JP2820807B2 - Image data encoding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data encoding method, and more particularly, to an image to be encoded after dividing a multi-valued image into blocks composed of a plurality of pixels, and performing orthogonal transformation on the gradation values of the pixels in the block. The present invention relates to a data encoding method.

In order to store image data whose information amount is orders of magnitude larger than numerical data, in particular, data of halftone images and color images, or to transmit them at high speed and with high quality, it is necessary to use a pixel-by-pixel level. The key values need to be encoded efficiently.

A highly efficient compression method for image data is as follows.
For example, adaptive discrete cosine transform coding (Adaptive Dis
The Crete Cosine Transform (hereinafter abbreviated as "ADCT") is known.

In the ADCT method, an image is divided into blocks each having 8 × 8 pixels, and an image signal of each block is converted into DCT coefficients of a spatial frequency distribution by a two-dimensional discrete cosine transform (hereinafter, referred to as “DCT transform”). Then, the quantized DCT coefficients are quantized by a threshold value suitable for the visual sense, and the obtained quantized DCT coefficients are encoded by a Huffman table statistically obtained.

[0005]

2. Description of the Related Art FIG. 5 is a block diagram of a conventional ADCT encoding circuit. The image is divided into blocks of 8 × 8 pixels shown in FIG.
To enter. In the two-dimensional DCT transform unit 32, the input image signal is orthogonally transformed by DCT, converted into DCT coefficients of a spatial frequency distribution shown in FIG. 10 and output to the linear quantization unit 33. Specifically, as shown in FIG. 6, the image signal input from the terminal 31 is converted into one-dimensional data by the one-dimensional DCT converter 32a.
Dimensional DCT is performed and output to the transposition unit 32b. The transposition unit 32b exchanges the rows and columns of the coefficients in one block and outputs the result to the one-dimensional DCT transform unit 32c. One-dimensional DCT transform section 32c converts input transposed coefficients into one-dimensional D
CT conversion is performed and output to the transposition unit 32d. Transposed part 32d,
Similarly to the transposition unit 32b, a signal obtained by exchanging the rows and columns of the coefficients in one block again is output from the terminal 31 'to the linear quantization unit 33.

Referring to FIG. 5 again, the linear quantization unit 3
In step 3, the input DCT coefficient is converted into a quantization matrix 3 composed of the threshold values shown in FIG.
4 is multiplied by a quantization control parameter to obtain a quantization threshold, and a DCT coefficient is divided by the quantization threshold to perform a linear quantization for obtaining a quantized DCT coefficient. As a result of this quantization, as shown in FIG. 12, the DCT coefficient equal to or smaller than the threshold value becomes zero, and a quantized DCT coefficient having only DC component and a small AC component has a value.

[0007] The quantized DCT coefficients arranged two-dimensionally are converted into one dimension by zigzag scan shown in FIG. 13 and input to the variable length coding unit 35. The variable length coding unit 35 calculates the DC component at the head of each block and the D component of the previous block.
The difference from the C component is variable-length coded. For the AC component, a variable-length coding is performed on the value of a non-zero effective coefficient (hereinafter, referred to as an “index”) and the length of a run of an invalid coefficient that becomes zero up to the index (hereinafter, referred to as a “run”). Further, DC and AC components are encoded using a code table 36 composed of a Huffman table created based on statistics for each image, and the obtained code data is sequentially output from a terminal 37. Is done.

On the other hand, the code data is restored to an image by the following method. FIG. 7 shows a block diagram of the ADCT restoration circuit. In FIG. 7, code data input from a terminal 10 is input to a variable-length decoding unit 11. Variable length decoding unit 1
In step 1, the input code data is decoded into fixed-length data of an index and a run by the decoding table 12 composed of a table opposite to the Huffman table of the code table 36 (see FIG. 5), and the inverse quantum Output to the conversion unit 51. The inverse quantization unit 51
By multiplying each of the quantization matrices 34 by a value (quantization threshold) multiplied by a quantization control parameter, the input quantized DCT coefficient is inversely quantized to restore the DCT coefficient, and the two-dimensional inverse DCT is performed. Output to the conversion unit 52.

[0009] The two-dimensional inverse DCT transform unit 52 orthogonally transforms the input DCT coefficients by inverse DCT transform, and converts the DCT coefficients of the spatial frequency distribution into image signals. In particular,
As shown in FIG. 8, the DCT coefficient input from the terminal 21 is one-dimensional inverse DCT transformed by the one-dimensional inverse DCT transformer 23 and output to the transpose unit 25. The transposition unit 25 interchanges the rows and columns of the coefficients in one block and outputs the result to the one-dimensional inverse DCT transform unit 26. One-dimensional inverse DCT unit 26 performs one-dimensional inverse DCT on the input transposed coefficients again and outputs the result to transposed unit 28. The transposition unit 28 outputs a signal obtained by exchanging the rows and columns of the coefficients in one block again from the terminal 14 similarly to the transposition unit 25, thereby restoring the image.

In such an ADCT method, there is a case where it is necessary to control a code amount when an image is coded so as to be a desired code amount. Such needs include, for example,
This occurs when a predetermined number of images are input to a limited-capacity storage medium (for example, an IC memory or a floppy disk).

Conventionally, after DCT coefficients are quantized, variable length coding is performed to determine the code amount. If the code amount is different from the target code amount, the quantization roughness is changed (by changing the quantization threshold). ,
The quantization and variable length coding are performed again, and thereafter, the coding is repeated a plurality of times to control the code amount to a desired value.

As another method, encoding is performed only once, the code amount of the variable-length code is counted, the quantization roughness is predicted from the code amount, and the code amount for each block is limited. Quantization and variable length coding are performed again.

[0013]

As described above, in the prior art, in order to perform encoding so as to obtain a desired code amount, variable-length encoding is performed at least once, and code amount and quantization are performed. The relationship with the roughness is determined, and the roughness of the quantization that provides a desired code amount is determined from the relationship. This means that feedback control is performed between variable-length coding and quantization, and there is a problem that processing becomes complicated.

Also, variable-length encoding requires a considerable amount of time,
There is a problem that it takes a considerable time to obtain the desired quantization roughness. In particular, the first method requires several times of variable-length encoding, so that such a problem becomes conspicuous.

The object of the present invention has been made in view of such conventional problems, and there is no need to perform variable-length encoding, and encoding is performed with a simple configuration and processing so that a desired code amount is obtained. It is an object of the present invention to provide a method for encoding image data.

[0016]

FIG. 1 is a diagram illustrating the principle of the present invention. Numeral 1 denotes a two-dimensional D for orthogonally transforming an input image signal by DCT to convert to a DCT coefficient of a spatial frequency distribution.
The CT transform unit 3 is a linear quantization unit that performs a linear quantization process on DCT coefficients and outputs quantized DCT coefficients. 5 is a code table composed of Huffman tables. 6 is a code table using a code table. A variable length coding unit for coding the quantized DCT coefficient into a variable length code. The variable length coding unit 7 accumulates the number of effective coefficients included in the quantized DCT coefficient obtained by quantization and calculates the number n of effective coefficients for one screen. The effective coefficient accumulating unit 8 for calculating the code amount is a code amount control unit that determines the quantization threshold value q (i) based on the number n of effective coefficients and controls the code amount so that the designated code amount F is obtained.

[0017]

The original image is divided into a plurality of blocks of N × M pixels, and the tone values of the plurality of pixels in each block are orthogonally transformed by the two-dimensional DCT transform unit 1 to obtain an N × M DCT.
Output the coefficient. The linear quantization unit 3 performs N × per block.
Each of the M DCT coefficients is represented by N × M corresponding quantization thresholds q (i) (q (i) is a positive integer of 0 or more, i = 0 to N ×
The effective coefficient accumulating unit 7 accumulates the number of effective coefficients included in the quantized DCT coefficients obtained by quantization, and obtains the total number n of effective coefficients for one screen. . The code amount control unit 8 determines the quantization threshold value q (i) based on the number n of effective coefficients so that the code amount F becomes the designated code amount F, and controls the variable length code output from the variable length coding unit 6. Control the code amount. In this way, it is not necessary to perform variable-length encoding, encoding can be performed with a simple configuration and processing to obtain a desired code amount, and a quantization threshold value q (i ) Can be obtained in a short time.

[0018]

EXAMPLES In the overall configuration diagram 2 is an example block diagram of an image data coding method of the present invention, it is denoted by the same reference numerals as in FIG. 1.

Reference numeral 1 denotes an image signal data P of 8 × 8 pixels for each block when the image is divided into blocks of 8 × 8 pixels.
A two-dimensional DCT conversion unit which receives CS, and orthogonally transforms the image signal data by DCT to convert it into 8 × 8 DCT coefficients of a spatial frequency distribution. 2 is a DCT coefficient which stores 8 × 8 DCT coefficients. It is a holding unit.

Reference numeral 3 denotes a linear quantization unit that performs a linear quantization process on the DCT coefficient and outputs a quantized DCT coefficient.
Each of 8 × 8 DCT coefficients per block is 8 × 8
The corresponding quantization thresholds q (i) (where q (i) is a positive integer greater than or equal to 0, i = 0 to 63) are quantized.

Reference numeral 4 denotes a quantized DCT coefficient holding unit for storing the quantized DCT coefficients obtained by the quantization.
A code table 6 composed of tables is a variable-length coding unit. When the 8 × 8 quantized DCT coefficients arranged two-dimensionally are converted into one-dimensional by zigzag scanning and input, the input is performed. Each quantum DCT coefficient is encoded using the code table 5 in order.

Reference numeral 7 denotes an effective coefficient accumulator for accumulating the number of effective coefficients included in the quantized DCT coefficients obtained by quantization to calculate the total number n of effective coefficients for one screen, and 8 denotes a designated code amount. A code amount control unit that determines a quantization threshold value q (i) based on the number n of effective coefficients to control the code amount so as to be F, a code amount / effective coefficient conversion unit 81 and a quantization threshold value determination unit 82.

The code amount / effective coefficient conversion section 81 calculates the total number m of effective coefficients for one screen from the designated desired code amount F. Since a proportional relationship is established between the code amount and the number of average effective coefficients per block as shown in FIG. 3, the code amount / effective coefficient conversion unit 81 calculates the code amount and the effective coefficient according to the designated coding amount F from the proportional relationship. Calculate the total number m of effective coefficients for one screen

The quantization threshold value determination unit 82 obtains 1 based on the total number n of effective coefficients for one screen and the designated code amount.
Quantization threshold q by difference of total number m of effective coefficients for screen
(i) is determined, and 8 × 8 quantization thresholds Q (i) (Q (i) is a positive integer equal to or greater than 0, i = 0 to 0)
63), and a quantization threshold Q based on a difference between the total number n of effective coefficients for one screen and the total number m of effective coefficients for one screen according to the designated code amount.
A scaling control unit 82b for determining a scaling factor (weighting factor) SF for scaling (i), and multiplying the quantization threshold Q (i) by the weighting factor SF to obtain an actual quantization threshold q ( i) is provided.

If the weight coefficient SF is increased, the quantization threshold q
(i) increases (in other words, the roughness of quantization increases), the total number n of effective coefficients decreases, and if the weighting coefficient SF is reduced, the quantization threshold q (i) decreases ( In other words, the roughness of the quantization is reduced), and the total number n of effective coefficients is increased. That is, the relationship between the weight coefficient and the total number of effective coefficients is as shown in FIG. Therefore, the scaling control unit 82b increases the weight coefficient SF when the difference between the effective coefficients n and m is not within the set range, and when n> m, and when the difference is not within the set range. , And n <m, the weighting factor SF is reduced.

Overall Quantization Threshold q (i) (q (i) is a positive integer greater than or equal to 0, i = 0, so that the code amount of data obtained by variable length coding becomes substantially equal to the desired code amount. To 63) are determined by the following steps. Note that the quantization threshold value holding unit 9a stores in advance the quantization threshold value Q (i) (Q (i) is a positive integer of 0 or more, i = 0 to 63).
, And the weighting factor SF is 1 at the initial stage.

(1) The two-dimensional DCT transform unit 1 orthogonally transforms the input image signal data of 8 × 8 pixels of one block by DCT to obtain 8 × 8 DCT coefficients. Remember.

(2) The code amount / effective coefficient conversion section 81 calculates the total number m of effective coefficients for one screen from the input desired code amount F based on the relationship shown in FIG. Input to the section 82b.

(3) The DCT coefficient holding unit 2 sequentially outputs DCT coefficients one by one to the divider 3a of the linear quantization unit 3 according to a data read signal RDT from a timing control unit (not shown).

(4) Simultaneously with the reading of the DCT coefficients, the quantization threshold value Q (i) read from the quantization threshold value holding section 82a is set to the initial weight coefficient SF (1) output from the scaling control section 82b. ) To obtain a quantization threshold value q (i), which is input to the divider 3a.

(5) The divider 3a divides the received DCT coefficient by the quantization threshold value q (i), and outputs the division result (quantized DCT coefficient) to the quantized DCT coefficient holding unit 4 and the effective coefficient accumulating unit 7. Output.

(6) The effective coefficient accumulating section 7 counts up when a non-zero coefficient among the quantized DCT coefficients is input. (7) When the quantization of one DCT coefficient is completed, the timing control unit instructs the DCT coefficient holding unit 2 to read the next DCT coefficient, and quantizes the next DCT coefficient.

(8) In this way, the DCT coefficient holding unit 2
Is read out one by one, and the result is output as a quantized DCT coefficient.
By repeating the process of accumulating the number of effective coefficients in the effective coefficient accumulating unit 7 for one screen in block units, the number n of effective coefficients for one screen is obtained.

(9) The calculated total number n of effective coefficients is
Output to the ring control unit 82b. (10) The scaling control unit 82b determines whether to update the weighting coefficient SF based on the magnitude relationship between the total number m of effective coefficients corresponding to the designated code amount and the total number n of obtained effective coefficients. .

(11) That is, if the difference between n and m is within the preset allowable range, the updating of the weighting factor SF is terminated. If the difference is out of the allowable range and n> m, the weighting factor SF is updated.
Is increased by a predetermined amount, and when the value is out of the allowable range and n <m, the weight coefficient SF is reduced.

(12) When the weighting coefficient SF is updated, the above-described processing is performed after clearing the accumulated value n of the effective coefficient accumulating unit 7 to zero, and the difference between n and m becomes smaller and the allowable value becomes smaller. Repeat until within range.

(13) If the difference between n and m falls within the allowable range,
The DCT coefficient is converted into a quantized DCT coefficient using the quantization threshold value q (i) at that time, and the quantized DCT coefficient is encoded by the variable-length encoding unit 6 and output.

When updating the weighting coefficients, the correspondence between the weighting coefficients and the number of effective coefficients may be determined in advance, and the correspondence may be used.

The present invention has been described with reference to the embodiments.
The present invention can be variously modified in accordance with the gist of the present invention described in the claims, and the present invention does not exclude these.

[0040]

As described above, according to the present invention, the original image is divided into a plurality of blocks each consisting of a plurality of pixels, and for each block, the gradation values of the plurality of pixels in the block are orthogonally transformed, and Since the transform coefficients obtained by the transform are quantized to obtain quantized coefficients, the number of non-zero effective coefficients among the quantized coefficients is accumulated, and the code amount is controlled based on the number of effective coefficients. , It is possible to obtain a quantization threshold value that provides a desired code amount without performing variable-length coding.
Encoding can be performed with a simple configuration and processing to obtain a desired code amount.

Also, according to the present invention, N ×
Each of the M transform coefficients is represented by N × M corresponding quantization thresholds q (i) (q (i) is a positive integer of 0 or more, i = 0 to N × M
-1), quantize, and accumulate the number of effective coefficients included in the quantized coefficients obtained by the quantization to obtain the number n of effective coefficients for one screen, and the number n of effective coefficients for one screen And the quantization threshold q (i) is changed according to the difference between the number m of effective coefficients obtained based on the specified code amount, and when the difference becomes small and falls within a set range, the quantization threshold q at that time is changed.
Since variable-length coding is performed using (i), it is not necessary to perform variable-length coding until the quantization threshold q (i) is obtained. In other words, it is possible to perform variable-length coding without performing variable-length coding. It is possible to obtain a quantization threshold value which is a code amount, and it is possible to obtain a quantization threshold value q (i) for obtaining a desired code amount in a short time.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention.

FIG. 3 is a relationship diagram between an effective coefficient and a code amount.

FIG. 4 is a relationship diagram between a weighting coefficient SF and the number of effective coefficients.

FIG. 5 is a block diagram of a conventional ADCT encoding circuit.

FIG. 6 is a block diagram of a two-dimensional DCT transform unit.

FIG. 7 is a block diagram of a conventional ADCT restoration circuit.

FIG. 8 is a block diagram of a two-dimensional inverse DCT transform unit.

FIG. 9 is an explanatory diagram of an original image signal.

FIG. 10 is an explanatory diagram of DCT coefficients.

FIG. 11 is an explanatory diagram of a threshold value for a DCT coefficient.

FIG. 12 is an explanatory diagram of quantized DCT coefficients.

FIG. 13 is an explanatory diagram of a scanning order at the time of quantization.

[Explanation of symbols]

 1 2D DCT transform unit 3 Linear quantization unit 5 Code table 6 Variable length coding unit 7 Effective coefficient accumulating unit 8 Code amount control unit 81 Code amount effective coefficient Transform unit 82 Quantization threshold value decision unit

Claims (5)

(57) [Claims] 1. An original image is divided into a plurality of blocks each composed of a plurality of pixels, and for each block, tone values of a plurality of pixels in the block are orthogonally transformed, and a transform coefficient obtained by the orthogonal transform is obtained. In the image data encoding method of obtaining a quantized coefficient by quantizing the quantized coefficient and accumulating the number of non-zero effective coefficients among the quantized coefficients obtained by quantizing the transform coefficient, An image data encoding method comprising controlling a code amount based on the number of image data. 2. An image obtained by dividing an original image into a plurality of blocks each composed of N.times.M pixels, and for each block, orthogonally transforming gradation values of a plurality of pixels in the block, and performing orthogonal transformation. In an image data encoding method for quantizing a transform coefficient to obtain a quantized coefficient and encoding the quantized coefficient, each of the N × M transform coefficients in the block is converted to N × M corresponding quantization thresholds. q (i) (q (i) is a positive integer of 0 or more, i = 0
ＮN × M−1), quantize, and accumulate the number of effective coefficients included in the quantized coefficients obtained by quantization to obtain the number n of effective coefficients for one screen. An image data encoding method, wherein a quantization threshold value q (i) is changed based on a difference between the number n of coefficients and the number m of effective coefficients calculated based on a designated code amount. 3. The image data encoding method according to claim 2, wherein the quantization threshold is changed until the difference falls within a set range. 4. The method according to claim 2, wherein a relationship between the code amount and the number of effective coefficients is determined in advance, and the number m of effective coefficients according to the designated code amount is calculated using the relationship. The image data encoding method described in the above. 5. N × M quantization thresholds Q (i) (Q
(i) is a positive integer equal to or greater than 0, i = 0 to N × M−1), and the quantization threshold q (i) is multiplied by a weighting coefficient SF to obtain the quantization threshold q ( In calculating i), when the difference is not within the set range and n> m, the weight coefficient SF is increased, and the difference is not within the set range,
4. The image data encoding method according to claim 3, wherein when n <m, the quantization threshold value q (i) is changed by decreasing the weight coefficient SF.