JPH03262382A - Method and apparatus for picture data coding and decoding - Google Patents

Abstract

PURPOSE:To attain high speed quantization and inverse quantization processing by obtaining an effective order of non-null coefficient as to each block and giving zero to a coefficient in excess of the effective order. CONSTITUTION:A proper effective order is estimated in most blocks by obtaining an effective order as OR processing of a highest order of non-null coefficient of 1st row (and/or column) and an 8th row (and/or column). When the 1st and 8th columns and rows of the DCT coefficient are scanned, the highest order of row scanning is 4th and the highest order of the column scanning is 5th, the effective order is 4th for the row and 5th for the column and the quantization calculation of the DCT coefficient after the effective order shown in hatched lines is omitted and zero is set as the quantization DCT coefficient. That is, the part shown in hatched lines is masked zero. Then the quantization calculation of the part is skipped and high speed processing is attained.

Description

[Detailed Description of the Invention] [Summary] A field code that is obtained by quantizing DCT coefficients obtained by performing two-dimensional discrete cosine transform on the pixels in the blocks obtained by dividing a multivalued image into blocks each consisting of a plurality of pixels. Regarding an image data encoding and restoring method and device for converting the encoded data into inverse transforms and restoring the image, the purpose of this is to speed up the encoding and restoring process. The area where the non-zero coefficients are distributed is estimated by calculating the effective order from the logical sum of the highest order among the non-zero coefficients, and the coefficients after the effective order are all zero during quantization and dequantization. By doing so, the calculation processing of quantization and inverse quantization is omitted to increase the speed.

[Industrial Application Fields] The present invention is a method for dividing a multivalued image into blocks each consisting of a plurality of pixels, and then performing a two-dimensional discrete cosine transform on the pixels in each block, and then quantizing the DCT coefficients. encode,
The present invention further relates to an image data encoding and restoring method and apparatus for inversely transforming encoded data to restore an image.

For example, the adaptive discrete cosine transform coding method (A
adaptive Discrete CCo51neT
(hereinafter abbreviated as "rADCT method").

The ADCT method divides an image into blocks consisting of, for example, 8 x 8 pixels, and converts the image signal of each block into coefficients of a spatial frequency distribution using two-dimensional discrete cosine transformation (hereinafter referred to as rDcTJ), which is adapted to visual perception. The Huffman coefficient was quantized using a threshold value, and the obtained quantized DCT coefficients were statistically obtained.
It is encoded using a table.

The encoding operation will be explained according to the block diagram of the ADCT method shown in FIG.

First, an image is divided into blocks each consisting of 8×8 pixels as shown in FIG.

The DCT transform unit 12 orthogonally transforms the input image signal by DCT to obtain a DC signal with a spatial frequency distribution shown in FIG.
It is converted into T coefficients and output to the linear quantization section 14.

Specifically, as shown in FIG.
The block is CT-transformed, the rows and columns of the coefficients within the block are transposed (transposed) by the transposition unit 114, and the result is output to the one-dimensional DCT transformation unit 116. In the one-dimensional DCT conversion unit 1165, one-dimensional D
It is subjected to one-dimensional DCT transformation in the same manner as the CT transformation unit 112 and output to the transposition unit +18. In the transposing section 11g, transposing section +14
Performs the same transposition process and outputs to terminal +2[1.

In this way, for all blocks of image data, the two-dimensional D
It is converted into DCT coefficients by performing CT conversion processing.

The linear quantization unit 14 linearly quantizes the input DCT coefficients using a quantization matrix 16 configured with threshold values shown in FIG.
Find the coefficient. FIG. 20 shows the quantized DCT coefficients obtained from the DCT coefficients in FIG. 18.

As shown in FIG. 20, DCT coefficients with values smaller than the threshold value become 0, and quantized DCT coefficients having values only for DCFfc and a small AC component are generated.

The two-dimensionally arranged quantized DCT coefficients are converted into one-dimensional data according to a scanning order called zigzag scan shown in FIG. 21, and are input to the variable length encoding unit 18. The variable length encoding unit 18 performs variable length encoding on the difference between the DC component at the beginning of each block and the DC component of the previous block. For the AC component, the values of non-zero coefficients (effective coefficients whose value is not zero) (hereinafter referred to as "index") and the length of the run of coefficients whose value is zero up to the index (hereinafter referred to as "run")
is variable-length coded for each block. Further, each of the DC and AC components is encoded using a code table 20 consisting of a Huffman table created based on statistics for each image, and the obtained code data is sequentially output from a terminal 22.

FIG. 16 shows a block diagram of a restoration circuit using the ADCT method, and encoded data is restored to an image by the following method.

First, code data input from the terminal 54 is input to the variable length code section 56. The variable length decoding unit 56 decodes the input code data into fixed length data of indexes and runs using a decoding table 58 that is an inverse table to the Huffman table of the code table 20 in FIG. conversion part 6
Output to 0.

The inverse quantization section 60 dequantizes the input quantized DCT coefficients by multiplying them by each of the quantization matrices 62 to restore the DCT coefficients, and the two-dimensional inverse DCT transformation section 64
Output to. The two-dimensional inverse DCT transform unit 64 orthogonally transforms the input DCT coefficients by inverse DCT transform, and transforms the coefficients of the spatial frequency distribution into an image signal.

Specifically, as shown in FIG. 23, the DCT coefficients input from the terminal 122 are converted into 1
The dimensional inverse DCT transform is performed and output to the transposing unit 126. The transposing unit 126 transposes the rows and columns of the coefficients within one block and outputs the result to the one-dimensional inverse DCT transform unit 128. one-dimensional inverted D
The CT conversion unit +28 converts the input transposed coefficients to 1 again.
The dimensional inverse DCT transform is performed and output to the transposing unit 130. Similar to the transposing unit 126, the transposing unit +30 outputs from the terminal 132 a signal obtained by transposing the rows and columns of the coefficients within one block again, thereby restoring the image.

[Conventional technology] In the conventional ADCT method, quantization DCT at the time of encoding
The coefficients are determined by a quantization operation that divides the DCT coefficients by a quantization threshold. FIG. 24 shows a block diagram of a conventional linear quantization circuit.

In FIG. 24, the OCT coefficients inputted manually from the terminal 26 are stored on the DCT coefficient holding section.

In accordance with the data read signal RED from the DCT coefficient coefficient holding section 8 timing control section 40, the input DCT coefficients are sequentially output pixel by pixel to the division section 36 as a quantization section. Further, the quantization threshold holding unit 32 similarly follows the data succession signal RED from the timing control unit 40.
The quantization threshold corresponding to each pixel held is sequentially output to the division unit 36. The division unit 36 divides D of each input pixel.
The CT coefficient is quantized by dividing it by a quantization threshold, and the result is output to the latch unit 44 as a quantized DCT coefficient (QUD). The timing control unit 40 calculates the access time of the division unit 36 and supplies the data latch signal L to the latch unit 44.
Generate AT. By this latch signal LAT, the quantized OCT coefficient is latched in the latch section 44 and output to the variable length encoding section 18.

When the quantization of the coefficients for one pixel is completed, the timing control unit 40 controls the DCT coefficient coefficient holding unit upper octetization threshold holding unit 3.
2 to read out the DCT coefficient and quantization threshold of the next pixel, and performs a quantization operation on the coefficient of the next pixel. In this way, the DCT coefficients held in the DCT coefficient coefficient holding section are read out in units of one pixel, divided by the quantization threshold held in the quantization threshold holding section 32, and the result is used to quantize the target pixel. By repeating the process of outputting DCT coefficients for each pixel and block for one screen, the DCT coefficient for one screen is
The coefficients are quantized.

On the other hand, the DCT coefficients at the time of restoration are obtained by inverse quantization in which the quantized DCT coefficients decoded from the encoded data are multiplied by a quantization threshold. FIG. 25 shows a block diagram of a conventional linear inverse quantization circuit.

In FIG. 25, the quantized DCT coefficients decoded from code data by the variable length decoding section 56 are inputted to the quantized coefficient holding section 70 from a terminal 75. The quantized coefficient holding section 70 sequentially outputs the input and held quantized DCT coefficients one by one to the multiplication section 78 as an inverse quantization section in accordance with the data read signal RED from the timing control section 80. Similarly, the quantization threshold holding unit 74 sequentially outputs the quantization threshold corresponding to each held DCT coefficient to the multiplication unit 78. The multiplier 78 performs an inverse quantization operation to obtain a DCT coefficient by multiplying the input quantized DCT coefficient of each pixel by a quantization threshold. The timing control section 80 calculates the access time of the multiplication section 78 and supplies the data latch signal L to the latch section 86.
Generate AT.

This latch signal LAT causes the latch section 86 to latch the DCT coefficients, and output them from the terminal 85.

When the inverse quantization of one quantization coefficient is completed, the timing control section 80 instructs the quantization coefficient holding section 70 and the quantization threshold holding section 74 to read out the next quantized DCT coefficient and quantization threshold. , performs an inverse quantization operation on the next quantized DCT coefficient. In this way, the process of reading out the quantized DCT coefficients held in the quantization coefficient holding unit 70 at the first side vehicle position, inversely quantizing them using the quantization threshold held in the quantization threshold holding unit 2, and outputting them is blocked. By repeating one screen per unit, the quantized DCT coefficients for one screen are dequantized. The DCT coefficients obtained by inverse quantization are subjected to two-dimensional inverse DCT transformation as shown in FIGS. 22 and 23, and thereby restored to image data.

[Problems to be solved by the invention] As described above, in the conventional image data encoding and restoration method using the ADCT method, when DCT coefficients are quantized during encoding, the DCT coefficients of all pixels are There was a problem that it took a long time to process because it was divided by the quantization threshold. In other words, the access speed of a divider that performs quantization is generally slow;
Furthermore, even if the divider is configured with memory such as ROM to increase speed, the memory capacity is large because the number of signal bits required for division is large, and the quantization speed is determined by the memory access speed. However, there was a problem in that it was difficult to increase the speed.

Similarly, in inverse quantization during restoration, all quantization D
Since the CT coefficients were multiplied by the quantization threshold, the processing took time and was difficult to speed up.

The present invention has been made in view of these conventional problems, and it is an object of the present invention to provide an image data encoding and restoring method and apparatus that can speed up encoding and restoring processing without increasing the circuit scale. purpose.

[Means to solve the problem] FIG. 1A is a diagram explaining the principle of the present invention.

First, the present invention divides an original image into a plurality of blocks each consisting of a plurality of pixels, and for each block, the gradation values of a plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. The present invention is directed to an encoding and restoring method and apparatus for quantizing DCT coefficients, encoding the quantized DC coefficients, and restoring an image by inversely transforming the encoded data.

Regarding such an encoding and decompression method and apparatus, the apparatus according to the present invention is configured as follows, for example.

First, in encoding, as shown in Figure 1A (a),
D to input and temporarily hold the DCT coefficients after DCT transformation
CT coefficient holding means 200 and DCT coefficient holding means 200
n of the blocks of DCT coefficients held in (n is 1
A non-zero coefficient detection means 300 that scans the rows (and/or columns) of the positive integer above and detects the highest order of non-zero coefficients; Effective order detection means 400 that calculates an effective order from a logical sum, quantization address holding means 500 that holds sequential addresses for quantizing DCT coefficients, and effective coefficient detection means 400
a zero selection means 600 that compares the effective order address of the quantization address with the address of the quantization address holding means and selects zero when the quantization address is equal to or higher than the effective order; quantization is performed by comparing the values and selecting zero for the quantized DCT coefficients exceeding the effective order, and speeding up is achieved by omitting the quantization operation (division) for the DCT coefficients that become zero.

On the other hand, in restoration, as shown in Figure 1A (b),
Quantization threshold holding means 210 temporarily holds the quantized DCT coefficients decoded from encoded data, and n (n is 1) of the quantized DCT coefficient blocks held in the quantization threshold holding means 210 The highest order detection means 310 scans the rows (and/or columns) of positive integers) and detects the highest order among the non-zero coefficients, and the highest order detection means 3111 scans the rows (and/or columns) of the non-zero coefficients. It includes an effective order calculation means 410 that calculates an effective order from a logical sum, and a zero coefficient selection means 610 that selects zero for DCT coefficients of orders exceeding the effective order obtained by the effective order calculation means 410, and performs inverse quantization processing. By selecting zero for DCT coefficients that sometimes exceed the effective order, dequantization operations (multiplications) are omitted to increase speed.

Further, the effective order obtained during encoding is added to the quantized DCT coefficient as effective order information and encoded, and during restoration, the effective order is detected from the effective upper order information separated from the encoded data and processed in the same manner. The configuration is such that detection of effective coefficients by scanning quantized DCT coefficients at the time is omitted.

[Operations] According to the image data encoding and restoration method and apparatus of the present invention having such a configuration, the following effects can be obtained.

First, in the present invention, based on the fact that DCT coefficients tend to be concentrated in low orders, the DCT coefficients in a block are scanned over several rows (and/or columns), and the highest order logic The area where non-zero coefficients are distributed is estimated by the effective order obtained from the sum, and the results are set to zero without performing quantization and inverse quantization for the area after the effective order, and this quantization, By skipping the dequantization operation, it is possible to speed up the encoding and restoration processing.

For example, in a block of 8×8 DCT coefficients in which the gradation changes gradually, non-zero coefficients are concentrated in DC and a few AC around it. In addition, in areas with sharp gradation changes such as edge areas, non-zero coefficients are concentrated in high-order AC areas, so as shown in Figure 1B, the 1st row (and/or column) and the 8th By finding the effective order as the logical sum of the highest orders of the non-zero coefficients of the rows (and/or columns), it is possible to appropriately estimate the effective order in most blocks.

Of course, the effective order can also be reliably determined by searching all rows (and/or columns).

When scanning the first and eighth columns and rows of DCT coefficients in Figure 1B, the highest order of row scanning is 4th order, and the highest order of column scanning is 5th order, so the effective order is The quantization operation for the DCT coefficients in the area after the effective order shown by the shaded area is omitted, and all quantized DCT coefficients are set to zero.

That is, the shaded portions in the figure are masked to zero.

This region is 46.7% (30/64) of the total, and the quantization operation for this portion can be skipped to speed up the processing.

[Embodiment] FIG. 2 is a block diagram showing an embodiment of the image data encoding device of the present invention.

In FIG. 2, for example, an image signal divided into 8×8 blocks as shown in FIG. D.C.
The T-transformer 12 orthogonally transforms the manually input image signal using DCT to convert it into DCT coefficients having a spatial frequency distribution shown in FIG. 6, and outputs the DCT coefficients to the linear quantizer 14.

The linear quantization unit 14 linearly quantizes (divides) the input DCT coefficients using a quantization matrix 16 configured with the threshold values shown in FIG. 19 determined through visual experiments to obtain quantized DCT coefficients.

The quantized DCT coefficients arranged two-dimensionally by the quantization operation of the linear quantizer 14 are converted into one-dimensional one according to the scanning order called zigzag scan shown in FIG.
The signal is input to the variable length encoder 18. Variable length encoding unit 18
variable-length encodes the difference between the DC component at the beginning of each block and the DC component of the previous block. As for the AC component, the index and run are variable-length coded for each block. Further, each of the DC and AC components is encoded using a code table 20 consisting of a Huffman table created based on statistics for each image, and the obtained code data is sequentially output from a terminal 22.

Such a configuration is the same as the conventional device, but in the present invention, an effective order detection section 24 is newly provided. This allows the conversion operation to be omitted.

FIG. 3 is a block diagram of an embodiment of the effective order detection section 24 of the present invention shown in FIG. 2 in the DCT coefficient coefficient holding section 8 of the linear quantization section 14. The DCT coefficient input from the terminal 26 is first held in the DCT coefficient holding section 8.

In this embodiment, the effective order detection section 24 scans the first and eighth rows of the DCT coefficients in one block held by the DCT coefficient coefficient holding section 8 to find the non-zero coefficient, that is, the highest non-zero coefficient. The order is detected and the effective order is determined from the logical sum of each highest order. Therefore, the effective order detection section 24 has 1
th highest degree search unit 46 and eighth highest degree detection unit 48
is provided.

For example, the DC of FIG. 6 has 8 DCT coefficient coefficient holding sections.
Assuming that the highest order is detected by scanning rows and rows of T coefficients, the first and eighth rows are first scanned to obtain the following results.

1st row highest degree = 4th order 8th line highest degree - 〇 degree Next, the first and eighth columns are scanned to obtain the following results.

1st column highest degree = 5th column 8th column highest degree = 0th degree These results are given to the logical sum calculating section 50, and the effective degree is determined as the logical sum of the first and eighth highest degrees. In this case, the effective order of rows is 4th and the effective order of columns is 5th. Each effective order of the row and column determined by the logical sum calculation section 6 is outputted to the effective address determination section 52. The effective address determination section 52 sets an effective address for determining the effective order, and outputs it to the linear quantization section 14 for a comparison operation in synchronization with the quantization address during the quantization operation.

FIG. 4 is a block diagram of an embodiment showing the linear quantization section 14 in FIG. 2 together with the effective order detection section 24.

In FIG. 4, prior to the quantization of one block of DCT coefficients, the effective order detection unit 24 performs a DC
T coefficient coefficient holding section 8 scans the input DCT coefficients and determines the effective order (effective address). The quantization address generation section 10 outputs the quantization address to the DCT coefficient coefficient holding section 8 comparator 32 in accordance with the data read signal RED from the timing control section 40 . At the same time, effective order detection unit 8
is the data read signal R from the timing control section 40
The effective address determined according to ED is output to comparator 32. The comparator 32 compares the input quantized address and the effective address, and outputs a comparison result CMP to the timing control unit 40 to determine whether both lower addresses exceed the effective address. Timing control unit 40 according to comparison result CMP
outputs a selection signal SEL to the multiplexer 38. Here, if a comparison result CMP in which the quantization address exceeds the effective address is obtained, the selection signal SEL is output, the multiplexer 38 switches to the output of the zero generator 38, and selects the zero signal of the zero generator 42. and outputs it to the latch section 44.

On the other hand, when a comparison result CMP is obtained in which the quantization address is within the effective address, the DCT coefficient read out according to the quantization address of the DCT coefficient coefficient holding section 8 is input to the division section 36 that performs the quantization operation, and at the same time A quantization threshold is input from the quantization threshold holding unit 32, and a quantization operation is performed. This calculation result is output as a quantized DCT coefficient QUD to the multiplexer 38, and the multiplexer 38 outputs the quantized DCT coefficient QUD to the latch section 44 in accordance with the selection signal SEL from the timing control section 40. Then, the timing control section 40 outputs a data latch signal LAT to the latch section 44, the quantized DCT coefficient QUD is latched by the latch section 44, and is output to the variable length encoding section 18 at the next stage.

When the quantization of the coefficients for one pixel is completed, the timing control section 40 controls the DCT coefficient coefficient holding section upper octetization threshold holding section 3.
2 to read out the DCT coefficient and quantization threshold of the next pixel, and performs a quantization operation on the coefficient of the next pixel.

In this way, by repeating the quantization process for each pixel and each block for one screen, the DCT coefficients of the (1) screen needle are quantized.

In the case of quantization of DCT coefficients in FIG.
Since the quantization operation according to the method is not necessary, and the zero region reaches as much as 46.7%, it can greatly contribute to speeding up the quantization operation.

FIG. 7 is a block diagram of an embodiment of an image data restoration/decoding apparatus of the present invention corresponding to the image data encoding apparatus of FIG. 2.

In FIG. 7, coded data input from the terminal 54 is manually inputted to the variable length coder 56.

In the variable length decoding unit 56, the input code data is decoded into fixed length data of index and run using a decoding table 58 which is an inverse table to the human table in the code table 20 of FIG. It is output to the inverse quantization section 60. The linear dequantization unit 60 dequantizes the quantized DCT coefficients by multiplying each of the quantized DCT coefficients by each of the threshold values of the quantization matrix 62 to restore the DCT coefficients, and
It is output to the dimensional inverse DCT transform unit 64. The two-dimensional inverse DCT transform unit 64 orthogonally transforms the input DCT coefficients by inverse DCT transform, and outputs the coefficients of the spatial frequency distribution from a terminal 66 as an image signal.

Although such a configuration is the same as the conventional one, in addition to this, in the present invention, an effective order detection section 68 is newly provided.

This effective order detecting section 68 is the effective order detecting section 24 of FIG.
However, unlike the DCT coefficient when the scanning target for detecting the effective order is encoding, it is a quantized DCT coefficient decoded from encoded data.

FIG. 8 is a block diagram showing an embodiment of the linear inverse quantization section 60 in the image data restoration apparatus of FIG. 7 together with the effective order detection section 68.

In FIG. 8, first, prior to dequantization, the quantized D input from the terminal 75 and held in the quantization coefficient holding unit 70
The CT coefficients are scanned by the effective order detection unit 68 to obtain the effective order and determine the effective address.

Subsequently, the inverse quantization address generated by the inverse quantization address generation section 72 is given to the quantization coefficient holding section 70, the comparator 76, and the quantization threshold holding section 74 in response to the readout signal RED from the timing control section 80. At the same time, the effective order detection unit 6
8 is supplied with a read signal RED from the timing control section 80, and the already determined effective address is set in the comparator 76. A comparator 76 compares the effective address from the effective order detector 68 with the dequantized address generated by the dequantized address generator 72. Comparison result C of comparator 76
According to MP, the timing control unit 80 controls the multiplexer 8
A selection signal SEL is output to the terminal 2. If the dequantized address exceeds the effective address, multiplexer 82 selects the zero signal generated by zero generator 84 and outputs it to latch section 86 .

On the other hand, if the dequantization address is within the effective address, the multiplier 78 serving as the dequantization section performs dequantization operation to obtain the DCT coefficient by multiplying the quantized DCT coefficient by the quantization threshold, and the multiplexer 82 is D as the calculation result
The CT coefficient is output to the latch section 86. Then, the timing control section 80 sends the data latch signal LA to the latch section 86.
T is output to latch the DCT coefficients and output to the next stage two-dimensional inverse DCT transform section 64. Quantization DCT for one pixel
When the inverse quantization of the coefficients is completed, the timing-based side unit 80 instructs the quantization coefficient holding unit 70 and the quantization threshold holding unit 74 to read out the quantized DCT coefficient and quantization threshold of the next pixel, and Perform inverse quantization of the quantized DCT coefficients of pixels.

By repeating the dequantization process pixel by block for one screen in this way, the quantized DCT for one screen is
The coefficients are dequantized.

FIG. 9 shows an embodiment of an image data encoding device for adding effective order information to encoded code data, and shows an embodiment of an image data restoring device corresponding to this image data encoding device. It is shown in FIG.

In FIG. 9, the configuration of the linear quantization unit 14 is the same as the embodiment shown in FIG. be. Further, this embodiment differs from the embodiment shown in FIG. 2 in that, on the variable length encoding section 18 side, a quantization coefficient holding section 90 is provided at the input stage, and a multiplexer 92 is provided at the output stage.

In the process of adding effective order information to code data, first, the effective order information detected by the effective order detector 20 is held in the effective order holding section 88, and the effective order information is output to the multiplexer 92 for each block. When the multiplexer 92 outputs code data from the variable length encoder 18 for one pixel, it switches to the effective order holding section 88 side and repeats the operation of outputting effective order information for each block.

In the image data restoration device shown in FIG. 10, a demultiplexer 94 is provided upstream of the variable length decoding section 56. The demultiplexer 94 separates the code data input from the terminal 54 into effective order information and image data, and the image data is decoded by the variable length decoding section 56, while the effective threshold information is held in the effective order information holding section 96. Ru. The effective order information holding section 96 receives the read signal RED from the timing control section 80.
In response to this, an effective address generated from the effective order information is set in the comparator 76. From now on, the 8th FI! By performing the same processing as the restoring device of J, it is inversely transformed into DCT coefficients.

Figures 11 and 12 show circuit block diagrams of ADCT encoding circuits as another embodiment for adding effective order information to encoded data.

Here, FIG. 11 shows an example in which the effective order information for each pixel is directly added to the encoded data, and FIG. 12 shows an example in which the effective order information for the entire image is encoded all at once. .

First, the following encoding operation is performed in both of FIGS. Although input from terminal 10, the signal is two-dimensional D
A CT conversion unit 12 performs DCT conversion. Effective order detection unit 2
4 scans the DCT coefficients to determine the effective order as shown in FIG. The linear quantization unit 14 linearly quantizes the DCT coefficients according to the effective order information obtained by the effective order detection unit 24. That is, if the quantized address is within the effective address based on the effective order, the quantized DCT coefficient is calculated by dividing the DCT coefficient by the quantization threshold, and if the quantized address is after the effective address, the quantized DCT coefficient is calculated. Set the zero signal without performing.

The quantized DCT coefficients output from the linear quantization unit 14 are variable-length encoded by the variable-length encoding unit 18, held in the variable-length code holding unit 98, and then output to the multiplexer 92.

On the other hand, the effective order information held in the effective order information holding unit 88 is directly output to the multiplexer 92 in the case of FIG. Output. The multiplexer 92 adds the two input data and outputs the result from a terminal 95 as a code.

Here, if the code data is H1 effective order information is Y, the variable length code holding section 98 is configured with buffer A, and the effective order information holding section 88 is configured with buffer B, then from terminal 95 in FIG. The following types of codes can be considered to be output.

■If there are no buffers A and B; HY, HY, HY... ■If there is no buffer A; HHH... YYY... ■If there is no buffer B; YYY... HHH... Similarly, the terminal shown in FIG. 12 outputs the following pattern.

■If buffer A exists; YYY... HHH... ■If buffer A does not exist; HHH... YYY... In other words, in the above, code data H and effective order information Y are output in pixel units, In the above ■~■, the code data H is displayed in units of screens.
and outputs effective order information Y.

Figure 13 shows A corresponding to the ADCT encoding circuit in Figure 11.
An embodiment of the DCT restoration circuit is shown, and FIG.
An embodiment of an ADCT restoration circuit corresponding to the ADCT encoding circuit shown in the figure is shown.

That is, FIG. 13 shows a method for restoring a code when effective order information is directly added to code data, and FIG.
The figure shows a method for restoring a code when the effective order information of the entire image is encoded all at once.

In FIGS. 13 and 14, the code input from the terminal 54 is separated into code data and effective order information by a demultiplexer 94. The code data is input to the variable length decoding section 56, and the quantized DCT coefficients are decoded. On the other hand, in the case of FIG. 13, the effective order information is directly input to the effective order holding unit 96, and in the case of FIG. be done.

The subsequent operations are the same as in FIGS. 13 and 14, and the linear inverse quantization section 60 inversely quantizes the quantized DCT coefficients according to the effective address from the effective order detection section 96 and outputs it to the DCT inverse T inverse transform 4. . In other words, if it is before the valid address,
An inverse quantization operation is performed in which the quantized DCT coefficient is multiplied by a quantization threshold, and if the address is after the effective address, a zero signal is set in the DCT coefficient without performing the inverse quantization operation. The two-dimensional DCT inverse transform unit 64 performs DCT inverse transform on the input DCT coefficients,
Restore the image signal. In this way, one screen of image data is restored.

In the above embodiment, in both encoding and restoration, the first and last eighth rows and columns in a block are scanned to determine the effective order, assuming that the gradation changes are large. In this example, we scan n rows or columns that are positive integers greater than or equal to 1 in a block, and then calculate Alternatively, the effective order may be determined, and the effective address may be set from this effective order to perform quantization and inverse quantization.

[Effects of the Invention] As explained above, according to the present invention, in quantization processing and inverse quantization processing, the effective order of non-zero coefficients is determined for each block, and the coefficients exceeding the effective order are given zero. Significantly reduces the amount of calculations and inverse quantization operations,
Quantization and inverse quantization processing can be performed at high speed.

[Brief explanation of drawings]

Fig. 1A is an explanatory diagram of the principle of the present invention; Fig. 1B is an explanatory diagram of the operation of the present invention; Fig. 2 is a configuration diagram of an embodiment of the image data encoding device of the present invention; Fig. 3 is an effective order in Fig. 2. An example of actual measurement of the detection section; FIG. 4 is a circuit diagram of an embodiment of the linear quantization section in FIG. 2; FIG. 5 is an explanatory diagram of image data encoded by the present invention;
The figure is an explanatory diagram of DCT coefficients obtained from the image data in Figure 5; Figure 7 is a configuration diagram of an embodiment of the image data restoration device of the present invention; Figure 8 is an implementation of the linear inverse quantization section in Figure 7. Example circuit diagram; Figure 9 is an example configuration diagram of an image data encoding device when effective order information is added to encoded data;
A block diagram of an embodiment of an image data restoring device for restoring the code of the image data voice conversion device shown in the figure; Fig. 11 is a circuit block diagram of an ADCT encoding circuit according to the present invention in which effective order information is added as is to the code data. ; Figure 12 is a circuit block diagram of an ADCT encoding circuit according to the present invention when effective order information is encoded and added to encoded data; Figure 13 is a circuit block diagram of an ADCT encoding circuit corresponding to the ADCT encoding circuit of Figure 11;
A block diagram of the DCT decoding circuit; Figure 14 is a block diagram of the ADCT encoding circuit shown in Figure 12.
A block diagram of a DCT decoding circuit; FIG. 15 is a block diagram of a conventional ADCT encoding circuit; FIG. 16 is a block diagram of a two-dimensional DCT transform unit in FIG. 15; FIG. 17 is a block diagram of one block of original image signal Figure 18 is an illustration of the DCT coefficient when the image signal in Figure 8 is subjected to DCT transformation; Figure 19 is an illustration of the quantization threshold adapted to visual perception; Figure 20 is the quantization of Figure 19 An explanatory diagram of quantized DCT coefficients when the DCT coefficients of FIG. 18 are quantized using quantity values; 2115th! ff is a diagram explaining the scanning order for quantizing DCT coefficients; Fig. 22 is a block diagram of a conventional ADCT method restoration circuit; Fig. 23 is a block diagram of the two-dimensional inverse DCT transform unit shown in Fig. 22; Fig. 24 is a block diagram of a conventional linear quantization circuit; No. 25
The figure is a block diagram of a linear inverse quantization circuit. In the figure, 12: Quadratic DCT transform unit 14: Linear quantization unit 16.62: Quantization matrix 18: Variable length encoding unit 20: Code table 24.68: Effective order detection unit 28: DCT coefficient holding unit 30: Quantization address generation section 32.74: Quantization threshold holding section 34.76: Comparator 36: Division section (quantization section) 38.82.92: Multiplexer 40.80: Timing control section 42.84: Zero generator 44.86: Latch section 46: 1st highest degree search section 48: 8th highest degree search section 50: OR calculation section 52: Effective address determination section 56: Variable length encoding section 58: Decoding table 60: Linear inverse quantum Converting unit 64: Two-dimensional inverse DCT transform unit 70: Quantization coefficient holding unit 72: Inverse quantization address generation unit 78: Multiplication unit (inverse quantization unit) 88: Effective order holding unit 94: Demultiplexer 98: Variable length code Holding section 100: Variable-length code section (for effective order information) 102: Variable-length decoding section (for effective order information) (b) Layer of the invention, Hi-Ikari Ming Decoy Figure 1A/$4! Ming Dynasty 11th Roaring Light Group Day 1B Figure Thread 2 (8) Button Illusion Nod Pressing Work Death's Crossfire Ikai Relationship Figure 3 JP 12 Group 171 Engraving Call Volume ÷ 1 and Death's Containment 1 De 1 Figure 4 LtllFl of II imitation data in this R light Figure 5 Figure 6 Figure 1 Arotu 7 (7'1lll [W image signal ♂ light surface Figure 17 DCT coefficient division +'FII!!1 18th Figure 19: Quantization DCT gAe alFIII Figure 19 Scanning order of quantization gA number Adjacent brightness reduction Block diagram of a conventional tooth-shaped quantization circuit Figure 19

Claims (14)

[Claims] (1) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. In an image data encoding and restoring method that quantizes DCT coefficients, performs two-dimensional variable-length encoding on the quantized DCT coefficients, and restores an image by inversely transforming the encoded coded data, the method includes: Scan the n rows and/or columns in each block of DCT coefficients, calculate the effective order by ORing the highest order among the non-zero coefficients whose value is not zero, and then calculate the effective order by scanning the n rows and/or columns in each block of DCT coefficients. All quantized DCT coefficients are set to zero, and during restoration, n rows and/or columns in each block of decoded quantized DCT coefficients are scanned and the highest order among the non-zero coefficients whose value is not zero is found. An image data encoding and restoring method characterized in that an effective order is determined from the logical sum of the quantized DCT coefficients, and all DCT coefficients in rows and/or columns after the effective order are set to zero during inverse quantization of quantized DCT coefficients. (2) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. An image data encoding and restoring method that quantizes DCT coefficients, performs two-dimensional variable length encoding on the quantized DCT coefficients, and further transforms the encoded coded data inversely to restore an image, the method comprising: Scan n rows and/or columns of the DCT coefficients, find the effective order from the logical sum of the highest order among the non-zero coefficients whose value is not zero, and calculate the quanta in the rows and/or columns after the effective order at the time of quantization. The quantized DCT coefficients are all set to zero, and the information on the effective order is added to the coded quantized DCT coefficients to form coded data, and when restoring, the effective order information is separated from the coded data and made effective. An image data encoding and restoring method characterized by detecting an order and setting all DCT coefficients in rows and/or columns after the effective order to zero during dequantization of quantized DCT coefficients. (3) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. In an image data encoding method for quantizing DCT coefficients and performing two-dimensional variable length encoding on the quantized DCT coefficients, a first step of temporarily holding the DCT coefficients after DCT transformation; a second step of scanning n rows and/or columns of the block of retained DCT coefficients and detecting the highest order among the non-zero coefficients; the highest order of the non-zero coefficients detected in said second step; a third step of determining the effective order from the logical sum of; quantization DCT exceeding the effective order obtained in the third step;
a fourth step of selecting zero for the coefficient; and selecting zero for the quantized DCT coefficient exceeding the effective order obtained in the third step and omitting the quantization operation of the DCT coefficient. Image data encoding method. (4) In the image data encoding method according to claim 3, in the second step, the first row and/or column and the last row and/or column in the block are scanned to obtain the effective order. An image data encoding method characterized by: (5) In the image data encoding method according to claim 3, the effective order obtained in the third step is added to the encoded quantized DCT coefficient as effective order information to form code data. An image data encoding method characterized by: (6) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. In an image data encoding device that quantizes DCT coefficients and performs two-dimensional variable length encoding of the quantized DCT coefficients, the DCT coefficient holding means (200) temporarily holds the DCT coefficients after DCT transformation; DCT held in the DCT coefficient holding means (200)
scan n rows and/or columns of the block of coefficients;
Highest order detection means for detecting the highest order among non-zero coefficients (
300); effective order detection means (400) for determining an effective order from the logical sum of the highest order of the non-zero coefficients detected by the highest order detection means (300); quantization address holding means (500); Compares the effective order of the effective order detection means (400) with the address of the quantization address holding means (500), and sets it to zero if the quantization address is equal to or higher than the effective order. An image data encoding device, comprising: a zero selecting means (600) for selecting; and selecting zero for quantized DCT coefficients of the effective order or later to omit a quantization operation of the DCT coefficients. (7) In the image data encoding device according to claim 6, the highest order detecting means (300) is configured to detect the first order in the block.
An image data encoding device characterized in that the highest order is determined by scanning the th row and/or column and the last row and/or column. (8) In the image data encoding device according to claim 6, the effective order obtained by the effective order detecting means (400) is added to the encoded quantized DCT coefficient as effective order information to generate code data. An image data encoding device characterized in that: (9) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. In an image data restoration method that quantizes DCT coefficients and restores an image by inverse transformation of encoded data obtained by encoding the quantized DCT coefficients, a first step of temporarily holding the decoded quantized DCT coefficients; a second step of scanning n rows and/or columns of the block of quantized DCT coefficients retained in the first step to detect the highest order among the non-zero coefficients; a third step of calculating an effective order from the logical sum of the detected highest orders; a fourth step of selecting zero for DCT coefficients exceeding the effective order obtained in the third step; An image data restoring method characterized in that zero is selected for the coefficients and inverse quantization calculation of quantized DCT coefficients is omitted. (10) In the image data restoration method according to claim 9, the second step is characterized in that the first row and/or column and the last row and/or column are scanned to find the highest order. How to restore image data. (11) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. Quantize the DCT coefficients and apply the quantized DCT
In an image data restoration method of restoring an image by inverse transformation of coded data in which coefficients are encoded, the first step is to temporarily hold the quantized DCT coefficients decoded from the coded data; and extract effective order information from the coded data. a second step of separating and detecting the effective order; a third step of generating zero as a dequantized DCT coefficient from the quantized DCT coefficient; and the effective order obtained in the second step. An image data restoring method, characterized in that zeros generated in the third step are selected as DCT coefficients of subsequent orders, and inverse quantization of the quantized DCT coefficients is omitted. (12) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. Quantize the DCT coefficients and apply the quantized DCT
An image data restoration device that restores an image by inverse transformation of encoded coefficient data includes: a quantized coefficient holding means (210) that temporarily holds decoded quantized DCT coefficients; and the quantized coefficients. highest order detection means (310) for scanning n rows and/or columns of the block of quantized DCT coefficients held in the holding means (210) and detecting the highest order among the non-zero coefficients; effective order detection means (410) that calculates an effective order from the logical sum of the highest orders detected by the highest order detection means (310);
); dequantization address holding means (510) for holding an order address for dequantizing the quantized DCT coefficient; and the highest order of the effective order detection means (410) and the dequantization address holding means (510). and zero coefficient selection means (610) for selecting zero for the DCT coefficient when the dequantization address is equal to or higher than the effective order; and zero for the DCT coefficient exceeding the effective order. An image data restoring device characterized in that an inverse quantization operation of quantized DCT coefficients is selectively omitted. (13) In the image data restoration device according to claim 1, the highest order detecting means (310) scans the first row and/or column and the last row and/or column to find the effective order. An image data restoration device characterized by: (14) For each block obtained by dividing the original image into a plurality of blocks each consisting of a plurality of pixels, the tone values of the plurality of pixels in the block are obtained by two-dimensional discrete cosine transformation. Quantize the DCT coefficients and apply the quantized DCT
In an image data restoration device that restores an image by inversely transforming code data obtained by encoding coefficients, the apparatus comprises: a quantized coefficient holding means for temporarily holding decoded quantized DCT coefficients; and effective order information from the coded data. effective order information separating means for separating and detecting an effective order; dequantization address holding means for holding an order address for dequantizing the quantized DCT coefficient; comparing the effective coefficient and the dequantization address; and a zero coefficient selection means for selecting zero for the DCT coefficient when the inverse quantization address is equal to or greater than the effective order; An image data restoration device characterized by omitting quantization operations.