JPH04342371A - Image processor - Google Patents

Abstract

PURPOSE:To present the image processor which can quantize a color image again with more fidelity and can correct the change of a recording characteristic. CONSTITUTION:Analog image signals in the three colors of R, G and B read by a CCD 1 are quantized to the digital signals of 8 bits by an A/D converter 2 respectively, the nonuniformity of the sensor 1 and an image forming system not shown in the figure and the filter characteristic are corrected by a shading correction part 3, the data are converted to an rgb luminance data and afterwards, they are converted to density data at a logarithm transformation part 4. In this case, cmy density data are limited within a color reproducing area by a gamut limit part 5 corresponding to the recording characteristic stored in a memory 8 and quantized again by a requantizing part 6, and requantizing error is corrected. Then, the data are converted to one-bit data of CMYK as recordable recording colors and recorded by a recorder 7.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus that requantizes color image data according to a plurality of sets of color data that can be expressed by a recording device.

0002

[Prior Art] Conventionally, this type of device has a sensor that uses RGB
Image data that has been color-separated and read is subjected to data conversion called masking/UCR processing into four-color data of CMYK, which are recording colors, taking into account recording characteristics. Further, pseudo intermediate processing is performed independently in each color space separated into four colors.

0003

[Problems to be Solved by the Invention] However, the above conventional example has the following drawbacks. ■When performing pseudo-halftone processing using a conditional dithering method, such as the error diffusion method, it is difficult to say that a single point on the recording paper necessarily represents the color of the input image.
Black dots slightly generated in the bare color, resulting from so-called black generation, create an unnatural texture and cause deterioration in image quality.

[0004] When the recording characteristics change, so-called masking/U
If an attempt is made to modify the CR coefficient, a huge amount of data processing is required, and this cannot be realized inexpensively on an actual machine. ■
When requantizing data after pseudo halftone processing is compressed, it is only possible to compress the requantized data independently for each of the four correlated colors.

The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide an image processing device that can requantize an input color image more faithfully and can compensate for fluctuations in recording characteristics. purpose.

[0006]

[Means to solve the problem] and

[Operation] In order to achieve the above object, the image processing apparatus of the present invention has the following configuration. That is, it has a requantization means for requantizing color image data according to a plurality of sets of color data that can be expressed by a recording device, and a correction means for correcting a requantization error generated by the requantization means. .

Preferably, the requantization means includes a calculation means for calculating a plurality of weighted average values of the pixel of interest according to the recording characteristics of the recording device, and converts the weighted average values obtained by the calculation means into the input data. The present invention is characterized by comprising a selective requantization means for selecting and requantizing the closest weighted average value.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of an image processing apparatus in this embodiment. first,
Analog image signals separated into three colors of R, G, and B by the CCD 1 and read are quantized by the A/D converter 2 into digital signals each having a width of 8 bits. Then, in the shading correction section 3, the unevenness and filter characteristics of the sensor 1 and the imaging system (not shown) are corrected, and after the data is converted into so-called NTSC/RGB luminance data, the density data c
', m', y'. When reading the recording characteristics, the memory 8 is configured to record one point with the actual recording device 7 described later, for example, red (R), green (
G), Blue (B), Cyan (C), Magenta (M),
8 of yellow (Y), black (K), white (W)
Recording Colors Each color is stored as r, g, and b data that has been logarithmically converted. Note that if the recording device 7 has the above-mentioned data in advance, the data may be input to the memory 8 from the recording device 7.

The density data converted according to the recording characteristic data described above is limited within the color reproduction area by the gamut limiting section 5. The gamut-limited c, m, and y data is input to the requantization unit 6, which determines the recording characteristic data described above, that is, which recording color out of the eight recordable colors is selected for one point of input image data. It is determined whether the recorded case is closest to the color tone of the input image. Here, 1-bit data for each of CMYK, which are recording colors, is decoded from the determined eight color data and recorded by the recording device 7.

[Description of Requantization Unit] Next, the detailed configuration of the above-mentioned requantization unit 6 will be explained with reference to FIG. 2. In the figure, the 3-bit data quantized by the selective requantization unit 62 is decoded into 1-bit data for each of CMYK by the recording signal conversion unit 64, and based on the 1-bit data for each of CMY, a weighted average is calculated in the vicinity of the pixel of interest. The value is the average value calculation unit 61
is required. Note that the signals are correlated according to Table 1.

[0011]

[Table 1]

Eight types of weighted average values are obtained in order to predict the requantization of the pixel of interest into eight colors. Therefore, the selective requantization unit 62 finishes the requantization by selecting the predicted value closest to the input data, but at this time, the difference between the selected predicted average value and the input data is calculated in each cmy space. The error correction section 63 corrects the difference, that is, the requantization error. The above-mentioned weighted average value is calculated by replacing a plurality of data that have already been requantized in eight ways with cmy multilevel data obtained from the memory 8 and obtained by the actual recording device 7.

Next, a more detailed explanation will be given with reference to FIG. In the figure, the memory 610 is a FiFo that delays image data requantized to 3 bits by one line, and the data at its input and output terminals are stored in independently provided c, m, and y
It is connected to three color weighted average calculation units 614a to 614c. In other words, the requantized data is stored in the RAMs 615a to 615c as colors c, m, and c, respectively, obtained by the actual recording device 7.
After being converted to y multi-value data, for example, in the case of cyan, F
/F611a-c are used to further delay and hold several pixels, and a weighted average value based on requantized data for four pixels adjacent to the pixel of interest is obtained by weighted average calculation units 614a-c. Here, Table 2 shows the recording characteristics used in this example.

[0014]

[Table 2]

[0015] Also, the weights for determining the average value are shown in Fig. 4.
As shown in (a), A adjacent to the position * of the pixel of interest
~D and the pixel of interest, as shown in the same figure (b), 1/16
~6/16 value. For cyan, F/
F611c output is A, F/F611a input is B, F/F
The 611a output corresponds to position C, and the F/F 611b output corresponds to position D. Therefore, the weighted average value calculation unit 614a performs a so-called product-sum calculation using the weighting coefficients shown in FIG. do.

[0016]

[Table 3]

For example, if the requantization results of ABCD are cyan, white, yellow, and white, the recording characteristics shown in Table 2 become Table 3, and the weighted average value calculation unit 614
From a to c, the average value mc, mm for 4 pixels is shown in the following formula.
, my are obtained, respectively. mc = (4×206+1×15+4×19+1×
15) x 1/16 = 58 mm = (4 x 91 + 1 x
15+4×16+1×15)×1/16=29 my
=(4×37+1×15+4×196+1×15)×
1/16=105 Next, the selective requantization unit 62 calculates the eight types of predicted average values Mc shown in Table 4 based on the above-mentioned average values mc , mm , my and the weighting coefficient 6/16 of the pixel of interest.
, Mm, My are obtained.

[0018]

[Table 4]

As an example, when the pixel of interest is predicted to be cyan, the following equation is obtained from the data in Table 2, and the predicted data in Table 4 is similarly obtained. Mc=mc+206×6/16=135Mm=m
m +91×6/16=63My=my+37×6
/16=119 Then, the selective requantization unit 62 compares the input data (more precisely, the error-corrected input data) with the eight types of data shown in Table 4. For the comparison, when the input data are Ic, Im, and Iy, the distance Δd between the eight types of predicted average values and the input data is determined based on the following equation, and the requantized color exhibiting the minimum value is determined.

Δd=(Ic−Mc)2+(Im−M
m )2 + (Iy - My )2 where Ic =
81, Im = 100, Iy = 140, then
Obviously, that of magenta will be the minimum, thus requantizing the pixel of interest to magenta. By the way, there is a finite difference between the predicted average value of magenta and the input data,
The differences Ec, Em, and Ey are output to the distributor 634 as requantization errors. In the above case, the following equation is obtained.

Ec =81-79=2 Em =100-112=
-2Ey = 140-134 = 6 This requantization error is divided into two by the distributor 634, one is input to the line delay memory 630 and delayed by one line, and the other is read out from the same memory 630. That is, after the adder 631 adds the error that occurred one line before, the F/F 632 synchronizes with the input timing of the next input data, and then corrects the next pixel data. It goes without saying that the error correction section 63 described above processes each color of cyan, magenta, and yellow.

By sequentially executing the above series of processes for each input pixel data, requantization of all image data is completed. [Modified example of requantization unit] Next, a case where the requantization unit 6 is implemented using the error diffusion method will be described in detail with reference to FIG. 5.

As shown in FIG. 5, the selective requantization unit 162
inputs the input cmy data after re-quantization error correction, which will be described later, and compares it with the recording characteristic data shown in Table 2 input from the recording characteristic data memory 8 shown in FIG. Choose from 8 types. For example, as mentioned above, if the input data is Ic = 81, Im
= 100, Iy = 140, when the pixel of interest is requantized and predicted as cyan, the distance △d from cyan is
△d=(206-81)2 +(91-100)
2 + (37-140)2 = 263
15 Similarly, a comparison is made with the predictions made for the other seven species, and it is determined that red has the smallest distance. The requantization result is decoded by the recording signal converter 64 and output to the recording device as CMYK binary data, as in the case described above.

On the other hand, the requantization errors are as follows, and are input to error distribution units 163 to 165, and each error distribution unit 163 to 165 distributes them to four adjacent pixels at the distribution ratio shown in the figure. error memory 166-168
is stored in Ec =81-60=21 Em =100-219=-119 Ey =140-190=-50 Here, for the next pixel, after reading out the distribution error and the previously distributed error from the memory and integrating them, Adder 1
Input image data is corrected in steps 70-173.

By executing the above processing for each pixel,
Similar to the above-described example, it is possible to perform requantization with color reproduction characteristics that are faithful to the colors of the input image. [Gamut Limit] In the process described above, if the input image data exceeds the color reproduction range due to the recording characteristics, for example, Ic
If data such as =Im =Iy =255 is input continuously, the predicted average value will always show a value less than that, so
Any requantization error that occurs is always added to the input data as a positive value. As a result, the requantization error becomes a larger positive value, causing the processing to diverge. Therefore, the gamut limiting section 5 in this embodiment is configured for the purpose of preventing processing divergence.

FIG. 5 is a diagram showing the configuration of the gamut limiting section 5 in this embodiment. In FIG. 5, RAMs 50 to 52 for each color are so-called look-up tables (L
UT), and the data is converted according to the following equation. Taking cyan as an example, when C'≧Cmax, C=CmaxC'≦Cmi
When n, C=Cmin and other cases
C=C' Here, Cmax and Cmin are the maximum value 210 and minimum value 15 of the cyan data in Table 2, respectively. The same applies to magenta and yellow.

[Modified example of gamut restriction] In the above embodiment, data exceeding the reproduction color space is clamped at Cmax and Cmin, but as shown in the following equation, input data 0 to 255 is clamped within Cmin to Cmax. It may be mapped to C=Cmin+(Cmax-Cmin)/255×C'
Further, as non-linear transformation, embodiments using various algorithms such as the following equations can be easily realized using FIG.

C=f(Cmin, Cmax, C')Although this is done independently for each color, in order to do it more accurately, For a solid that connects eight points, any point that falls outside the solid may be detected and mapped to the closest surface of the solid. Specifically, this can be achieved by using a RAM that inputs c'm'y' 24-bit data to its address terminal, and converting it based on the look-up table data in the RAM.

[Storage of Recording Characteristic Data] As explained above, according to this embodiment, recording is performed by requantizing while correcting errors based on the tint data recorded by the actual recording device 7. It is also possible to easily respond to changes in device characteristics. In other words, when requantizing into binary data, the six colors of cyan, magenta, and yellow are combined with black and white, where white is the color data of the recording paper used, and is recorded as a color pattern. This can be achieved by reading the pattern with a sensor and storing it in the data memory 8.

[0030]

[Other Embodiments] In the above-mentioned embodiments, the recording characteristic data of a monochromatic black recording material was used for black recording, but in order to further improve the color reproduction of input image data, Kc, K Ky due to and Y, Km due to K and M 3
It is also possible to requantize into 11 colors by adding different color data and using recording characteristic data of 11 colors. Furthermore, even if black data Kcmy based on CMY is added and requantized to 12 colors, color reproduction is further improved. In this case, the decoding to the recording signal is based on Table 5.

[0031]

[Table 5]

Furthermore, although the above-mentioned embodiment has been described as an example in which requantization is performed into binary data, it can be implemented in the same manner in the case where, for example, it is possible to record in each color in 3-value to 16-value data. In other words, Figure 3
In order to extend the embodiment to three values, set the sum of the weighting coefficients during the average value calculation from 1 to 1/2, select the requantization level from 0, 1, and 2 for each color, and There are 27 types of combinations, and the prediction of the pixel of interest can also be selected from among the 27 types. As in this embodiment, by increasing the number of requantization levels by increasing the number of multilevel levels, more faithful requantization of the input image is possible in order to further reduce the requantization error.

The present invention may be applied to a system made up of a plurality of devices, or to a device made up of one device. It goes without saying that the present invention can also be applied to a case where the present invention is achieved by supplying a program to a system or device.

[0034]

[Effects of the Invention] As explained above, according to the present invention,
The following effects can be obtained. ■By requantizing using color data from a recording device that can actually be expressed, requantization that is more faithful to the input image becomes possible. (2) By directly using the characteristics of the recording device for requantization, it is possible to requantize faithfully to the input image even if the characteristics of the recording device vary.

■Depending on the expressible color characteristics of the recording device,
By controlling the input image data, divergence in the requantization process can be prevented.

[Brief explanation of drawings]

FIG. 1 is a schematic block diagram showing the configuration of an image processing device in this embodiment.

FIG. 2 is a block diagram showing the configuration of a requantization section in this embodiment.

FIG. 3 is a diagram showing a detailed configuration of a requantization section in this embodiment.

FIG. 4(a) is a diagram showing the positions of a pixel of interest and adjacent pixels, and FIG. 4(b) is a diagram showing weights for determining the average value of the pixel of interest.

FIG. 5 is a diagram showing the configuration of a gamut restriction section in this embodiment.

FIG. 6 is a diagram showing a configuration when a requantization section is implemented using an error diffusion method.

Claims (2)

[Claims] 1. Requantization means for requantizing color image data according to a plurality of sets of color data that can be expressed by a recording device, and correction means for correcting requantization errors generated by the requantization means. An image processing device comprising: 2. The requantization means includes a calculation means for calculating a plurality of weighted average values of the pixel of interest according to the recording characteristics of the recording device, and a calculation means for calculating a plurality of weighted average values of the pixel of interest according to the recording characteristics of the recording device, and a calculation means for determining the weighted average value closest to the input data from each weighted average value calculated by the calculation means. 2. The image processing apparatus according to claim 1, further comprising selective requantization means for selecting and requantizing the weighted average value.