JPH0799853B2 - Image data compression / decompression method for printing - Google Patents

Description

The present invention relates to a method and apparatus for compressing / decompressing image data for printing.

(Prior Art) Conventionally, when transmitting printing image data, each color of CMYK (cyan, magenta, yellow, and smear) is handled in the form of 8 bits.

On the other hand, image data is widely used in the field of television / facsimile for the purpose of band compression and the like, and two methods of complete restoration coding and non-restoration coding are adopted.
Of these, the former uses Huffman codes, linear codes, and arithmetic, and completely restores the information of the information source. The latter includes exchange coding, vector quantization, predictive coding, and the like, and codes information with a certain amount of reduction from the information source.

(Problems to be Solved by the Invention) As described above, since the transmission of the printing image data is not compressed, the data amount is enormous, for example, A4 size is about 100 MB, and the data handling is complicated.

Further, although image compression in televisions and facsimiles can perform high-rate compression, it is not suitable for printing due to poor image quality.

An object of the present invention is to provide a method and apparatus for compressing / decompressing image data that can secure image quality suitable for printing and reduce the amount of data.

(Means for Solving Problems) In order to achieve the above object, in the present invention, RGB image data is converted into H (hue), S (saturation), and V (lightness) data according to human visual sense. However, each of the converted data is quantized and compressed to a degree that does not cause visual deterioration, and then the variable length coding is performed to obtain the compressed data. This data is later encoded and HSV
It is converted into data and further converted into RGB data, and a sumi version is generated based on the V data therein.

(Embodiment) FIG. 1 is a block diagram showing a configuration of an apparatus according to the present invention, which shows a system for converting into C'M'Y'K 'data based on RGB data as a whole. Data compression and decompression are performed during the process. This apparatus is configured to carry out the method of the present invention, and the method of the present invention will be described through its operation description.

The RGB data to be supplied to this system is, for example, a signal after scanning a photographic film and performing color separation with a filter. The data extracted from this system is not based on the above CMY but RGB
It can also be due to. Then, from the input to the output, (1) HSV conversion (2) quantization (3) RICE conversion (4) RICE inverse conversion (5) inverse HSV conversion is performed, and color inversion is added as necessary.

Next, these conversion operations (1) to (5) will be sequentially described.

(1) HSV conversion This means an operation of converting RGB data into H (hue), S (saturation), and V (brightness) data, which is performed by the HSV converter 1 in FIG. This HSV conversion is an algorithm that was devised to facilitate the handling of color information in the field of computer graphics, and although it has not been rigorously discussed in terms of color, it has been recognized for its simplicity and effectiveness in various fields. There is. And the contents of the algorithm are as follows.

When 0 ≦ R, G, B ≦ 1, MAX = R, G, B maximum value MIN = R, G, B minimum value, the deviation range RNG is RNG = MAX-MIN, and the brightness V is V = Required as MAX. Then, when MAX = 0, S = 0 and H = undefined. On the other hand, when MAX ≠ 0, the saturation S
Becomes S = RNG / MAX, and the following operators r, g, and b are obtained.

That is, r = (MAX-R) / RNG g = (MAX-G) / RNG b = (MAX-B) / RNG From the operators r, g, and b, when R = MAX, h = b-g G = In the case of MAX, h = r−b + 2 When B = MAX, h = g−r + 4 is obtained, and the hue H is calculated as H = 60 × h + 360 when h ≦ 0 and H = 60 × h when h> 0.

Figures 2 (a), (b) and (c) show HSV from this RGB data.
An example of data analysis that is the target of data conversion is shown. The example shown in FIG. 2 is, for example, a photographic film of a natural landscape. Generally, R, G, B data It can be said that the concentration distribution is centered on the intermediate value.

3 (a), (b), and (c) show the RGB data of FIG. 2 as H.
The result of SV conversion is shown. In the hue H, red (R) has the highest frequency, and the others are almost averaged. The saturation S is
It is concentrated around 0.03. The lightness V has almost the same characteristics as the density distribution of RGB data.

(2) Quantization This is the operation performed by the quantizer 2 in FIG.
Since the data obtained by the HSV conversion has visual redundancy on each axis, first-stage compression can be performed by allocating the number of bits corresponding to the data and performing quantization. The quantization bit numbers are H, S, and V of 5, 2, and 7, respectively, so that the S / N ratio is saturated and no deterioration is visually perceived on the printed matter.
Good to have a bit. This is determined so that the sum of squared errors is minimized by the quantization of MAX.

When the original pixel value of the data obtained by the above HSV conversion is replaced with its quantized representative number, it becomes as shown in FIG. The quantization representative number is one of 32 (5 bit) kinds of colors in terms of H (hue), S
In terms of (saturation), it indicates which of the 4 (2 bits) stages, and regarding V (brightness), which of the 128 (7 bits) stages. In the example of FIG. 4, for example, H has a large amount of hue near R between R and Y, S has low saturation, and V
Also has a low brightness, which means that the portion of the target image has such a tendency of H, S, and V. As can be seen from this example, pixels in the vicinity have the same or similar numbers and form a group. Therefore, by utilizing this property, more efficient compression can be performed by a method such as run length. here,
To perform compression with an emphasis on image quality, complete restoration coding for returning to this stage is used.

(3) RICE Conversion (Variable Length Coding) In this embodiment, the RICE converter 3 shown in FIG. 1 performs the variable length coding. RICE encoding is perfect restoration encoding for compressing one-dimensional data (scan data, etc.). In this encoding,
The minimum unit of coding is called a block, and it is desirable that this data length J can be shortened. For that purpose, the characteristic of the original data is examined and the encoding is performed according to the characteristic. Here, the data length J = 21.

FIG. 5 shows an example of this characteristic, and shows that the frequency is higher as the difference between adjacent pixels is closer to zero. From this, the difference between pixels is obtained, and the shorter the absolute value of the difference, the shorter the code assigned. This is shown in the table below.

FIG. 6 shows an example of the structure of the data obtained as a result of this. In this format, the data is composed of four parts: ID, RF, variable code, and E. ID is the code number of this block data, RF is the reference data of the difference, and E is the block terminator.

FIG. 7 shows an example of a device configuration for creating such variable length data. This device includes a difference calculator 31, an FS generator 32, a code selector 33 and an 8-word coder.
It consists of 34.

The difference calculator 31 is composed of a subtraction unit,
For example, for one pixel block composed of 21 pixels, the difference between adjacent pixels is calculated. The calculation result is given to the FS generator 32.

The FS generator 32 is realized by one vinyl counter and a comparator, and assigns a variable length code as shown in Table 1 above according to the difference value from the difference calculator 31. This variable length code, that is, the code is given to the code selector 33.

The code selector 33 is composed of a shift register of 3J stages (3 × 21 = 63 stages in this example), and encodes the FS code from the FS generator 32. This encoding is performed by selecting one of the three modes according to the bit length F of the FS code as follows. This is as shown in Table 2 below. Then, this code is sent to the 8-word coder 34 or is output as it is.

The 8-word coder 34 is composed of a ROM having an appropriate table, and divides the bit string code from the code selector 33 into 3 bits and converts them. If the code string length is not divisible by 3, a dummy bit is added. Table 3 below shows this conversion operation.

What has been described so far is the data compression process. The compressed data is decompressed after being transmitted, for example, stored, communicated, etc., by the transmission means 4 in FIG. Next, the restoration operation will be described.

(4) RICE inverse conversion The decoding procedure by RICE inverse conversion is as follows, and basically, the RICE conversion procedure is followed in reverse. 1) Mode selection, 2) 3-bit data reproduction, 3) Differential data playback,
4) Each stage of reverse difference reproduction.

1) Mode selection This corresponds to the mode selection shown in Table 2 above and determines whether or not processing by an 8-word decoder is required.

2) 3-bit data reproduction This is performed when the mode of bit data reproduction by the 8-word decoder is selected as a result of 1) mode selection.

3) Reproduction of difference data Reproduction of difference data is performed using a table having a reverse relationship to the first table.

4) Reverse difference reproduction Based on the difference reference data in the block data, each difference value is added to this as the value of the first pixel in the block, and the H, S, and V values of each pixel are reproduced.

Of the restored data obtained in this way, V is originally
It is the maximum value of RGB. In terms of CMY, it can be said that it is the minimum value of CMY because CMY is the inverse of RGB (obtained by subtracting from 255). And the Sumi data in printing is basically CMY
Since it is a non-linear function of the minimum value of, the V-data can be passed through an appropriate look-up table (LUT) 8 to obtain sumi data K ′ as shown in FIG. The smidata has a desired gradation such as skeleton black and full black.

(5) Inverse HSV conversion This is performed by the inverse HSV converter 6 in FIG. 1 and is performed by the following method. That is, first, assuming that H (hue) has a spread of 360 ° and S (saturation) and V (brightness) are positive numbers of 1 or less, they are calculated as follows.

0 ≤ H ≤ 360, 0 ≤ S ≤ 1, 0 ≤ V ≤ 1, but all real numbers h '= H / 60 h "= integer part of h'f = h'-h" (fractional part of h') p = (1−S) × V q = (1−S × f) × V t = (1− (S × (1−f)) × V. Next, when h ″ = 0, (R, G , B) = (V, t, q) = 1 = (q, V, p) = 2 = (p, V, t) = 3 = (p, q, V) = 4 = (t, p, V) ) = 5 = (V, p, q) The RGB data thus obtained is converted into CMY data by the color reversing device 7 in FIG. 1 or subjected to normal scanner processing.

Fig. 8 shows the overall system from the import of RGB data from the target image to its compression and decompression, color correction, and final output. The part surrounded by the one-dot chain line is It is the device of FIG.

In this system, the RGB data is taken out by the image data capturing device 10 and sent to the compression system 20, and the data is compressed and given to the transmission system 30 as HSV data.
To the CMYK data, and the color corrector 50 corrects the color. Further gradation correction may be performed if necessary.

It goes without saying that the CMYK data in the above example may be replaced with RGB data.

〔The invention's effect〕

As described above, the present invention converts RGB data into H (hue), S
(Saturation) and V (brightness) data are converted, quantized and compressed, respectively, and converted into variable-length codes for transmission. Therefore, by decompressing this transmission data by a process reverse to that at the time of compression, The quality of the restored image can be improved and the amount of data can be minimized. In particular, when HSV data is replaced with a quantized representative value number, neighboring pixels tend to form a group of identical or close numbers, and this is converted to variable length code data, so complete restoration is possible with a small amount of data. Compression is performed. In addition, it is possible to form sumi version data using the V data in the restored HSV data.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an apparatus according to the present invention,
2 (a), (b), and (c) are RGB histograms of indoor photographed images, which are examples of images, and FIGS. 3 (a), (b), and
(C) is an HSV histogram of the same image, FIG. 4 is an explanatory diagram of the HSV distribution situation for a part of an image, and FIG.
Pixel difference histogram in HSV distribution in the figure, FIG. 6 is an explanatory diagram of data after compression in the apparatus of FIG. 1, FIG. 7 is a block diagram showing the configuration of the RICE converter in FIG. 1, and FIG. FIG. 2 is an explanatory diagram of a printing image data creating device including the device of FIG. 1.

Claims (2)

[Claims] Claim: What is claimed is: 1. Convert RGB data having a density for each color in pixel units extracted from an image to be printed into HSV data having hue, saturation and lightness, and add the HSV data to the sum of squared errors. Is quantized by calculating the quantized representative value number so as to minimize and replacing it with the quantized representative value number, and the quantized data is converted into variable length code data according to a predetermined conversion table and transmitted. , The method for compressing image data for printing. 2. RGB data having a density for each color extracted from an image to be printed in pixel units is converted into HSV data having hue, saturation, and lightness, and the HSV data is quantized as a representative value. Quantize by operating the number and replacing it with the quantized representative value number, convert this quantized data to variable length code data according to a predetermined conversion table, and use this inverse conversion table to convert this variable length code data. A method of compressing / decompressing image data for printing, wherein the HSV quantized data is inversely converted, the HSV quantized data is converted to RGB data, and the Sumi data is formed by the V data in the HSV quantized data.