a result, tiles with a larger evaluation parameter k have lower compression distortion (quantization distortion in this case) in the low frequency band.

Steps S49 to S53: Same as steps S7 to S11 described in the first embodiment.

Step S54: The microprocessor 39 adds data relating to the quantization step width determined for each tile to the compressed data to generate a compressed file in preparation for using data at the time of image expansion. The recording unit 32 records and stores the generated compressed file on the memory card 33.

With the series of operations described above, the imaging operation of the electronic force camera 11 and the image compression processing are completed.

[Effects of Third Embodiment]

With the operation described above, the same effects as in the first embodiment can be obtained in the third embodiment.

In particular, in the third embodiment, tiles in which tile boundaries are conspicuous (for example, tiles such as blue sky) are appropriately selected and low-frequency compression distortion (that is, occurrence of tile boundaries) which is a problem is suppressed in tile units. Will be possible. Conversely, it is possible to appropriately select tiles whose tile boundaries are not visually conspicuous (for example, tiles with complicated patterns) and to allow moderately low-frequency compression distortion. As a result, for tiles with complicated patterns, the image information of the mid- and high-frequency components is compressed as accurately as possible to the extent that tile boundaries are allowed to occur, and image information deterioration due to compression is prevented as much as possible. It becomes possible.

As described above, in the third embodiment, the frequency distribution of the compression code amount is adjusted for each tile based on the conspicuousness of the tile boundary. As a result, tile boundaries are a real problem At extremely high compression ratios, it is possible to achieve the desired adjustment by minimizing image information degradation while improving tile boundaries.

Furthermore, in the third embodiment, low-frequency compression distortion is suppressed also around tiles with noticeable tile boundaries. As a result, it is possible to improve a situation in which low-frequency distortion occurs around a flat tile, and as a result, a tile boundary becomes conspicuous.

[Supplementary information of the embodiment]

In the above-described embodiment, the conversion coefficient and the standard deviation (square root of the variance) of the pixel value are obtained to evaluate the conspicuousness of the gradation shift. However, the present invention is not limited to this. In general, any statistical property of the present invention may be used as long as the quantity correlates with “conspicuousness of gradation deviation”. For example, statistical properties indicating gradation flatness, gradation uniformity, gradation monotony, and the like may be used.

In addition, the “variance” of the present invention may be any measure that indicates the degree of dispersion of values, and is not limited to the statistical calculation formula expressed by the formula [9].

By the way, the dynamic range of the pixel value is a value that fluctuates greatly even if there is a large gradation change even momentarily. Therefore, the dynamic range of the pixel value is not a value indicating the flatness of the gradation, and is not an amount correlating with “conspicuousness of gradation deviation”.

In the above-described embodiment, the quantization step size of the transform coefficient is adjusted and set as the image compression parameter. However, the present invention is not limited to this. In general, any image compression parameter of the present invention may be used as long as it affects the amount of appearance of compression distortion at a tile boundary. For example, the image compression parameter may be a parameter for adjusting the lower digit truncation of the bit plane, a parameter for adjusting the bit stream truncation, or a parameter for adjusting a selected region of ROI coding.

In the first and second embodiments, the image compression parameters are uniformly changed for all the plurality of tiles. In this case, since the image compression is uniformly applied to the entire plurality of tiles, there is little possibility that the compression distortion will appear on the screen unevenly. However, the present invention is not limited to this. For example, the third embodiment As shown in the figure, the image compression parameters may be adjusted and set for each individual tile. Further, the image compression parameters may be individually adjusted and set at the tile boundary and inside the tile.

Further, in the above-described embodiment, the case where the image compression of the present invention is performed in the electronic camera 11 has been described. However, the present invention is not limited to this. For example, the image compression device of the present invention may be realized on a computer. In this case, an image compression program (corresponding to claim 9) may be created by converting the image compression procedure in the above-described embodiment into a program code. Note that the present invention may be embodied in various other forms without departing from the spirit or main characteristics thereof. Therefore, the above-described embodiment is merely an example in every aspect, and should not be construed as limiting. The scope of the present invention is defined by the scope of the claims, and is not limited by the text of the specification. Further, all modifications and changes belonging to the equivalent scope of the claims are within the scope of the present invention. Industrial applicability

In the image compression of the present invention, a statistical property correlating with “visual conspicuousness of gradation shift” is obtained from image data, and in accordance with the conspicuous prominence of a tile boundary predicted by the statistical property, Adjust and set image compression parameters that affect the appearance of compression distortion.

Therefore, for image data in which tile boundaries are likely to be conspicuous at the time of image expansion, appropriate measures such as suppressing compression distortion in advance are facilitated.

Claims

The scope of the claims

(1) An image compression device that divides image data into a plurality of tiles and performs image compression for each of the tiles,

A statistical processing unit that processes the image data and obtains a statistical property that is correlated with “conspicuousness of the tone shift when tone shift occurs in the image data”; A parameter setting unit for predicting the conspicuousness of the tile boundaries at the time of image expansion and adjusting and setting `` image compression parameters that affect the appearance of compression distortion '' according to the conspicuousness of the tile boundaries;

An image compression unit configured to perform image compression on the image data using the adjusted image compression parameter;

An image compression device comprising:

(2) In the image compression device according to claim 1,

The statistical property is a value indicating the flatness of the gradation of the image data.The higher the degree of flatness, the smaller the compression distortion of the low-frequency spatial frequency component. Adjust and set the image compression parameters

An image compression apparatus characterized by the above-mentioned.

(3) In the image compression device according to claim 1,

The statistical property is a ratio of a gradation flat tile (tile that can be regarded as a flat gradation) to the plurality of tiles,

The image compression device, wherein the parameter setting unit adjusts and sets the image compression parameter in a direction to reduce a compression distortion of a low-frequency spatial frequency component as the ratio of the gradation flat tile increases.

(4) In the image compression device according to claim 1,

The statistical property is a variance obtained for a transform coefficient obtained by converting the image data into a spatial frequency domain.

The image compression device, wherein the parameter setting unit adjusts and sets the image compression parameter in a direction to reduce compression distortion of a low-frequency spatial frequency component as the variance of the transform coefficient is smaller.

(5) In the image compression device according to claim 4, The statistical property is a variance obtained for a transform coefficient obtained by performing a wavelet transform on the image data.

An image compression apparatus characterized by the above-mentioned.

(6) The image compression device according to any one of Claims 1 to 5, wherein the parameter setting unit uniformly sets the image compression parameters for the entire plurality of tiles.

An image compression apparatus characterized by the above-mentioned.

(7) The image compression device according to any one of claims 1, 2, 4, and 5, wherein the statistical processing unit obtains the statistical property for each tile.

The parameter setting unit predicts the conspicuousness of the tile boundary in the tile unit based on the statistical property of the tile unit, and the compression distortion of the low-frequency spatial frequency component increases as the tile is more conspicuous. Adjust and set the image compression parameter in the direction to reduce

An image compression apparatus characterized by the above-mentioned.

(8) In the image compression device according to claim 7,

The parameter setting unit adjusts and sets the image compression parameter in a direction to reduce compression distortion of a low-frequency spatial frequency component also for a tile adjacent to a tile in which the tile boundary is conspicuous.

An image compression apparatus characterized in that:

(9) An image compression program for causing a computer to function as the statistical processing unit, the parameter setting unit, and the image compression unit according to any one of claims 1 to 8.

(10) An image compression method for dividing image data into a plurality of tiles and performing image compression on each of the tiles,

A statistical processing step of processing the image data to determine a statistical property correlating with “inconspicuousness when a gradation shift occurs in the image data”;

Predict the conspicuousness of the tile boundary at the time of image expansion based on the statistical properties, and adjust the `` image compression parameter that controls the appearance of compression distortion according to the predicted conspicuousness of the tile boundary '' Parameter setting step to be set. An image compression step of compressing the image data using the adjusted image compression parameter;

An image compression method comprising: