JP2009010612A - Image compressing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the size of compressed data by eliminating data of an undesired high spatial frequency component in image data by attenuating an AC component in a part (dark part) at a low luminance level of an image. <P>SOLUTION: This image compressing device for compressing image data Id in each block in a fixed size is provided with a buffer 101 for outputting the image data Id in each block, a DCT device 102 for converting image data Bd corresponding to a block outputted from the buffer 101 into data Td of a frequency domain by DCT processing, and a quantizing part 103 for quantizing the data Td of the frequency domain corresponding to each block converted by the DCT device 102, wherein the quantizing part 103 changes components of a quantization table to be used to quantize the data Bd of the frequency domain of the each block in each block so as to increase a compression rate of a part darker than reference brightness of a corresponding image. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image compression apparatus and an image compression method, and more particularly to a high-efficiency image encoding process that efficiently compresses an image by increasing a compression ratio of a dark portion of an image.

  An example of image compression processing compliant with the JPEG (Joint Photographic Expert Group) standard is disclosed in Patent Document 1.

  In this image compression process, an image is divided into a region of a certain size, here a block of 8 × 8 pixels, and the image data is divided into block units using a discrete cosine transform (DCT: Discrete Cosine Transform), Transform from time domain to frequency domain.

  Data obtained by the conversion to the frequency domain (hereinafter referred to as DCT data) is compressed by Huffman coding after the amount of information is reduced by quantization.

  In this conversion to the frequency domain, the amount of information is not reduced by the conversion itself, but by using the fact that energy is collected in low frequency components, the amount of information is reduced by quantization of DCT data and entropy coding such as Huffman coding is performed. The compression rate of the image data is improved by the data compression according to.

  FIG. 9 is a block diagram for explaining a conventional image compression apparatus that performs JPEG-compliant image compression processing as described above.

  The image compression apparatus 200 buffers input image data Id, divides the image data so as to correspond to the block, and outputs image data (hereinafter also referred to as block data) Bd corresponding to the block. A buffer 201 and a DCT unit 202 that converts block data Bd from the buffer 201 into frequency domain data (DCT data) Td by DCT. In addition, the image compression apparatus 200 quantizes the DCT data Td from the DCT unit 202 using a fixed quantization table Qt to generate the quantized data Qd, and the quantized data Qd And a Huffman encoder 204 that performs Huffman encoding processing using the Huffman code table Ht to generate compressed data Cd. The image compression apparatus 200 includes a quantization table storage unit 203a that stores the quantization table Qt, and a Huffman table storage unit 204a that stores the Huffman table.

  Next, the operation will be briefly described.

  When image data Id corresponding to one screen is input to the image compression apparatus 200, the image data Id is divided by the buffer 201 so as to correspond to, for example, an 8 × 8 pixel size block. Then, image data is output for each block. When the image data (block data) Bd corresponding to each block is input to the DCT unit 202, the block data Bd is converted into a frequency composed of 8 × 8 frequency components by discrete cosine transform. The data is converted into area data (DCT data) Td and input to the quantizer 203. The quantizer 203 divides each frequency component in the DCT data Td by the corresponding quantization component of the quantization table, whereby quantized data Qd consisting of 8 × 8 frequency components quantized is obtained. Is generated and output to the Huffman encoder 204. The Huffman encoder 204 performs a Huffman encoding process on the quantized data Qd using the Huffman table Ht, and the compressed data Cd obtained thereby is output from the image compression apparatus 200.

  By the way, if normal image data is quantized as it is, a great deterioration in image quality occurs. However, by converting the frequency of the image data as described above, the information that is conspicuous for humans is collected in the low frequency component. By quantizing the frequency-converted image data, it is possible to compress the image data while retaining the properties of the original image.

  Specifically, there are two quantization tables used for compressing such image data, that is, a luminance table and a color table. Data obtained by DCT processing of image data is represented by the quantization table. Information compression (information truncation) is performed by dividing by the table value.

  Basically, the low frequency component is not compressed so much, and the high frequency component is compressed with a high compression rate.

  Therefore, in this quantization table, the quantization level (quantization component) for the low frequency part of the DCT data is a small value, and the quantization level (quantization component) for the high frequency component of the DCT data is a large value. Here, increasing the compression rate means increasing the set values of the quantization components in these tables as a whole.

Since the quantization table values used in such image data encoding processing are necessary when reproducing an image, that is, when decoding encoded data, a JPEG file, that is, a JPEG-compliant code, is used. It is written in the header portion of the encoded data and sent to the decoding side together with the encoded data.
Japanese Patent Laid-Open No. 7-123413

  However, the conventional compression method has the following problems.

  Since compression is performed on a single image using a fixed quantization table, a dark portion of one image, that is, a portion with a small amount of information originally, and a portion with a bright amount of information as an image. It was treated with the same compression rate.

  In this case, there is a possibility that the data amount of one image can be made smaller by the compression process. However, in the method using the same quantization table for one image as described above, the image quality deterioration due to the compression is caused. Cannot be further compressed.

  The present invention has been made to solve such a problem, and it is possible to perform further compression so that unnecessary high spatial frequency component data is removed, thereby reducing the size of the compressed image data. An object is to obtain an image compression apparatus and an image compression method that can be further reduced.

  An image compression apparatus according to the present invention is an image compression apparatus that generates compressed data by compressing an image, an image dividing unit that divides the image into blocks that are units of the compression, and image data of the image A quantization unit that performs quantization for each block, and the quantization unit is a component of a quantization table used for quantization of the image data so that a compression ratio of a darker portion of the image is higher than a reference brightness Is determined for each block, and the above object is achieved.

  The present invention includes an encoder that encodes quantized data obtained by quantizing the image data and generates compressed data in the image compression device, and a quantization table corresponding to each block, It is preferable to transmit the compressed data of each block to the decoding side.

  In the image compression apparatus according to the present invention, it is preferable that a quantization table corresponding to each block is included in a header portion corresponding to each block of the compressed data and transmitted.

  According to the present invention, in the image compression apparatus, the quantization unit holds a reference quantization table, and corrects a component of the reference quantization table according to a luminance level of each block, thereby correcting the quantization table. Is preferably created for each block, and the image data of each block is quantized using the created correction quantization table.

  In the image compression apparatus, it is preferable that the correction quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.

  In the image compression apparatus, it is preferable that the reference quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.

  The present invention provides the above image compression apparatus, further comprising: a converter that performs discrete cosine transform on the image data of the image for each block to generate converted data, and the quantization unit quantizes the converted data as the image data. Preferably, the correction quantization table is created by changing the AC component corresponding to the AC component of the conversion data among the components of the reference quantization table according to the luminance level of each block. .

  According to the present invention, in the image compression apparatus, the quantization unit converts a high-frequency component corresponding to a component having a high spatial frequency of the converted data, among the AC components of the reference quantization table, to the luminance level of each block. It is preferable that the correction quantization table is created by changing according to the above.

  According to the present invention, in the image compression device, the quantization unit holds a luminance level extraction unit that extracts a luminance level of each block based on the image data, and the reference quantization table. A quantization table creating unit that creates a corrected quantization table for each block by correcting high-frequency components of the reference quantization table according to the extracted luminance level, and the created correction data for the converted data of each block It is preferable to have a quantizer that quantizes using a quantization table.

  According to the present invention, in the image compression device, the quantization table creation unit creates the correction quantization table by multiplying a high frequency component of the reference quantization table by a coefficient according to a luminance level of each block. The luminance level extraction unit compares the detection level supplied from the outside with the luminance level corresponding to each block, and when the luminance level of the block is equal to or higher than the detection level, When a correction control signal indicating that the reference quantization table should be used as a quantization table for the block is output, the coefficient is set to 1, and the luminance level of the block is smaller than the detection level Outputs a correction control signal indicating that the correction quantization table should be used as the quantization table of the block, and sets the coefficient to the block It is preferred that the degree level is set to lower the larger value.

  An image compression method according to the present invention is an image compression method for compressing an image to generate compressed data, an image dividing step for dividing the image into blocks serving as compression units, and image data of the image. A quantization step for performing quantization for each block, and in the quantization step, components of a quantization table used for quantization of the image data so that a compression ratio of a darker portion of the image is higher than a reference brightness Is determined for each block, thereby achieving the above object.

  The present invention includes an encoding step of generating compressed data by encoding quantized data obtained by quantizing the image data in the image compressing method, and including a quantization table corresponding to each block, It is preferable to transmit to the decoding side together with the compressed data of each block.

  In the image compression method according to the present invention, it is preferable that a quantization table corresponding to each block is included in a header portion corresponding to each block of the compressed data and transmitted.

  According to the present invention, in the image compression method, in the quantization step, a correction quantization table is created for each block by correcting the components of the reference quantization table according to the luminance level of each block. It is preferable to quantize the image data using the created correction quantization table.

  In the image compression method according to the present invention, it is preferable that the correction quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.

  In the image compression method according to the present invention, it is preferable that the reference quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.

  The image compression method includes a conversion step of performing discrete cosine conversion on the image data of the image for each block to generate conversion data. In the quantization step, the conversion data is quantized as the image data. Preferably, the correction quantization table is created by changing the AC component corresponding to the AC component of the conversion data among the components of the reference quantization table according to the luminance level of each block. .

  According to the present invention, in the image compression method, in the quantization step, among the alternating current components of the reference quantization table, a high frequency component corresponding to a component having a high spatial frequency of the converted data is converted into a luminance level of each block. It is preferable that the correction quantization table is created by changing according to the above.

  According to the present invention, in the image compression method, the quantization step extracts a luminance level of each block based on the image data, and stores the reference quantization table according to the extracted luminance level of each block. Preferably, the correction quantization table is created for each block by correcting the high-frequency component, and the converted data of each block is quantized using the created correction quantization table.

  According to the present invention, in the image compression method, in the quantization step, the correction quantization table is created by multiplying a high frequency component of the reference quantization table by a coefficient according to a luminance level of each block, The determined detection level is compared with the luminance level corresponding to each block. When the luminance level of the block is equal to or higher than the detection level, the reference quantization table is used as the quantization table for the block. A correction control signal indicating that the block should be generated, and when the coefficient is set to 1 and the luminance level of the block is smaller than the detection level, the correction quantization table as the quantization table of the block It is preferable to generate a correction control signal indicating that the block should be used, and to set the coefficient to a larger value as the luminance level of the block is lower. There.

  The operation of the present invention will be described below with the above configuration.

  In the present invention, the image compression apparatus includes a quantization unit that quantizes the image data for each block that divides the image, and the quantization unit compresses a portion of the image that is darker than the reference brightness. Since the components of the quantization table used for quantization of the image data are changed for each block so that the rate is increased, the amount of information in the portion of the compressed data with a low luminance level (dark portion) is reduced. Thus, it is possible to further reduce the size of the compressed data while suppressing image quality deterioration.

  In the present invention, among the alternating current components of the quantization table used for quantization of each block, the high frequency corresponding to the high spatial frequency component of the converted data obtained by converting the image data from the time domain to the frequency domain. Since the component is changed in accordance with the luminance level of each block, the AC component in the dark portion of the image data is further reduced compared to the AC component in the bright portion. Thereby, the effect that the noise component of the dark part of an image in the image data obtained by decompress | decompressing compressed data is removed is acquired.

  As described above, according to the present invention, image compression is performed in a direction in which unnecessary high spatial frequency component data is removed by attenuating the AC component of the low luminance level portion (dark portion) in the image data. Accordingly, an image compression apparatus and an image compression method that can further reduce the compression size can be obtained.

  In addition, when the compressed data obtained by compressing the image data in this way is decoded and reproduced and displayed, an effect of removing a noise component in a dark portion of the image can be obtained.

Hereinafter, embodiments of the present invention will be described.
(Embodiment 1)
FIG. 1 is a diagram for explaining an image compression apparatus according to Embodiment 1 of the present invention. FIG. 1 (a) shows the overall configuration, and FIG. 1 (b) shows luminance level extraction that constitutes the image compression apparatus. Shows the part.

  The image compression apparatus 100 according to the first embodiment performs a buffer 101 that divides image data Id so as to correspond to a block of 8 × 8 pixels, and image data (block data) Db corresponding to the block from the buffer. A DCT unit 102 that performs DCT processing, a quantizer 103 that quantizes DCT-processed image data (DCT data) Td, and a Huffman code that performs Huffman encoding of the quantized DCT data (quantized data) Qd These are the same as the buffer 201, the DCT unit 202, the quantizer 203, and the Huffman encoder 204 in the conventional image compression apparatus 200.

  When the luminance signal (image data) Id having a luminance level lower than the reference level (detection level) Lh supplied from the outside is input to the image compression apparatus 100 of the first embodiment, the quantization table AC is variable. A luminance level extractor 105 that outputs a control signal Cs to the creation unit 103a, and an AC component of a reference quantization table (hereinafter simply referred to as a normal table) Ta based on the control signal Cs from the luminance level extractor 105. It includes a quantization table (AC variable) creation unit 103a that creates a changed correction quantization table (hereinafter also simply referred to as a correction table) Tb. Here, the AC component of the quantization table is a component corresponding to the high frequency component of the 8 × 8 components of the DCT data Td in the quantization table.

  FIG. 1B is a diagram for explaining the luminance level extractor 105 in detail.

  The luminance level extractor 105 calculates the average luminance level of each block and the detection level Lh based on the detection level Lh supplied from the outside and the image data (block data) Bd corresponding to each block from the buffer 101. And a comparator 105a that outputs an AC component correction target signal AS and a coefficient E for the AC component of the quantization table as the control signal Cs in accordance with the comparison result. Here, the AC component correction target signal AS is a signal indicating whether or not the AC component should be corrected for the quantization table corresponding to the block to be quantized. The coefficient E is a multiplication coefficient (hereinafter also referred to as an X coefficient) used for correcting the AC component of the quantization table, and its value is from 1 in proportion to the low average luminance level for each block. It will be bigger.

  FIG. 2 is a diagram for explaining a control signal output from the luminance level extractor 105. FIG. 2A shows a change in the average luminance level (hereinafter referred to as luminance level) of each block with respect to the image position X. FIG. 2B shows the change of the AC component correction target signal AS with respect to the image position X, and FIG. 2C shows the change of the coefficient E with respect to the image position X. 2A to 2C, the image position X on the horizontal axis is the position in the entire image of the block to be compressed.

  FIG. 3 shows the relationship between the luminance level L and the coefficient Cx. As can be seen from this figure, when the luminance level L is equal to or higher than a certain detection level Lh, the coefficient Cx is set to 1, and when the luminance level L falls below the detection level Lh, the coefficient Cx decreases as the luminance level decreases. Large values A to G (1 <A <B <C <D <E <F <G) are set.

  Further, when the quantization table AC variable creation unit 103a receives the control signal Cs from the luminance level extraction unit 105, the quantization table AC variable creation unit 103a multiplies the AC component of the normal table Ta serving as a reference by the above coefficient to obtain a predetermined AC component other than the DC component. An AC conversion table (correction table) Tb having a larger value is output to the quantizer 103.

  FIG. 4 shows how the normal table Ta and the correction table Tb obtained by multiplying the predetermined AC component by the above coefficient are switched in the quantization table AC variable creation unit 103a.

  In FIG. 4, the horizontal axis represents the image position X, that is, the position of the block in the entire image to be compressed, and the vertical axis represents the average luminance level L of the block at that position.

  As shown in FIG. 4, for a block having an average luminance level L higher than the detection level Lh, the normal table Ta is used when quantizing the DCT data, and the average luminance level lower than the detection level Lh. For a block having DCT, a correction table Tb obtained by changing the predetermined AC component of the normal table Ta so that the AC component of the DCT data is further attenuated when the DCT data is quantized is used.

  FIG. 5 shows an example of specific conversion of the quantization table in the quantization table AC variable creation unit 103a.

  For a block having a luminance level equal to or higher than the detection level Lh, as shown in FIG. 5A, the X coefficient Cx is set to Cx = 1, so that the normal table Ta is multiplied by the X coefficient Cx. The obtained correction table Tb1 is the same as the original normal table Ta.

  For a block having a luminance level lower than the detection level Lh, as shown in FIG. 5B, the X coefficient Cx is set to Cx = 2, for example, and the original quantization table Ta of The block is quantized using a correction table Tb2 in which the AC component of a predetermined portion is multiplied by the X coefficient Cx (= 2) to increase the AC component.

  FIG. 6 shows the data structure of a compressed file Fc output as an encoded stream from the image compression apparatus according to the first embodiment.

  This compressed file Fc has a block corresponding part Bc corresponding to each block, and each block corresponding part Bc is composed of a header part Hb and a data part Dp, and the header part Hp corresponds to each block. The quantization table Qt used for quantization of the DCT data to be performed is included together with other headers Ha1 and Ha2, and the data portion Dp includes compressed data Cd corresponding to each block.

  Note that the specific numerical values of the components of the normal table Ta, the correction tables (tables after coefficient multiplication) Tb1 and Tb2 shown in FIG. 5 and the quantization table Qt shown in FIG. It goes without saying that the specific numerical value can take other values.

  Next, the operation will be described.

  When the image data Id is input to the image compression apparatus 100 according to the first embodiment, the image data Id is temporarily stored in the buffer 101, and the image data is divided into blocks, that is, horizontal 8 pixels × vertical 8 pixels. Is output for each area consisting of

  The image data (block data) Bd corresponding to each block is supplied from the buffer 101 to the DCT unit 102 and the luminance level extraction unit 105.

  Then, the block data Bd is subjected to DCT processing by the DCT unit 102 and converted to frequency domain data (DCT data) Td. At this time, the luminance level extraction unit 105 extracts the average luminance level in each block based on the block data Bd, and compares the extracted luminance level with the detection level Lh given from the outside. A control signal Cs corresponding to the comparison result is supplied to the quantization table generating unit 103a. In the quantization table generation unit 103a, the normal table Ta or a correction table Tb in which the AC component of a predetermined portion of the normal table Ta is increased is generated according to the control signal Cs from the luminance level extraction unit 105.

  Then, the DCT data from the DCT unit 102 is quantized by the quantizer 103 using the quantization table Qt (Ta or Tb) from the quantization table generator 103a, and the quantized data Qd is This is output to the Huffman encoder 104. The Huffman encoder 104 performs a Huffman encoding process on the quantized data Qd using the Huffman table Ht held in the Huffman table holding unit 104a, and outputs the compressed image data Cd obtained thereby. The

  Hereinafter, operations of the luminance level extraction unit 105, the quantization table creation unit 103a, and the quantizer 103 will be described in detail.

  That is, as shown in FIGS. 2A and 2B, when the luminance level (input image level) L obtained from the block data Bd is equal to or higher than the detection level Lh, it is output from the luminance level extraction unit 105. The AC component correction target signal AS, which is one of the control signals, is at the low level, and 1 is set to the coefficient Cx.

  In this case, as shown in FIG. 7A, each frequency component of the 8 × 8 array, which is the block data Td that has been subjected to the DCT processing, is a quantization coefficient at a corresponding position of the 8 × 8 array in the normal table Ta. Divided by (quantization component), quantized data Qd composed of division values of an 8 × 8 array is generated. Here, in FIG. 7A, the specific numerical values of the quantization coefficients (quantization components) of the normal table Ta are the normal tables shown in FIGS. 5A and 5B for convenience of explanation. The Ta quantization coefficient (quantization component) is a different value from the specific value.

  The quantization table Ta used for quantization is added to the header portion of the data area corresponding to each block in the compressed data Cd and transmitted to the receiving side. Therefore, when decompressing the compressed data on the receiving side, as shown in FIG. 7B, it is included in the data (quantized data) Qd before inverse quantization and the header of the compressed data. Dequantized data is created by multiplication with the quantization table Ta.

  On the other hand, when the luminance level (input image level) L obtained from the block data Bd is smaller than the detection level Lh, the AC component correction target signal AS, which is one of the control signals output from the luminance level extraction unit 105, is High. The coefficient Cx is set to a value corresponding to the low luminance level L. Specifically, as shown in FIG. 3, as the luminance level is smaller, the coefficient Cx is set to a larger value A to G (1 <A <B <C <D <E <F <G).

  In this case, as shown in FIG. 8A, each frequency component of the 8 × 8 array, which is the block data Td subjected to the DCT processing, has a coefficient Cx (Cx = 2) in the high-frequency part of the quantization coefficient of the normal table. ) Is multiplied by the quantization coefficient at the corresponding position in the 8 × 8 array of the correction table Tb obtained by multiplying () to generate quantized data Qd composed of the divided values in the 8 × 8 array. Here, the quantization coefficients to be multiplied by the coefficients of the normal table are the coefficients from the third row, the third row to the eighth column of the normal table, and the coefficients from the fourth column to the eighth column.

  At this time, not the correction table Tb used for quantization but the normal table before correction is added to the header portion of the data area corresponding to each block in the compressed data Cd and transmitted to the receiving side.

  That is, the quantization table attached to the compressed data is attached with a table before the quantization coefficient of the normal table Ta is multiplied by 2 (that is, the same as the normal table Ta). In this case, the attached normal table is used when the receiving side performs inverse quantization in the decompression process of the compressed data. As a result, the inverse quantized data obtained by inverse quantizing the quantized data has an AC component smaller than that of the inverse quantized data before the quantization coefficient of the quantized data is multiplied by 2.

  That is, as shown in FIG. 8B, the result (frequency component of the inverse quantized data) obtained by multiplying the frequency component of the quantized data by the quantization coefficient of the quantization table attached to the compressed file is FIG. 7 (b) shows a comparison with the result (frequency component of inverse quantized data) obtained by multiplying the frequency component of quantized data by the quantization coefficient of the quantization table attached to the compressed file. Ingredients become smaller.

  When the AC component is attenuated in this manner, the band is narrowed, but there should be no finely patterned image as the original image in the dark portion. If there is, the noise of the image sensor can be considered, but it can be deleted by attenuation in this way. Further, since the AC component of the low luminance component is further reduced as compared with the normal compression, there is an effect that the noise component in the dark part is removed when the compressed data is restored.

  Thus, in the present embodiment, unnecessary high spatial frequency component data is removed by attenuating the AC component in the block of the low luminance level (dark portion) in the image, and the compression size is reduced. It can be further reduced.

  Also, the high spatial frequency component is removed by attenuating the AC component of the low luminance level portion (dark portion) in the image. As a result, a fine noise component in an apparently dark portion is removed.

  Although not specifically described in the above embodiment, the image compression apparatus of the above embodiment is a digital camera such as a digital video camera or a digital still camera used in the data communication unit, an image input camera, a scanner, or a facsimile. An electronic information device having an image input device such as a camera-equipped mobile phone will be described. An electronic information device according to the present invention includes a compression processing unit that compresses image data obtained by an imaging unit, a memory unit such as a recording medium that records the compressed image data, and decompresses compressed data recorded in the recording unit. A decompression processing unit, a display unit such as a liquid crystal display device that displays the decompressed image data on a display screen such as a liquid crystal display screen after predetermined signal processing for display, and the compressed image data for communication At least the compression processing unit of the communication means such as a transmission / reception device for performing communication processing after predetermined signal processing and the image output means for printing (printing) and outputting (printing out) the image data. The compression processing unit includes the image compression apparatus according to the embodiment of the invention.

  As mentioned above, although this invention has been illustrated using preferable embodiment of this invention, this invention should not be limited and limited to this embodiment. It is understood that the scope of the present invention should be construed only by the claims. It is understood that those skilled in the art can implement an equivalent range from the description of specific preferred embodiments of the present invention based on the description of the present invention and common general technical knowledge. Patents, patent applications, and documents cited herein should be incorporated by reference in their entirety, as if the contents themselves were specifically described herein. Understood.

  The present invention can reduce the size of compressed data while suppressing deterioration of image quality in an electronic apparatus having a data compression unit that compresses and records or transmits an image, such as a camera-equipped mobile phone or a digital still camera.

FIG. 1 is a diagram for explaining an image compression apparatus according to an embodiment of the present invention. FIG. 1 (a) shows an overall configuration, and FIG. 1 (b) shows a luminance level extractor constituting the image compression apparatus. It is a figure for demonstrating in detail. FIG. 2 is a diagram for explaining a control signal output from the luminance level extractor of the above embodiment, in which the luminance level of each block (FIG. (A)), the AC component correction target signal AS (FIG. (B)), And the coefficient (FIG. (C)) which corrects a quantization table is shown. FIG. 3 is a diagram for explaining a method of determining a specific value of the coefficient Cx from the luminance level L. FIG. 4 is a diagram illustrating an operation of switching between the normal table Ta and the correction table Tb in the quantization table AC variable creation unit in the image compression apparatus of the above embodiment. FIG. 5 is a diagram for explaining a specific process for correcting a quantization table using a coefficient in the quantization table AC variable creation unit in the image compression apparatus of the above embodiment, and the quantization when the coefficient Cx = 1. Table correction (FIG. (A)) and quantization table correction when the coefficient is 2 (FIG. (B)) are shown. FIG. 6 is a diagram illustrating a quantization table attached to the header of the compressed file output from the image compression apparatus according to the embodiment. FIG. 7 is a diagram for explaining the operation of the image compression apparatus according to the above embodiment. Quantization using a normal table (FIG. 7A) and inverse quantization using a normal table (FIG. 7B). Show. FIG. 8 is a diagram for explaining the operation of the image compression apparatus according to the embodiment described above. The quantization using the correction table (FIG. 8A) and the inverse quantization using the normal table (FIG. 8B). Show. FIG. 9 is a block diagram illustrating a conventional image compression apparatus that performs JPEG-compliant image compression processing.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Image compression apparatus 101 Buffer 102 DCT device 103 Quantizer 103a Quantization table (AC variable) creation part 104 Huffman encoder 104a Huffman table holding part 105 Luminance level extractor 105a Comparator AS AC component correction object signal Bc block correspondence Part Bd block data Cd compressed data Cs control signal Cx coefficient Dp data part Fc compressed file Ha1, Ha2 other header information Hp header part Ht Huffman code table Id image data Qd quantized data Qt quantization table Ta normal table Tb correction table Td DCT data

Claims (20)

An image compression device that compresses an image to generate compressed data,
An image dividing unit that divides the image into blocks serving as compression units;
A quantization unit that quantizes image data of the image for each block;
The image compression apparatus, wherein the quantization unit determines a component of a quantization table used for quantization of the image data for each block so that a compression ratio of a darker portion of the image than a reference brightness is increased. An encoder for generating compressed data by encoding quantized data obtained by quantizing the image data;
The image compression apparatus according to claim 1, wherein a quantization table corresponding to each block is transmitted to the decoding side together with the compressed data of each block.   The image compression apparatus according to claim 2, wherein the quantization table corresponding to each block is transmitted by being included in a header portion corresponding to each block of the compressed data. The quantization unit is
A reference quantization table is held, a correction quantization table is created for each block by correcting the components of the reference quantization table according to the luminance level of each block, and the image data of each block is quantized The image compression apparatus according to claim 1, wherein the image compression apparatus performs the correction using the created correction quantization table.   The image compression apparatus according to claim 4, wherein the correction quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.   The image compression apparatus according to claim 4, wherein the reference quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block. A converter that generates discrete data by performing discrete cosine transform on the image data of the image for each block,
The quantization unit quantizes the conversion data as the image data, and an AC component corresponding to an AC component of the conversion data among the components of the reference quantization table according to the luminance level of each block. The image compression apparatus according to claim 4, wherein the correction quantization table is created by changing.   The quantization unit changes the high-frequency component corresponding to a component having a high spatial frequency of the conversion data, among the AC components of the reference quantization table, according to the luminance level of each block, and performs the correction quantization The image compression apparatus according to claim 7, which creates a table. The quantization unit is
A luminance level extraction unit that extracts the luminance level of each block based on the image data;
A quantization table creation unit that holds the reference quantization table and corrects high-frequency components of the reference quantization table according to the extracted luminance level of each block to create the corrected quantization table for each block;
The image compression apparatus according to claim 8, further comprising: a quantizer that quantizes the converted data of each block using the created correction quantization table. The quantization table creating unit creates the corrected quantization table by multiplying a high frequency component of the reference quantization table by a coefficient according to a luminance level of each block,
The luminance level extraction unit compares a detection level supplied from the outside with a luminance level corresponding to each block, and if the luminance level of the block is equal to or higher than the detection level, quantization for the block is performed When a correction control signal indicating that the reference quantization table should be used as a table is output and the coefficient is set to 1 and the luminance level of the block is smaller than the detection level, The image compression according to claim 9, wherein a correction control signal indicating that the correction quantization table should be used is output as a quantization table, and the coefficient is set to a larger value as the luminance level of the block is lower. apparatus. An image compression method for generating compressed data by compressing an image,
An image dividing step for dividing the image into blocks that are units of compression;
A quantization step for quantizing the image data of the image for each block,
An image compression method in which, in the quantization step, a component of a quantization table used for quantization of the image data is determined for each block so that a compression ratio of a portion darker than a reference brightness of the image is increased. An encoding step of generating compressed data by encoding quantized data obtained by quantizing the image data;
The image compression method according to claim 11, wherein the quantization table corresponding to each block is transmitted to the decoding side together with the compressed data of each block.   The image compression method according to claim 12, wherein the quantization table corresponding to each block is transmitted by being included in a header portion corresponding to each block of the compressed data. In the quantization step,
A correction quantization table is created for each block by correcting the components of the reference quantization table according to the luminance level of each block, and the image data of each block is quantized using the created correction quantization table. The image compression method according to claim 11, wherein the image compression method is performed.   The image compression method according to claim 14, wherein the correction quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block.   The image compression method according to claim 14, wherein the reference quantization table is transmitted to the decoding side together with the compressed data of each block as a quantization table corresponding to each block. A transform step for generating transform data by performing discrete cosine transform on the image data of the image for each block;
In the quantization step, the conversion data is quantized as the image data, and among the components of the reference quantization table, an AC component corresponding to an AC component of the conversion data is determined according to a luminance level of each block. The image compression method according to claim 14, wherein the correction quantization table is created by changing.   In the quantization step, the correction quantization is performed by changing a high frequency component corresponding to a component having a high spatial frequency of the conversion data, among AC components of the reference quantization table, according to a luminance level of each block. The image compression method according to claim 17, wherein a table is created. The quantization step includes:
Based on the image data, extract the luminance level of each block,
The correction quantization table is created for each block by correcting the high frequency component of the reference quantization table according to the extracted luminance level of each block,
The image compression method according to claim 18, wherein the converted data of each block is quantized using the created correction quantization table. In the quantization step, a high-frequency component of the reference quantization table is multiplied by a coefficient according to the luminance level of each block to create the corrected quantization table,
The determined detection level is compared with the luminance level corresponding to each block. When the luminance level of the block is equal to or higher than the detection level, the reference quantization table is used as the quantization table for the block. A correction control signal indicating that the block should be generated, and when the coefficient is set to 1 and the luminance level of the block is smaller than the detection level, the correction quantization table as the quantization table of the block The image compression method according to claim 19, wherein a correction control signal indicating that the block is to be used is generated, and the coefficient is set to a larger value as the luminance level of the block is lower.

Images