JPS639283A - Color picture coding system - Google Patents

Abstract

PURPOSE:To improve the resolution, reproducibility of color gradation and compression taking the visuality into account by providing a coder comprising a block coder or the like and a coder comprising a block smoothing device or the like so as to facilitate the connection with a picture I/O or the like. CONSTITUTION:Primary color signals R, G, B outputted from a color picture input device 1 are converted into uniform color space signals L*, a*, b* at every picture element by a color converter 2 and stored in a buffer memory 3. Then coders 4, 5 segment the signals L*, a*, b* in the unit of (n)X(n) picture blocks from a memory 3 to apply coding at block. In such case, in coding the lightness data L*, it is processed via a block coder 4-1 and a vector quantizer 4-2 and stored in a buffer memory 6 as a code data C1. Further, the chromaticity information a*, b* are degenerated into a code data C2 via a block smoothing device 5-1 and a color coder 5-2 and stored in the memory 6. Then the data C1, C2 are synchronized and stored in a memory 7 as the degenerated code data. Thus, the efficient picture coding is applied.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to a color image encoding method for encoding image data suitable for transmission and recording of color images.

[Prior art]

In order to transmit or record color images, which generally require a huge amount of data compared to black and white images, there are problems such as long transmission and recording times, line usage costs associated with transmission time,
There were problems such as the high cost of the hardware of the recording medium.

Therefore, encoding is required to compress the amount of color image data.

In general, a color image is basically described using three types of signals, and a method that compresses data of these three types of signals at once has problems such as an increase in hardware scale.

Additionally, in general, image input/output devices often use three primary color signals R, G, and B as three types of signals, but these three primary color signals R, G, and B are unique to the image input/output device. In many cases, the characteristics of the device were included (and troublesome effort was required to connect the devices).

Furthermore, since there is almost no difference in the spatial frequency characteristics of the three primary colors corresponding to the three primary color signals R, G, and B, compression by reducing the spatial frequency has a problem in that the compression rate is poor.

Therefore, N T which linearly converted the three primary color signals R, G, and B
, chromaticity data I, using the SC system YIQ signal.
There is a method to reduce the spatial frequency of Q. However, Y
, I.

Q Each signal is a psychophysical quantity, and no consideration is given to linearity with respect to the visual system, uniformity between color discrimination, etc., and input image data is converted, quantized, encoded, and decoded. When doing this, there is a problem in that visually effective data is rounded up as an error, or data is allocated in a visually unnecessary range. Also, to solve this problem, Y.

Complex corrections are required in spaces of psychophysical quantities such as I and Q signal spaces.

〔the purpose〕

In view of the above-mentioned problems, the present invention provides a color image encoding method that facilitates connection with image input/output devices, etc., achieves high resolution and color gradation reproducibility in consideration of visual characteristics, and high compression rate. The purpose is to provide.

[Examples] The present invention will be described in detail below using preferred examples.

FIG. 1 is a block diagram of a color image encoding/decoding apparatus which is an embodiment of the present invention for encoding and decoding color image data. However, as an example here, a case will be described in which recording/reading from an image data memory or the like is performed.

Here, three primary color signals R (red) and G (green) are used as image data handled by the image input device and the image output device.

B (blue) is used, and the uniform color space is CIE197.
This will be explained using the 6L*a*b* uniform color space.

1 is a color image input device such as an image reading device that photoelectrically reads a color image using an image sensor such as COD, 2 is a signal R9G from the color image input device 1, and B is a uniform color space signal L*a*b*. 3 is a buffer memory that temporarily stores output data from the color converter for multiple lines; 4 is an encoder that encodes brightness data and *; here, a block encoder 4-1 and It is composed of a vector quantizer 4-2.

Reference numeral 5 denotes an encoder for encoding chromaticity data a* and b*, which is composed of a block smoother 5-1 and a color encoder 5-2. 6 is the output signal C1, C2 from the encoder 4.5
7 is an image data memory for storing encoded image data C for one screen, for example;
8 is a buffer memory for reading from the image data memory 7; 9 is a decoder having an inverse system to the encoder 4; 10;
is a color encoder 5-2 and an inverse system,
11 is a buffer memory that synchronizes the signal tree decoded by the decoder 9 with the signals a* and b* decoded by the decoder 10; 12 is a color conversion system that is the inverse of the color converter 2; 13 is a buffer memory for timing with the color image output device 14.

In addition, the color converters 2 and 12, the color coder 5-2, and the color coder lO
, the vector quantizer 4-2, and the decoder 9 can mainly constitute a look-up table including a memory table configured to read converted data when accessed with input data.

The operation in the above configuration will be explained.

The digitized three primary color signals R, G, and B, which are sequentially output pixel by pixel from the color image input device 1, are converted into uniform color space signals L*, a* for each pixel by a color converter 2.

b* and sequentially stored in the buffer memory 3.

On the other hand, the encoder 4.5 extracts L*, a*, and b* from the buffer memory 3 in units of nXn pixel blocks and encodes each block. This will be explained with reference to FIG.

In FIG. 2, a 4×4 pixel block will be explained as an nXn pixel block.

The brightness data L* is encoded so as to preserve resolution information and level information. First, using brightness data L*ij, which is multivalued data, there is a pattern code PC classified according to the characteristics of code data C8, which is level information, by block encoding 4-1. In addition, pattern code P
As C, variance or the like can be used as a statistic.

Further, the resolution data L''ij is obtained by reducing the number of bits of l*,i based on the code data C8, and here, for simplicity, each pixel is expressed as one bit.

Next, the vector quantizer 4-2 refers to the code data C6, degenerates L"ij into a vector quantization code VQC, and combines it with the code data C6 to create the code data C1.
Store in buffer memory. The vector quantization code VQC is created by a look-up table LTBI having a memory table accessed using C8 as a selector and L*' as an address.

On the other hand, each block is cut out 16 from the buffer memory 3 and smoothed for each block. Smoothed data i*, 5*
code data C2 by the color encoder 5-2 based on
is degenerated into. Note that code data C2 is smoothed data i*
, 5* are used as addresses.

The lookup table here is created by dividing the uniform color space L*a*b*, which can reproduce colors, onto the L*-0 plane into fine areas (areas), and calculating the numerical value (code data C2) that represents the fine areas. ) is a column table of uniform color space L*a*
An example of the reproduction range of b* and area division is shown in FIG.

In this way, the code data C, C2 obtained by the encoders 4 and 5 are asynchronously input to the buffer memory 6,
Thereafter, it is stored in the image data memory 7 as code data C in synchronization.

By sequentially performing the above operations block by block, the color image data is recorded in the image data memory 7 as degenerated code data C.

Next, decoding will be explained.

The degenerated code data C recorded in the image data memory 7 as described above is sequentially read out to the buffer memory 8, divided into code data C, C2, and decoded by the decoder 9 and decoder 10, respectively. do.

The details will be explained using FIG. 4.

First, decoding of the brightness data L* will be explained.

The address is the vector quantization code VQC in the buffer memory 8, and the average value "*" and the pattern code PC
A good method of vector quantization is to reduce such errors by using the selector as a lookup error for decoding. However, in consideration of hardware scale and calculation speed, in this embodiment, the vector quantizer 4-2 and decoder 9 are configured with a lookup table, so the contents of this lookup table can cause encoding/decoding errors. is decided.

The decoded resolution data L*'IJ is further divided into the average value L
Level information is added by * and pattern code PC, and the data is decoded into brightness data L*l of each pixel, which is stored in the buffer memory 11.

On the other hand, the chromaticity data (a*b*) is decoded using the lookup table LTB4°5 for the code data C in the buffer memory 8, using i*, B
* is decoded. i*, H* here are i* in Figure 2,
It does not necessarily match 5*. This is an error caused by degenerating a* and b* as shown in Figure 3, and H*
, B* are decoded into a* and b* of 17 representative points (indicated by different points) in the divided area in FIG. Therefore H
In order to set the difference in your P between * and 5* within a visually acceptable range, create a lookup table for the exemption number, taking into account the method of dividing the area and the method of selecting the representative point.

The decoded a-tree, b*, is the chromaticity data (a*) common to all pixels of this nXn pixel block (here, 4x4).

(i, j = 1-4).

The decoded chromaticity data (a*, b
*), there are two methods: storing only one set of i*, 5*, and storing nXn sets. The former is more advantageous in terms of data amount, and the latter is better in terms of consistency with L*.

L*, a*, b* decoded in this way and stored in the buffer memory 11 are sent to the color converter 1 for each pixel.
2, the signal is converted into three primary color signals R, G, and H, which are sequentially stored in the buffer memory 13.

The above operations are performed for each nXn (here, 4.
) Repeatedly for each pixel block, color image output device 1
4, the buffer memory 13 outputs three primary color signals R, G, and B, and a color image output device 14 such as a color printer reproduces and outputs the color image.

Although the embodiments have been described with reference to recording and reading from the image data memory, the present invention can easily be applied to the transmission of color images by connecting a communication device or a transmission line in place of the image data memory.

Furthermore, in this embodiment, the block encoder 4-1 and the vector quantizer 4-2 are used as the encoder 4 for the brightness data L, but this is not the case (for example, one pull-up table operates at once). It is also conceivable to configure it to do so.

On the other hand, as the encoder 5 for the chromaticity data (a*, b*), the block smoother 5-1 is used to reduce the spatial frequency, but this is not limited to this, for example, pixel thinning, average value in the vicinity, etc. It can also be configured by selecting representative colors. Again, the method of creating the lookup table of the color encoder 5-2 is not limited to that shown in FIG. 3; for example, concentric and radial divisions may also be considered.

Furthermore, in addition to the CIE1976L*a*b* uniform color space, it is also possible to consider the uniform color space as CIEl, 9.76L*u*v* uniform color space, or CIE1764U*V*W* uniform color space. Also, instead of a*, b*, CI
It is also possible to use u and v on the E1960vC8 diagram.

Furthermore, in this embodiment, the lightness data L* and the chromaticity data (a*, b*) are encoded and decoded separately, but for example, I .

It is also possible to change the encoding lookup table using the selector. In this way, in encoding and decoding, lightness data L* is converted to chromaticity data (8 years old).

b*) and the chromaticity data (a*, b*) can also be performed with reference to the lightness data L*.

〔effect〕

As explained above, resolution information is stored by lightness L*, color tone is stored by chromaticity (a*, b*), and color gradation is stored by lightness L* and chromaticity (a*, b*). ) Color image encoding can be configured at low cost.

Furthermore, the introduction of a color converter facilitates connection with color image input/output devices having various color signals.

[Brief explanation of drawings]

Figure 1 is a block diagram of a configuration example of a color image encoding/decoding device to which the present invention is applied, Figure 2 is a diagram showing the flow of image data during encoding, and Figure 3 is a uniform color representation of the color reproduction range. A diagram showing an example of how to encode spatial CIF1976L *a *b * and (a*b*) necessary for color encoding and how to select representative points. Figure 4 is a diagram showing the flow of image data by decoding. 2.12 is a color converter, 4.5 is an encoder, 9 is a decoder, and 10 is a decoder.

Claims (1)

[Claims] After converting the color image data to uniform color space coordinates, the brightness data is encoded to preserve the spatial frequency on the image,
A color image encoding method characterized in that the chromaticity data is encoded so as to reduce the spatial frequency on the image.