JP6204171B2 - Image encoding apparatus and image decoding apparatus - Google Patents

Description

  The present invention relates to an image encoding device that compresses and encodes image data, and an image decoding device that decompresses and decodes image encoded data.

  2. Description of the Related Art Conventionally, there is an image encoding device that compresses and encodes image data for the purpose of reducing the burden on communication, etc., and image decoding that decompresses the encoded image data that is the reverse process and restores the original image An apparatus is also known. In recent years, an encoding standard that can be compressed at a higher compression rate has been proposed (see Non-Patent Document 1).

  An image encoding device and an image decoding device according to the standard will be described. First, FIG. 11 shows a schematic configuration of the image encoding device. As shown in FIG. 11, the image encoding apparatus 100 includes a color space conversion unit 110, a frequency conversion unit 120, a quantization unit 130, a coefficient prediction unit 140, an adaptive scan unit 150, a VLC encoding unit 160, a bit. The stream output unit 170 is configured. Note that VLC in the name of the VLC encoding unit 160 is an abbreviation for Variable Length Coding (Variable Length Coding).

  The image data is composed of pixels (pixels), which are the minimum unit of an image, and a macroblock having an image area of 256 pixels divided by 16 pixels (horizontal direction) × 16 pixels (vertical direction) as one unit, The macro block is composed of 4 pixel × 4 pixel unit blocks obtained by dividing each macro block into four in the horizontal and vertical directions. In the image processing according to the standard, the processing proceeds in units of the macroblock and the block.

  The color space conversion unit 110 performs color space conversion processing on the input image data for each pixel in units of macroblocks. The color space conversion process depends on the pixel format defined in the standard of Non-Patent Document 1. When the pixel format requires color space conversion, the original image data in the RGB format is converted into the YUV format, and the original image data in the CMYK format is converted into the YUVK format. The Further, when the pixel format does not require color space conversion, the original image data is processed so as not to be color space converted. Then, the color space conversion unit 110 inputs the processed image data to the frequency conversion unit 120.

  The frequency conversion unit 120 converts the image data that has been color space processed by the color space conversion unit 110 into a raster order in units of macroblocks, and within each macroblock, in the order of the color components after the color space conversion in each macroblock. For each block, processing (frequency conversion) for converting the data of each pixel into coefficient data for each frequency band is performed. The frequency conversion unit 120 performs frequency conversion on the 16 pixel data of 4 × 4 pixels in each block and converts the pixel data into coefficient data, and then removes the remaining data except the coefficient data on the upper left of the 16 blocks. Fifteen coefficient data are collected to form first layer data. This is HP coefficient (high pass coefficient) data which is a high frequency component. Next, the coefficient data at the upper left of each of the 16 blocks are collected to form 4 × 4 16 coefficient data, and the collected 4 × 4 coefficient data is further frequency-converted to obtain 4 × 4. The second hierarchical data is formed by the remaining 15 coefficient data except the one coefficient data at the upper left of the 4 × 4 coefficient data subjected to frequency conversion, and after the second frequency conversion. Among the 4 × 4 coefficient data, the third hierarchical data is formed by one coefficient data at the upper left. The second layer data corresponds to LP coefficient (low-pass coefficient) data that is a low frequency component, and the third layer data corresponds to DC coefficient (DC coefficient) data that is a direct current component. As described above, the frequency conversion unit 120 generates the first layer data (HP coefficient), the second layer data (LP coefficient), and the third layer data (DC coefficient), and each of these layer data is quantized. Input to the unit 130.

  The quantization unit 130 quantizes the first layer data, the second layer data, and the third layer data after the frequency conversion input from the frequency conversion unit 120 using selected quantization parameters. Processing is performed, and the processed first layer data, second layer data, and third layer data are input to the coefficient prediction unit 140.

  As shown in FIG. 12, the quantization unit 130 includes a quantization parameter storage unit 111, an iMan value generation unit 112, an iExp value generation unit 113, a left barrel shifter 114, and a division unit 115.

  The quantization process is performed using a quantization parameter that is a numerical value for dividing the coefficient data frequency-converted by the frequency converter 120. The quantization parameter is input to the quantization parameter storage unit 111 from the external controller or the like before the compression process is started, and is stored in the quantization parameter storage unit 111.

  Different quantization parameters are used for each of the first layer data, the second layer data, and the third layer data. The quantization parameter is selected and output from the quantization parameter storage unit 111. This is indicated by QP [7: 0]. This quantization parameter QP consists of an 8-bit numerical value, and [7: 0] of the above-mentioned QP [7: 0] indicates that it has 8 bits in the arrangement order of bit 7 to bit 0. ing. The bit notation shown below is the same notation.

  In the standard of Non-Patent Document 1, the quantization process is performed by iMan generated from information indicating whether or not the upper bits (bits 7 to 4) of the quantization parameter (QP) are all 0 and the lower bits (bits 3 to 0). Quantization scaling factor (QSF), which is data obtained by bit-shifting the value data to the left by the value represented by the data called iExp value generated from the upper bits (bits 7 to 4) of quantization parameter QP In other words, the input coefficient data is divided by this quantization scaling factor and quantized.

  This process will be described with reference to FIG. 12. The quantization parameter QP [7: 0] output from the quantization parameter storage unit 111 is input to the iMan value generation unit 112 and the iExp value generation unit 113.

  Here, the configuration of the iMan value generation unit 112 is shown in FIG. As shown in FIG. 13, the iMan value generation unit 112 includes a bit cutout unit 116, an adder 117, a comparison unit 118, and a selector 119.

  The quantization parameter QP input from the quantization parameter storage unit 111 is input to the bit cutout unit 116 and the comparison unit 118. The bit cutout unit 116 cuts out QP [3: 0], which is the lower 4 bits, of the input quantization parameter QP [7: 0], and inputs the input port A of the selector 119 and the adder. The adder 117 adds the numerical value 16 to the lower bits QP [3: 0] of the input quantization parameter, and inputs the addition result to the input port B of the selector 119.

  The comparison unit 118 compares the input quantization parameter QP [7: 0] with the numerical value 16, and if the QP is smaller than 16 (all the high-order bits of QP are 0), the numerical value 1 (result is True), otherwise, the value 0 (result is false) is output, and the output is input to the input port C of the selector 119.

  If the value of the input port C input from the comparison unit 118 is 1 (the result is true), the selector 119 performs the quantization input to the input port A of the selector 119 from the bit cutout unit 116. If QP [3: 0], which is the lower 4 bits of the parameter, is selected and the value of the input port C is 0 (result is false), the input port B of the selector input from the adder 117 is input. A result obtained by adding QP [3: 0], which is the lower 4 bits of the input quantization parameter, and a numerical value 16 is selected, and the selected data is output as iMan [4: 0]. Input to the barrel shifter 114.

  Further, the iExp value generation unit 113 cuts out the upper 4 bits QP [7: 4] from the quantization parameter QP [7: 0] input from the quantization parameter storage unit 111, and extracts this bit. Is input to the left barrel shifter 144 as iExp [3: 0].

  The left barrel shifter 114 bit-shifts iMan [4: 0] input from the iMan generation unit to the left by the number represented by iExp [3: 0] input from the iExp generation unit. Then, a quantization scaling factor (QSF [19: 0]) that is a divisor of quantization is generated, and this QSF [19: 0] is input to the division unit 115. Note that iMan is composed of 5 bits, and iExp, which is a shift value composed of 4 bits, takes a numerical value of 0 to 15. Therefore, since the quantization scaling factor is generated by shifting the 5-bit iMan to the left by a maximum of 15 bits, the quantization scaling factor has a maximum bit length of 20 bits.

  The division unit 115 divides the first layer data, the second layer data, and the third layer data, which are coefficient data input from the frequency conversion unit, by a quantization scaling factor input from the left barrel shifter 114. Then, the division result is input to the coefficient prediction unit 140 as quantized coefficient data.

  The coefficient prediction unit 140 is configured to calculate coefficient values between adjacent blocks for the first layer data, the second layer data, and the third layer data, which are the quantized coefficient data input from the quantization unit 130, respectively. A prediction mode is calculated by comparison, and a process of subtracting between blocks having high correlation in the direction of the derived prediction mode is performed based on a predetermined arithmetic expression. Then, the coefficient prediction unit 140 inputs the processed first layer data, second layer data, and third layer data to the adaptive scan unit 150.

  The adaptive scan unit 150 performs a process of replacing the coefficient data for each block with respect to the first layer data and the second layer data input from the coefficient prediction unit 140. In this replacement process, non-zero coefficient data is collected on the side where the pixel position in the block after frequency conversion is small (low frequency side), and the zero value is continuous on the side where the pixel position in the block is large (high frequency side). Since coefficient data is collected, the compression rate of encoding can be increased. Then, the adaptive scan unit 150 inputs the first layer data, the second layer data, and the third layer data on which the replacement process is not performed to the VLC encoding unit 160.

  The VLC encoder 160 performs entropy encoding on the first layer data, the second layer data, and the third layer data processed by the adaptive scan unit 150 to generate a code stream. The code stream is input to the bit stream output unit 170.

  The bit stream 170 sequentially concatenates the code stream encoded by the VLC encoding unit 160 to the image setting data as header information, and collects one file as a bit stream for one image data. Create and output the formed file to the outside.

  Next, FIG. 14 shows a schematic configuration of an image decoding apparatus according to the standard. As shown in FIG. 14, the image decoding apparatus 200 includes a bit stream analysis unit 210, a VLC decoding unit 220, an adaptive scan unit 230, a coefficient inverse prediction unit 240, an inverse quantization unit 250, and a frequency inverse transform unit. 260 and a color space inverse transform unit 270. The process of the image decoding apparatus is performed in the reverse procedure of the process of the above-described image encoding apparatus, and the configuration of the image data used is the same as that of the image encoding apparatus. The first layer data (HP coefficient), second layer data (LP coefficient), and third layer data (DC coefficient) are restored to image data in units of macroblocks and blocks. Hereinafter, the operation of each component of the image decoding apparatus 200 will be described in order.

  The bitstream analysis unit 210 analyzes a bitstream that is input compressed data for one image, extracts image information, separates encoded image data, and inputs the encoded image data to the VLC decoding unit 220 To do.

  The VLC decoding unit 220 converts the first layer data (HP coefficient), the second layer data (LP coefficient), and the third layer data (DC coefficient), which are the image encoded data separated by the bit stream analysis unit 210. On the other hand, variable length decoding processing which is VLC decoding processing is performed, and the decoded coefficient data is output to the adaptive scan unit 230.

  The adaptive scan unit 230 replaces the coefficient data of the first layer data and the second layer data whose pixel order is replaced by a method determined at the time of compression by a procedure reverse to the compression, and returns the original pixel order. Process. This replacement process collects non-zero coefficients on the lower side of the frequency band at the upper left of the block during compression (conversely, those with zero coefficient values gather on the higher side of the lower right frequency band) This is performed in order to increase the compression rate in encoding.

  The coefficient inverse prediction unit 240 uses adjacent blocks for the first layer data, the second layer data, and the third layer data input from the VLC decoding unit 220, which are input from the adaptive scan unit 230, respectively. The prediction mode is derived by comparing the coefficients between them, and the corresponding pixel value of the highly correlated block is added to the target pixel according to the set arithmetic expression for the prediction mode. The pixel value before processing is restored (reverse prediction). Then, the first hierarchy data, the second hierarchy data, and the third hierarchy data after inverse prediction are input to the inverse quantization unit 250.

  The inverse quantization unit 250 performs inverse quantization on the first layer data, the second layer data, and the third layer data inversely predicted by the coefficient inverse prediction unit 240 using the set quantization parameters. And the processed first layer data, second layer data, and third layer data are input to the frequency inverse transform unit 260 for each block.

  As shown in FIG. 15, the inverse quantization unit 250 includes a quantization parameter storage unit 211, an iMan value generation unit 212, an iExp value generation unit 213, a left barrel shifter 214, and a multiplier 215.

  This is the same configuration as that of the quantization unit 130 of the image encoding device 100 in FIG. 12 except that the division unit 115 is replaced with a multiplier 215.

  That is, the iMan generating unit 212 generates information based on information indicating whether the upper 4 bits are all 0 and the lower 4 bits in the QP [7: 0] that is the quantization parameter held in the quantization parameter storage unit 211. The 5-bit length iMan [4: 0] and the 4-bit length iExp [3: 0] generated from the upper 4 bits are input to the left barrel shifter 214, and the left barrel shifter 214 receives the iMan [4 : 0] is shifted to the left by the number indicated by iExp [3: 0] to generate a quantized scaling factor (QSF [19: 0]), which is input to the multiplier 215.

  The multiplier 215 multiplies the inverse prediction coefficient data input from the coefficient inverse prediction unit 240 by the quantization scaling factor input from the left barrel shifter 214 to generate inverse quantization coefficient data, and the frequency Input to the inverse transform unit 260.

  The frequency inverse transform unit 260 sequentially processes the coefficient data inversely quantized by the inverse quantization unit 250 in the raster direction with 4 × 4 pixels as one block, and converts the coefficient data for each frequency band into pixel data. Perform the conversion process. The frequency inverse transform unit 260 converts the first layer data, the second layer data, and the third layer data separated for each frequency band into a two-dimensional space of one macroblock (16 × 16 pixels) for each color component. The data is returned to the color space inverse transform unit 270.

  The color space inverse conversion unit 270 processes the image data for each color component input from the frequency inverse conversion unit 260 for each pixel, and when color space reverse conversion is unnecessary, the color space reverse conversion is not performed. When space inverse transformation is necessary, color space inverse transformation is performed and the processed pixel data is output to the outside as image data for one image.

As described above, whether or not color space reverse conversion is necessary depends on information called a pixel format representing an image format defined in the standard of Non-Patent Document 1, and the pixel format requires conversion. In some cases, image data in YUV format is converted to RGB format, and image data in YUVK format is converted to CMYK format. Further, when the pixel format does not require color space reverse conversion, color space reverse conversion is not performed.

JPEG XR Recommendation ITU-T T.832: Information technology -JPEG XR image coding system-Image coding specification. [Online] .Recommendation T.832 (01/12). Approval 2012-01-13. Status: In force. 2012-9-14, ITU (International Telecommunication Union). [Retrieved on 2013-10-29]. Retrieved from the Internet: <URL: http://www.itu.int/rec/T-REC-T.832 -201201-I / en>.

  By the way, the coefficient data and its divisor data input to the division unit in the quantization unit of the image encoding device, and the coefficient and its multiplier data input to the multiplier in the inverse quantization unit of the image decoding device are the bits. It is known that as the width increases, the circuit scale constituting the division unit and multiplier increases, and the processing speed of the division unit and multiplier decreases.

  As the divisor data of the division unit 115 in the quantization unit 130 of the conventional image encoding device 100 shown in FIG. 12 and the multiplier data of the multiplier 215 in the inverse quantization unit 250 of the conventional image decoding device 200 shown in FIG. As described above, iMan generated from the information indicating whether all the upper bits of the quantization parameter are 0 and the lower bits and iExp generated from the upper bits of the quantization parameter are bit-shifted to the left. A 20-bit long quantization scaling factor is used.

FIG. 16 shows the relationship between the quantization parameter (QP) and the quantization scaling factor (QSF). While the quantization parameter varies from 1 to 255, the quantization scaling factor varies from 00002 _{(16) to} F8000 _{(16) in} hexadecimal notation and from 2 to 1015808 in decimal notation. Since bit shift in the left direction is performed using iExp generated from the higher-order bits of the quantization parameter, the change in the quantization scaling factor with respect to the change in the quantization parameter increases as the quantization parameter increases. Suddenly grows.

  The division data in the quantization unit of the conventional image coding apparatus and the multiplication data in the conventional image decoding apparatus are large values as represented by 20 bits as described above, and division and multiplication are performed using this. Therefore, there has been a problem that the circuit scale of the multiplier used in the division unit and the inverse quantization unit used in the quantization unit is increased, and the processing speed of the multiplier and the division unit is reduced.

  The present invention has been made in view of the above circumstances, and the division data in the quantization unit of the image encoding device and the value of the multiplication data in the inverse quantization unit of the image decoding device are converted into the values of the quantization parameters. Providing an image encoding device including a quantization unit capable of performing high-speed processing, and an image decoding device including an inverse quantization unit so as not to increase extremely with respect to the increase, the circuit scale being small And its purpose.

In order to solve the above-described problem, an image encoding device according to the present invention includes:
The image data is frequency-converted for a macroblock consisting of m blocks in each horizontal and vertical direction in units of a block consisting of n pixels in each horizontal and vertical direction, and the upper left pixel position of each block in the macroblock The data corresponding to the pixel position other than the above is used as the coefficient data of the high frequency component, the data corresponding to the upper left pixel position of each block in the macro block is collected, further frequency-converted, and the upper left pixel position after the frequency conversion. The frequency conversion unit which makes the corresponding data the coefficient data of the direct current component, and the data corresponding to the pixel position other than the upper left the coefficient data of the low frequency component,
Quantization of the coefficient data of each component frequency-converted by the frequency conversion unit by using a quantization parameter having a (p + q) bit length held in an internal or external storage area to obtain quantized coefficient data And
A coefficient scanning unit that performs replacement processing of pixels in the block of quantized coefficient data of the low frequency component and high frequency component quantized by the quantizing unit to obtain replacement coefficient data;
An encoding unit that generates encoded data by encoding the quantized coefficient data of the DC component after quantization by the quantizing unit and the low-frequency component and high-frequency component replacement coefficient data after the replacement processing by the coefficient scanning unit In an image encoding device comprising:
The quantization unit is
An iExp value generation unit for generating an iExp value having a p-bit length, which is data obtained by cutting out the upper p bits of the quantization parameter;
When all the upper p bits of the quantization parameter are 0, the lower q bits of the quantization parameter are cut out, and the value 0 is assigned to the bit above the most significant bit (Q + 1) bit length iMan value is generated, and when any one of the upper p bits of the quantization parameter is 1, the lower q bits of the quantization parameter are cut out and the An iMan value generation unit that generates an iMan value having a length of (q + 1) bits, which is data obtained by assigning a value of 1 to a bit above the upper bit,
A division unit that divides the coefficient data of the direct current component, the low frequency component, and the high frequency component that are frequency-converted by the frequency conversion unit by the iMan value to obtain divided data;
A right barrel shifter that shifts the data divided by the division unit to the right by the value represented by the iExp value generated by the iExp value generation unit and generates quantized coefficient data; However, n and m are integers of 2 or more, and p and q are integers of 1 or more.

  According to the image coding apparatus according to the present invention having the above-described configuration, first, image data for one image is input to the frequency conversion unit, and the image data is divided into n in each of the horizontal and vertical directions. Using a block of pixels as a unit, frequency conversion is performed on a macroblock consisting of m blocks in each of the horizontal and vertical directions to obtain coefficient data of a high frequency component, a low frequency component, and a direct current component, and input to the quantization unit Is done.

  Then, in the quantization unit, the frequency-converted coefficient data of each component is quantized using a quantization parameter held in an internal or external storage area.

  The quantized coefficient data is input to the coefficient prediction unit. The coefficient prediction unit derives a prediction mode and performs a process of subtracting between highly correlated blocks, and then inputs the coefficient data to the coefficient scan unit. In the scanning unit, processing for changing the arrangement order of the pixels in the block is performed so that non-zero coefficient data is concentrated on the low frequency side in the block, and the coefficient data in the block unit after the replacement is encoded in the encoding unit. And encoded data is generated.

  Then, in the quantization unit, in the iExp value generation unit, the upper p bits of the quantization parameter having a length of (p + q) bits are cut out to generate an iExp value having a p bit length, and in the iMan value generation unit. Then, an iMan value having a length of (q + 1) bits is generated from information indicating whether all the upper p bits of the quantization parameter are 0 and data obtained by cutting out the lower q bits of the quantization parameter. The coefficient data of each component frequency-converted by the frequency conversion unit is divided by the iMan value generated by the iMan value generation unit, and the division data divided by the division unit is converted to the iExp value in the right barrel shifter. The numerical value indicated by the iExp value generated by the generation unit is bit-shifted rightward and quantized.

  As described above, according to the configuration of the present invention, the coefficient data of each component frequency-converted by the frequency conversion unit is generated from iMan generated from information indicating whether or not all the upper bits of the quantization parameter are 0 and the lower bits. Since the division data is bit-shifted to the right by the iExp value generated from the higher-order bits of the quantization parameter, the coefficient data of each component is moved to the left by the iMan value as in the conventional configuration. In this case, the data is not divided by a large numerical value called a quantization scaling factor that is data bit-shifted by the iExp value, the circuit scale of the divider that performs the division can be reduced, and the processing speed of the division can be improved. Thus, the image encoding process can be optimized.

Furthermore, the division unit according to the image encoding device of the present invention is:
A reciprocal table unit that has a storage area that holds a reciprocal value corresponding to each iMan value, and selects a reciprocal value corresponding to the iMan value generated by the iMan value generation unit;
A multiplier that multiplies the coefficient data of the DC component, the low frequency component, and the high frequency component frequency-converted by the frequency conversion unit by the reciprocal value of the iMan value selected by the reciprocal table unit.

  According to the division unit in the image encoding device according to the present invention having the above configuration, first, in the reciprocal table unit, the reciprocal corresponding to the iMan value generated by the iMan value generating unit is stored in the storage area in the reciprocal table unit. The reciprocal of each iMan value held is selected, and the coefficient data of each component frequency-converted by the frequency converter in the multiplier is multiplied by the reciprocal value of iMan selected by the reciprocal table unit to obtain the coefficient Data quantization is performed.

  As described above, according to the configuration of the present invention, the reciprocal is generated by using the information indicating whether the upper bits of the quantization parameter are all 0 and the iMan value having a small bit width generated from the bit extraction data of the lower bits. Therefore, the reciprocal data of the reciprocal data when the reciprocal is taken compared to the case of using the quantized scaling factor that is the data obtained by bit shifting the iMan value to the left by the iExp value generated from the higher order bits of the quantization parameter. The bit width can be reduced. Further, since the iMan value is generated from the information indicating whether all the upper bits of the quantization parameter are 0 and the bit cut-out data of the lower bits, the iMan value is based on the reciprocal number compared to the conventional quantization parameter as a whole. The number of numerical values can be reduced, and the number of reciprocals to be held can also be reduced. Therefore, it is possible to reduce the circuit scale of the reciprocal table unit and the multiplier of this configuration, improve the processing speed of the division unit, and optimize the image encoding process.

The image decoding apparatus of the present invention is
Decodes compressed image data in which coefficient data consisting of DC components, low-frequency components, and high-frequency components in the image data after frequency conversion is encoded, and consists of DC components, low-frequency components, and high-frequency components A decryption unit for generating decrypted data;
A coefficient scanning unit that performs replacement processing of pixels in the block of the decoded data of the low-frequency component and the high-frequency component decoded by the decoding unit to obtain replacement coefficient data;
The DC component decoded data decoded by the decoding unit, and the low frequency component and high frequency component coefficient data in which the arrangement order of pixels in the block is changed by the coefficient scanning unit An inverse quantization unit that performs inverse quantization using an 8-bit length quantization parameter held in a storage area to obtain inverse quantization coefficient data;
An image area restored by inverse frequency transform of the inverse quantization coefficient data is composed of blocks of n pixels in each of the horizontal and vertical directions, and from a macro block having m blocks in each of the horizontal and vertical directions. As the image area, the inverse quantization coefficient data of the direct current component is arranged at the pixel position at the upper left of the m area in each of the horizontal and vertical directions, and the inverse quantization of the low frequency component is performed at a pixel position other than the upper left of the area. The coefficient data of each pixel position in the region subjected to frequency inverse transform for the region is disposed, and the coefficient data of each pixel position in the region subjected to the frequency inverse transform is converted into each block in the macroblock corresponding to the position corresponding to the pixel position in the region, It is arranged at the pixel position at the upper left, and the inverse quantization coefficient data of the high frequency component is arranged at a pixel position other than the pixel position at the upper left of each block in the macro block. Every further by frequency inverse conversion, the image decoding apparatus and a frequency inverse conversion unit for the frequency inversion data,
The inverse quantization unit includes:
An iExp value generation unit for generating an iExp value having a p-bit length, which is data obtained by cutting out the upper p bits of the quantization parameter;
When all the upper p bits of the quantization parameter are 0, the lower q bits of the quantization parameter are cut out, and the value 0 is assigned to the bit above the most significant bit (Q + 1) bit length iMan value is generated, and when any one of the upper p bits of the quantization parameter is 1, the lower q bits of the quantization parameter are cut out and the An iMan value generation unit that generates an iMan value having a length of (q + 1) bits, which is data obtained by assigning a value of 1 to a bit above the upper bit,
The iMan value is multiplied by the decoded data of the DC component decoded by the decoding unit and the coefficient data of the low frequency component and the high frequency component in which the arrangement order of the pixels in the block is changed by the coefficient scanning unit. A multiplier for multiplication data,
A left barrel shifter for generating dequantized coefficient data by bit-shifting the multiplication data multiplied by the multiplier to the left by a value represented by the iExp value generated by the iExp value generation unit; However, n and m are integers of 2 or more, and p and q are integers of 1 or more.

  According to the image decoding apparatus according to the present invention having the above-described configuration, first, encoded data corresponding to image data for one image is input to the decoding unit, and in the decoding unit, a DC component, a low frequency It is decoded for each component and high-frequency component, and is input to the coefficient scanning unit as decoded data. In the coefficient scanning unit, non-zero coefficient data is concentrated on the low frequency side in the block. The processing is performed to change the order of the pixels in the block, and the coefficient data of the block unit after the replacement is input to the coefficient inverse prediction unit, and the coefficient inverse prediction unit derives the prediction mode and adds back between the highly correlated blocks. Then, the inverse prediction data is input to the inverse quantization unit.

  Then, in the inverse quantization unit, the inversely predicted coefficient data of each component is inversely quantized by multiplication using a quantization parameter held in an internal or external storage area.

  The inversely quantized coefficient data of each component is input to the frequency inverse transform unit, where the frequency inverse transform unit includes a block of n pixels in each horizontal and vertical direction, and the block in each horizontal and vertical direction. The frequency is inversely converted into m macroblocks to obtain image data.

  Then, in the inverse quantization unit, in the iExp value generation unit, the upper p bits of the quantization parameter having a length of (p + q) bits are cut out to generate an iExp value having a p bit length, and the iMan value generation unit A qq) bit length iMan value is generated from the information indicating whether the upper p bits of the quantization parameter are all 0 and the lower q bits of the quantization parameter. , The coefficient data of each component subjected to the inverse prediction processing by the coefficient inverse prediction unit is multiplied by the iMan value generated by the iMan value generation unit, and the multiplication data multiplied by the multiplier is multiplied by the left barrel shifter. , Bit shift leftward by the numerical value indicated by the iExp value generated by the iExp value generation unit, and inverse quantization is performed.

  As described above, according to the configuration of the present invention, the coefficient data of each component subjected to the inverse prediction processing by the coefficient inverse prediction unit is generated from the information indicating whether the upper bits of the quantization parameter are all 0 and the lower bits. The iMan value is multiplied, and the multiplication data is bit-shifted leftward by the iExp value generated from the higher-order bits of the quantization parameter, so that the coefficient data of each component is converted to the iMan value as in the conventional configuration. Multiplication by a large value called a quantization scaling factor, which is data shifted bitwise by an iExp value in the left direction, is eliminated, the circuit scale of the multiplier that performs the multiplication can be reduced, and the multiplication processing speed is improved. Therefore, it is possible to optimize the image decoding process.

  As described above, according to the present invention, the quantization parameter used in the quantization unit and the inverse quantization unit in the quantization unit in the image encoding device and the inverse quantization unit in the image decoding device is the quantization parameter. An iMan value is generated from the information indicating whether the upper bits are all 0 and the lower bits, and an iExp value is generated from the upper bits of the quantization parameter. In the quantizing unit, the coefficient data of each component subjected to frequency conversion is generated. After dividing by the iMan value, the divided data is bit-shifted to the right by the iExp value and quantized, and the inverse quantization unit multiplies the coefficient data of each component subjected to inverse prediction processing by the iMan value, and then multiplies the data. Is bit-shifted to the left with the iExp value and is inversely quantized, so that the quantization unit in the image encoding device and the inverse quantization unit in the image decoding device To reduce the circuit scale, it is possible to improve the processing speed, it is possible to optimize the image encoding processing and image decoding processing.

It is the block diagram which showed schematic structure of the image coding apparatus which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the quantization part which concerns on this embodiment. It is the block diagram which showed the structure of the division part which concerns on this embodiment. It is explanatory drawing for demonstrating the relationship between the quantization parameter which concerns on this embodiment, and iMan. It is explanatory drawing for demonstrating an example of the reciprocal number of iMan concerning this embodiment. It is explanatory drawing for demonstrating the relationship between the quantization parameter which concerns on this embodiment, and iExp. It is the flowchart which showed the process sequence centering on the quantization part which concerns on this embodiment. It is the block diagram which showed schematic structure of the image decoding apparatus which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the inverse quantization part which concerns on this embodiment. It is the flowchart which showed the process sequence centering on the inverse quantization part which concerns on this embodiment. It is the block diagram which showed schematic structure of the conventional image coding apparatus. It is the block diagram which showed the structure of the conventional quantization part. It is the block diagram which showed the structure of the conventional iMan value production | generation part. It is the block diagram which showed schematic structure of the conventional image decoding apparatus. It is the block diagram which showed the structure of the conventional inverse quantization part. It is explanatory drawing for demonstrating the relationship between the conventional quantization parameter and a quantization scaling factor.

  Hereinafter, specific embodiments of an image encoding device and an image decoding device according to the present invention will be described with reference to the drawings. First, an image encoding device will be described.

[1. Image coding apparatus according to this embodiment]
FIG. 1 is a block diagram showing a configuration of an image encoding device 1 according to the present embodiment. As shown in FIG. 1, the image encoding device 1 of this embodiment replaces the quantization unit 130 in the conventional image encoding device 100 of FIG. 11 with a quantization unit 10 having a different configuration from this. Only in this respect, the configuration of the conventional image encoding device 100 is different. Therefore, the other components are denoted by the same reference numerals as those in the image encoding device 100, and detailed description thereof is omitted.

  As shown in FIG. 2, the quantization unit 10 includes a quantization parameter storage unit 11, an iMan value generation unit 12, an iExp value generation unit 13, a division unit 14, and a right barrel shifter 15, and the flow shown in FIG. The process will proceed. Hereinafter, each process of the quantization unit will be described with reference to each step of FIG.

  The quantization parameter storage unit 11 acquires and stores a quantization parameter necessary for processing from a controller or the like before starting the encoding processing (step S1).

  At the start of the encoding process, the coefficient value is counted and the macro block counter value m of the block counter (not shown) that updates the macro block, the color component, and the block counter value, the color component counter value c, The counter value n of the block is initialized to 1 (step S2). Hereinafter, the quantization process for each counter value of the macroblock, the color component, and the block updated together with the process is the same process, and therefore, the quantization process for each counter value in a state in which the process has progressed will be described.

  In the frequency conversion unit 120, the image data of the current macro block (corresponding to the counter value m) is frequency converted, and each coefficient data of a direct current (DC) component, a low frequency (LP) component, and a high frequency (HP) component. (Step S3). In this macroblock, there are 1, 3, or 4 color components. Within each color component, there is a DC component coefficient data (DC coefficient) of one pixel and a low frequency component consisting of one block of 15 pixels. Coefficient data (LP coefficient) and 15 pixels are one block, and there are high-frequency component coefficient data (HP coefficient) consisting of 16 blocks.

  The quantization parameter storage unit 11 selects a DC parameter, a low frequency component, and a high frequency component quantization parameter (each 8 bits long) corresponding to the current color component (corresponding to the counter value c) (step S4). ). The quantization process is performed in the order of DC coefficient data, LP coefficient data, and HP coefficient data.

  First, a DC component quantization parameter (8-bit length) selected by the quantization parameter storage unit 11 is input to the iMan value generation unit 12 and the iExp value generation unit 13.

  The configuration of the iMan value generation unit 12 is the same as that already described with reference to FIG. 13, but the iMan value generation unit 12 has a quantization parameter value less than 16 (the upper 4 bits are all 0). , The lower 4 bits of the quantization parameter are cut out, and data obtained by assigning 0 to the bit above the most significant bit is generated as a 5-bit iMan value, and the quantization parameter value is 16 or more (upper 4 If any one of the bits is 1), the lower 4 bits of the quantization parameter are cut out, and the value 16 is added (1 is assigned to the bit above the most significant bit). A 5-bit iMan value is generated and input to the division unit 14. The iExp value generation unit 13 cuts out the upper 4 bits of the quantization parameter, generates a 4-bit iExp value, and inputs it to the right barrel shifter 15 (step S5).

  Note that specific numerical values of iMan values corresponding to the quantization parameter (QP) are shown in FIG. When the quantization parameter is in the range of 1 to 31, the iMan value is the same as the quantization parameter. However, when the quantization parameter is 31 or later, every 16 repetitions of 32 to 47 and 48 to 63 are performed. , IMan values repeat values of 16 to 31, respectively. In FIG. 4, the notation of... Is an abbreviation of the numerical value, and indicates a numerical sequence in which the numerical value increases by one. As described above, the iMan value is in the range of 1 to 31 with respect to the quantization parameter in the range of 1 to 255, and the number range is small and limited. Further, specific numerical values of the iExp value corresponding to the quantization parameter (QP) are shown in FIG. The iExp value is obtained by cutting out the upper 4 bits of the quantization parameter, so that the quantization parameter value is 1 in the range of 1-31, 2 in the range of 32-47, and 3 in the range of 48-63. In addition, for every 16 numerical values (only the first quantization parameter is 15), the data is increased by 1 to 1-15. 6, the notation of... Is a number that is incremented by one in the quantization parameter (QP) column, and each number sequence in the range of 16 numbers such as 64-79 following 48-63. In the iExp column, numerical values increasing by 1 are shown.

  The division unit 14 divides the coefficient data input from the frequency conversion unit 120 by the iMan value input from the iMan value generation unit 12 and inputs the division data to the right barrel shifter 15. The right barrel shifter 15 bit-shifts the division data input from the division unit 14 to the right by the number of iExp values input from the iExp value generation unit 13 to generate quantization coefficient data, and It inputs into the coefficient estimation part 140 (step S6). As described above, the division unit 14 can set the divisor to 5-bit data in the range of 1 to 31 called iMan values, so that the circuit scale can be reduced and the processing speed can be improved. Although part of the division is bit-shifted by the right barrel shifter, the barrel shifter has a smaller circuit scale than the divider and can be configured to be capable of high-speed processing. it can.

  Moreover, although the said division part can also be comprised using a divider, it can be comprised using a multiplier like FIG. The division unit 14 includes an inverse table unit 16 and a multiplier 17.

  The reciprocal table unit 16 has a storage area for holding reciprocal values corresponding to integers 1 to 31, and selects a reciprocal value corresponding to data called iMan value input from the iMan value generation unit 12 and its inverse. Numerical data is input to the multiplier 17. The multiplier 17 multiplies the coefficient data input from the frequency conversion unit 120 by the reciprocal data of the iMan value input from the reciprocal table unit 16, and multiplies data (because it is multiplied by the reciprocal value. Is input to the right barrel shifter 15.

Note that specific numerical values of the reciprocal data corresponding to the iMan value are shown in FIG. FIG. 5 shows the reciprocal data corresponding to iMan values in the range of 1 to 31 in hexadecimal. The reciprocal value shown in this example is a fixed-point representation of a 16-bit number. The most significant bit is the one's digit, the other bits are the decimal part, and the decimal is one It is assumed to be between the numerical value and the decimal part. For example, when the iMan value is 1, the reciprocal is 8000 ₍₁₆₎ , and when the iMan value is 2, the reciprocal is a half value of 4000 ₍₁₆₎ . Thereafter, the iMan value is a value of 1 to iMan based on 8000 ₍₁₆₎ where the iMan value is a reciprocal of 1 with respect to the value of 3 to 31. In FIG. 5,... Indicates omission of numerical values from 5 to 27 for the iMan column and omission of reciprocal values of the numerical values from 5 to 27 for the reciprocal column. This example is an example of the reciprocal table, and the bit length that can be taken by the reciprocal value is not limited to 16 bits, and can be increased or decreased depending on the application and purpose.

  Thus, by configuring the division unit 14 using the reciprocal table 16 and the multiplier 17, the circuit scale and processing speed of the division unit 14 can be improved. Since the iMan value that is a divisor is in a small numerical range of 1 to 31 represented by a 5-bit length, the bit length of the reciprocal data can be reduced, and the number of reciprocals to be retained However, since it is sufficient to hold the reciprocal number of 31 pieces in the range of 1 to 31, the circuit scale of the reciprocal table unit 16 can be reduced.

  Next, the low-frequency component quantization parameter (8-bit length) selected by the quantization parameter storage unit 11 is input to the iMan value generation unit 12 and the iExp value generation unit 13.

  Similarly to the case of the DC component quantization parameter, the iMan value generation unit 12 generates iMan value data of a low frequency component, and the iExp value generation unit 13 generates data of the iExp value of the low frequency component. The 15 LP coefficient data in block units generated by the dividing unit 14 and input from the frequency converting unit 120 by the dividing unit 14 are divided by the iMan value data generated by the iMan value generating unit 12, and the 15 The divided data is bit-shifted rightward by the value of the iExp value data generated by the iExp generation unit 13 in the right barrel shifter 15 to generate quantized coefficient data, which is input to the coefficient prediction unit 140 (Step S8).

  Similarly to the case of the low-frequency component quantization parameter, the iMan value generation unit 12 generates high-frequency component iMan value data, and the iExp value generation unit 13 generates high-frequency component iExp value data. Is generated (step S9), and 15 HP coefficient data of the current block (corresponding to the counter value n) input from the frequency converting unit 120 by the dividing unit 14 is generated by the iMan value generating unit 12. Divided by the iMan value data, the 15 divided data are bit-shifted rightward by the value of the iExp value data generated by the iExp value generation unit 13 in the right barrel shifter 15 to generate quantized coefficient data. Then, it is input to the coefficient prediction unit 140 (step S10). Then, every time the quantization of the current block is completed, the counter value of the block is updated (step S12), and the quantization process is performed up to the final block of the current color component. The quantization processing of each frequency component is similarly performed until the final color component (step S13), and the quantization processing of each frequency component is performed similarly until the final macroblock is obtained. The process ends (step S15).

  According to the image encoding apparatus 1 of the present example having the above configuration, first, image data for one image is input to the color space conversion unit 110, and color space conversion is performed according to the pixel format of the image data, or Image data that has not undergone color space conversion and has undergone processing in the color space conversion unit 110 is input to the frequency conversion unit 120.

  The frequency conversion unit 120 performs processing for sequentially converting the data of each pixel into a coefficient for each frequency band in the raster direction in units of macroblocks (16 × 16 pixels), and the first layer data (HP coefficient) after processing, The second layer data (LP coefficient) and the third layer data (DC coefficient) are input to the quantization unit 10 in block units (4 × 4 pixels).

  The quantization unit 10 uses the first layer data, the second layer data, and the third layer data that have been coefficientized by the frequency conversion unit 120, the information about whether all the upper 4 bits of the quantization parameter are 0 and the lower 4 A process of dividing the divided data by a 5-bit iMan value generated from the bits and shifting the divided data to the right by the value of the 4-bit iExp value generated from the upper 4 bits of the quantization parameter to quantize the data. The first layer data, the second layer data, and the third layer data after processing are input to the coefficient prediction unit 140.

  The coefficient prediction unit 140 derives a prediction mode for the first layer data, the second layer data, and the third layer data, and performs a process of subtracting between the highly correlated blocks, The second layer data is input to the adaptive scan unit 150 for each block.

  The adaptive scanning unit 150 performs processing for replacing the coefficient data for each block with respect to the first layer data (HP coefficient) and the second layer data (LP coefficient). In this replacement process, after frequency conversion, non-zero coefficients are collected on the side where the pixel position in the block is small (low frequency side), and the zero value continues on the side where the pixel position in the block is large (high frequency side). This is to increase the compression rate of encoding by collecting data. The first layer data, the second layer data, and the third layer data on which the replacement process is not performed are input to the VLC encoding unit 160.

  The VLC encoding unit 160 performs entropy encoding processing on the first layer data, the second layer data, and the third layer data processed by the adaptive scan unit 150, generates a code stream, and outputs the code stream. The stream is input to the bit stream output unit 170.

  The bit stream output unit 170 sequentially concatenates the code streams encoded by the VLC encoding unit 160 to the image setting data as header information, and forms one file as a bit stream for one image data. And output the formed file to the outside.

[2. Image decoding apparatus according to this embodiment]
Next, an image decoding apparatus according to the present invention will be described. FIG. 8 is a block diagram showing the configuration of the image decoding apparatus 2 according to this embodiment. As shown in FIG. 8, the image decoding apparatus 2 of the present embodiment is replaced with the inverse quantization unit 250 in the conventional image decoding apparatus 200 of FIG. Only in this respect, the configuration of the conventional image decoding apparatus 200 is different. Accordingly, the other components are denoted by the same reference numerals as those in the image decoding apparatus 200, and detailed description thereof is omitted.

  As shown in FIG. 9, the inverse quantization unit 20 includes a quantization parameter storage unit 21, an iMan value generation unit 22, an iExp value generation unit 23, a multiplier 24, and a left barrel shifter 25, and is shown in FIG. Processing proceeds with the flow. Hereinafter, each process of the inverse quantization unit will be described with reference to the steps of FIG.

  At the start of the encoding process, the coefficient value is counted and the macro block counter value m of the block counter (not shown) that updates the macro block, the color component, and the block counter value, the color component counter value c, The counter value n of the block is initialized to 1 (step S21). Hereinafter, the quantization process for each counter value of the macroblock, the color component, and the block updated together with the process is the same process, and therefore, the quantization process for each counter value in a state in which the process has advanced will be described.

  The quantization parameter storage unit 21 acquires and stores the quantization parameter necessary for the process from the compressed data encoded after the start of the decoding process (step S22).

  The VLC decoding unit 220 decodes the encoded data of the current macroblock (corresponding to the counter value m), and generates a direct current (DC) component, a low frequency (LP) component, and a high frequency (HP) component. Each coefficient data is set (step S23). In this macroblock, there are 1, 3, or 4 color components. Within each color component, there is a DC component coefficient data (DC coefficient) of one pixel and a low frequency component consisting of one block of 15 pixels. Coefficient data (LP coefficient) and 15 pixels are one block, and there are high-frequency component coefficient data (HP coefficient) consisting of 16 blocks.

  The quantization parameter storage unit 21 selects a DC component, a low frequency component, and a high frequency component quantization parameter (each 8 bits long) corresponding to the current color component (corresponding to the counter value c) (step S24). ). The inverse quantization process is performed in the order of DC coefficient data, LP coefficient data, and HP coefficient data.

  First, the DC component quantization parameter (8-bit length) selected by the quantization parameter storage unit 21 is input to the iMan value generation unit 22 and the iExp value generation unit 23.

  The configuration of the iMan value generation unit is the same as that already described with reference to FIG. 13, but the iMan value generation unit has a quantization parameter value of less than 16 (all the upper 4 bits are all 0). The lower 4 bits of the quantization parameter are extracted, and data with 0 assigned to the upper bit of the most significant bit is generated as a 5-bit iMan value, and the quantization parameter value is 16 or more (upper 4 bits If any one of the bits is 1), the lower 4 bits of the quantization parameter are extracted, and the numerical value 16 is added (1 is assigned to the bit above the most significant bit). A bit length iMan value is generated and input to the multiplier 24. The iExp value generation unit 23 cuts out the upper 4 bits of the quantization parameter, generates a 4-bit iExp value, and inputs it to the left barrel shifter 25 (step S25).

  The multiplier 24 multiplies the coefficient data input from the coefficient inverse prediction unit 240 by the iMan value input from the iMan value generation unit 23 and inputs the multiplication data to the left barrel shifter 25. The left barrel shifter 25 bit-shifts the multiplication data input from the multiplier 24 to the left by the number of iExp values input from the iExp value generation unit 23 to generate inverse quantization coefficient data, The frequency is input to the inverse frequency converter 260 (step S26). Thus, since the multiplier 24 can set the multiplier to 5-bit length data in the range of 1 to 31 called iMan values, the circuit scale can be reduced and the processing speed can be improved. In addition, part of the multiplication is bit-shifted by the left barrel shifter, but the barrel shifter is smaller in circuit scale than the multiplier and can be configured to be capable of high-speed processing, so it is possible to improve processing performance. it can.

  Next, the low-frequency component quantization parameter (8-bit length) selected by the quantization parameter storage unit 21 is input to the iMan value generation unit 22 and the iExp value generation unit 23.

  Similar to the case of the DC component quantization parameter, the iMan value generation unit 22 generates low frequency component iMan value data, and the iExp value generation unit 23 generates low frequency component iExp value data. The 15 LP coefficient data in block units input from the coefficient inverse prediction unit 240 is generated by the multiplier 24 and multiplied by the iMan value data generated by the iMan value generation unit 22, The 15 multiplication data are bit-shifted leftward by the value of the iExp value data generated by the iExp generation unit 23 in the left barrel shifter 25 to generate inverse quantization coefficient data, and the frequency inverse conversion unit 260 Input (step S28).

  Similar to the case of the low-frequency component quantization parameter, the iMan value generation unit 22 generates high-frequency component iMan value data, and the iExp generation unit 23 generates high-frequency component iExp value data. The iMan value generation unit 22 generates 15 HP coefficient data of the current block (corresponding to the counter value n) input from the coefficient inverse prediction unit 240 by the multiplier 24 (step S29). The iMan value data is multiplied, and the 15 multiplication data are bit-shifted leftward by the value of the iExp value data generated by the iExp generation unit 23 in the left barrel shifter 25 to generate inverse quantization coefficient data. Then, it is input to the frequency inverse transform unit 260 (step S30). Then, every time the quantization of the current block is completed, the block counter value is updated (step S32), and the quantization process is performed up to the final block of the current color component. The quantization processing of each frequency component is similarly performed until the final color component is obtained (step S33), and the quantization processing of each frequency component is performed similarly until the final macroblock is obtained. The process ends (step S35).

  According to the image decoding apparatus 2 of the present example having the above-described configuration, first, compressed data for one image is input to the bitstream analysis unit 210, image information is extracted, and separated encoding is performed. Data is input to the VLC decoding unit 220.

  The VLC decoding unit 220 decodes the encoded data separated by the bitstream analysis unit 210, and performs first layer data (HP coefficient), second layer data (LP coefficient), and third layer data (DC coefficient). Are generated and input to the adaptive scanning unit 230.

  The adaptive scanning unit 230 performs processing for replacing the coefficient data for each block with respect to the first layer data (HP coefficient) and the second layer data (LP coefficient). This replacement process collects non-zero coefficients on the side where the pixel position in the block is small (low frequency side) during frequency conversion in compression, and the zero value is on the side where the pixel position in the block is large (high frequency side). This is performed in order to increase the compression rate of encoding by collecting continuous coefficient data. The first layer data, the second layer data, and the third layer data on which the replacement process is not performed are input to the coefficient inverse prediction unit 240.

  The coefficient inverse prediction unit 240 derives a prediction mode for the first layer data, the second layer data, and the third layer data, and performs a process of adding back between the highly correlated blocks. , Second layer data, and third layer data are input to the inverse quantization unit 20.

  The inverse quantization unit 20 is information on whether or not all the upper 4 bits of the quantization parameter are 0 for the first layer data, the second layer data, and the third layer data subjected to the inverse prediction processing by the coefficient inverse prediction unit 240 And the 5-bit iMan value generated from the lower 4 bits, and the multiplication data is bit-shifted to the left by the value of the 4-bit iExp value generated from the upper 4 bits of the quantization parameter. The first layer data, the second layer data, and the third layer data after processing are input to the frequency inverse transform unit 260.

  The frequency inverse transform unit 260 performs frequency inverse transform on the first layer data, the second layer data, and the third layer data inversely quantized by the inverse quantization unit 20, and performs macroblock unit (16 × 16 pixels). The converted image data is input to the color space inverse conversion unit 270 as image data.

  The color space inverse transform unit 270 performs the color space inverse transform on the image data in units of macro blocks frequency-transformed by the frequency inverse transform unit 260 according to the pixel format that is the image information extracted by the bit stream analysis unit 210, Alternatively, the processed image data is output to the outside as image data for one image without performing inverse color space conversion.

  As described above in detail, according to the image encoding device 1 of the present embodiment, in the quantization unit 10, the iMan value generated from the information indicating whether or not all the upper bits of the quantization parameter are 0 and the lower bits. Then, the coefficient data input from the frequency conversion unit is divided by the division unit, and the division data is bit-shifted to the right by the value of the iExp value generated from the upper bits of the quantization parameter to obtain the quantized coefficient data. Compared to the conventional configuration in which the coefficient data is divided by the quantization scaling factor, which is data obtained by bit shifting the iMan value to the left by the value of the iExp value, the divisor value given to the division unit is reduced. Therefore, it is possible to reduce the circuit scale of the division unit and improve the processing speed of the division unit.

  Then, the division unit in the quantization unit of the image encoding device 1 according to the present embodiment stores information on whether all the upper bits of the quantization parameter are 0 and the reciprocal value of the iMan value generated from the lower bits. Each time an iMan value is generated, an inverse number of the iMan value is selected from the inverse number table unit, and the coefficient data input from the frequency conversion unit by the inverse of the iMan value is input to the multiplier. Therefore, the multiplier multiplies the coefficient data input from the frequency conversion unit by a reciprocal of the quantization scaling factor which is data obtained by bit shifting the iMan value in the left direction by the value of the iExp value to the left. Compared to the configuration, the bit width of the reciprocal data to be retained can be reduced, the number of reciprocal data to be retained is reduced, the reciprocal table unit and multiplication Can reduce the circuit scale, it is possible to improve the processing speed of the divider. Thereby, the optimization of the image coding process of an image coding apparatus can be aimed at.

  Further, according to the image decoding device 2 of the present embodiment, the inverse quantization unit 20 performs frequency conversion based on information indicating whether all the upper bits of the quantization parameter are 0 and the iMan value generated from the lower bits. The coefficient data input from the unit is multiplied by a multiplier, and the multiplication data is bit-shifted to the left by the value of the iExp value generated from the upper bits of the quantization parameter to obtain inverse quantization coefficient data. Therefore, the multiplier value given to the multiplier can be reduced as compared with the configuration in which the coefficient data is multiplied by the quantization scaling factor which is data obtained by bit shifting the iMan value to the left by the value of the iExp value, It is possible to reduce the circuit scale of the multiplier and improve the processing speed of the multiplier. Thereby, optimization of the image decoding process of an image decoding apparatus can be aimed at.

  As mentioned above, although one Embodiment of this invention was described, the specific aspect which this invention can take is not limited to this at all. For example, in the above example, the quantization parameter is 8 bits long, and an iMan value having a length of 5 bits is generated from the information indicating whether the upper 4 bits of the quantization parameter are all 0 and the lower 4 bits. Although an iExp value having a length of 4 bits is generated from the upper 4 bits, the present invention is not limited to this. The quantization parameter is set to (p + q) bits, and whether the upper p bits of the quantization parameter are all 0 or not. An iMan having a length of (q + 1) bits may be generated from the information and the lower q bits, and an iExp value having a p bit length may be generated from the upper p bits of the quantization parameter.

  In the present embodiment, the image encoding device and the image decoding device have different configurations. However, the image encoding / decoding device may have a single configuration. In that case, the quantization unit in the image encoding device and the inverse quantization unit in the image decoding device may have different configurations, or may be configured as a single quantization / inverse quantization unit. When the quantization / inverse quantization unit has one configuration, the quantization parameter storage unit, iMan value generation unit, and iExp value generation unit have a common configuration, and a division unit, a multiplier, a right barrel shifter, and a left The direction barrel shifter is configured to perform quantization using a division unit and a right direction barrel shifter for encoding, and to perform inverse quantization using a multiplier and a left direction barrel shifter for decoding. Also good.

DESCRIPTION OF SYMBOLS 1 Image coding apparatus 2 Image decoding apparatus 10 Quantization part 11 Quantization parameter memory | storage part 12 iMan value generation part 13 iExp value generation part 14 Division part 15 Right direction barrel shifter 16 Reciprocal number table part 17 Multiplier 20 Inverse quantization part 21 Quantization parameter storage unit 22 iMan value generation unit 23 iExp value generation unit 24 Multiplier 25 Left barrel shifter 100 Image encoding device 110 Color space conversion unit 111 Quantization parameter storage unit 112 iMan value generation unit 113 iExp value generation unit 114 Left Direction barrel shifter 115 Division unit 116 Bit cutout unit 117 Adder 118 Comparison unit 119 Selector 120 Frequency conversion unit 130 Quantization unit 140 Coefficient prediction unit 150 Adaptive scan unit 160 VLC encoding unit 170 Bit stream output unit 200 Image decoding device 210 Bitstream Analysis Unit 211 Quantization Parameter Storage Unit 212 iMan Value Generation Unit 213 iExp Value Generation Unit 214 Left Barrel Shifter 215 Multiplier 220 VLC Decoding Unit 230 Adaptive Scan Unit 240 Coefficient Inverse Prediction Unit 250 Inverse Quantization Unit 260 Frequency Inverse transform unit 270 Color space inverse transform unit

Claims (3)

The image data is frequency-converted for a macroblock consisting of m blocks in each horizontal and vertical direction in units of a block consisting of n pixels in each horizontal and vertical direction, and the upper left pixel position of each block in the macroblock The data corresponding to the pixel position other than the above is used as the coefficient data of the high frequency component, the data corresponding to the upper left pixel position of each block in the macro block is collected, further frequency-converted, and the upper left pixel position after the frequency conversion. The frequency conversion unit which makes the corresponding data the coefficient data of the direct current component, and the data corresponding to the pixel position other than the upper left the coefficient data of the low frequency component,
Quantization of the coefficient data of each component frequency-converted by the frequency conversion unit by using a quantization parameter having a (p + q) bit length held in an internal or external storage area to obtain quantized coefficient data And
A coefficient scanning unit that performs replacement processing of pixels in the block of quantized coefficient data of the low frequency component and high frequency component quantized by the quantizing unit to obtain replacement coefficient data;
An encoding unit that generates encoded data by encoding the quantized coefficient data of the DC component after quantization by the quantizing unit and the low-frequency component and high-frequency component replacement coefficient data after the replacement processing by the coefficient scanning unit In an image encoding device comprising:
The quantization unit is
An iExp value generation unit for generating an iExp value having a p-bit length, which is data obtained by cutting out the upper p bits of the quantization parameter;
When all the upper p bits of the quantization parameter are 0, the lower q bits of the quantization parameter are cut out, and the value 0 is assigned to the bit above the most significant bit (Q + 1) bit length iMan value is generated, and when any one of the upper p bits of the quantization parameter is 1, the lower q bits of the quantization parameter are cut out and the An iMan value generation unit that generates an iMan value having a length of (q + 1) bits, which is data obtained by assigning a value of 1 to a bit above the upper bit,
A division unit that divides the coefficient data of the direct current component, the low frequency component, and the high frequency component that are frequency-converted by the frequency conversion unit by the iMan value to obtain divided data;
A right barrel shifter for generating quantized coefficient data by bit-shifting the division data divided by the division unit to the right by the value represented by the iExp value generated by the iExp value generation unit; An image encoding device characterized by that.
However, n and m are integers of 2 or more, and p and q are integers of 1 or more. The division unit is
A reciprocal table unit that has a storage area that holds a reciprocal value corresponding to each iMan value, and selects a reciprocal value corresponding to the iMan value generated by the iMan value generation unit;
A multiplier that multiplies the coefficient data of the DC component, the low frequency component, and the high frequency component frequency-converted by the frequency conversion unit by the reciprocal value of the iMan value selected by the reciprocal table unit. The image encoding device according to claim 1. Decodes compressed image data in which coefficient data consisting of DC components, low-frequency components, and high-frequency components in the image data after frequency conversion is encoded, and consists of DC components, low-frequency components, and high-frequency components A decryption unit for generating decrypted data;
A coefficient scanning unit that performs replacement processing of pixels in the block of the decoded data of the low-frequency component and the high-frequency component decoded by the decoding unit to obtain replacement coefficient data;
The DC component decoded data decoded by the decoding unit, and the low frequency component and high frequency component coefficient data in which the arrangement order of pixels in the block is changed by the coefficient scanning unit An inverse quantization unit that performs inverse quantization using an 8-bit length quantization parameter held in a storage area to obtain inverse quantization coefficient data;
An image area restored by inverse frequency transform of the inverse quantization coefficient data is composed of blocks of n pixels in each of the horizontal and vertical directions, and from a macro block having m blocks in each of the horizontal and vertical directions. As the image area, the inverse quantization coefficient data of the direct current component is arranged at the pixel position at the upper left of the m area in each of the horizontal and vertical directions, and the inverse quantization of the low frequency component is performed at a pixel position other than the upper left of the area. The coefficient data of each pixel position in the region subjected to frequency inverse transform for the region is disposed, and the coefficient data of each pixel position in the region subjected to the frequency inverse transform is converted into each block in the macroblock corresponding to the position corresponding to the pixel position in the region, It is arranged at the pixel position at the upper left, and the inverse quantization coefficient data of the high frequency component is arranged at a pixel position other than the pixel position at the upper left of each block in the macro block. Every further by frequency inverse conversion, the image decoding apparatus and a frequency inverse conversion unit for the frequency inversion data,
The inverse quantization unit includes:
An iExp value generation unit for generating an iExp value having a p-bit length, which is data obtained by cutting out the upper p bits of the quantization parameter;
When all the upper p bits of the quantization parameter are 0, the lower q bits of the quantization parameter are cut out, and the value 0 is assigned to the bit above the most significant bit (Q + 1) bit length iMan value is generated, and when any one of the upper p bits of the quantization parameter is 1, the lower q bits of the quantization parameter are cut out and the An iMan value generation unit that generates an iMan value having a length of (q + 1) bits, which is data obtained by assigning a value of 1 to a bit above the upper bit,
The iMan value is multiplied by the decoded data of the DC component decoded by the decoding unit and the coefficient data of the low frequency component and the high frequency component in which the arrangement order of the pixels in the block is changed by the coefficient scanning unit. A multiplier for multiplication data,
A left barrel shifter configured to bit-shift the multiplication data multiplied by the multiplier leftward by a value represented by the iExp value generated by the iExp value generation unit to generate inverse quantization coefficient data; The image decoding apparatus characterized by the above-mentioned.
However, n and m are integers of 2 or more, and p and q are integers of 1 or more.

Images