JPS6038972A - Input display device - Google Patents

Abstract

PURPOSE:To attain picture output display with excellent color reproduction by applying nonlinear processing to a luminance signal and a color difference signal and converting especially the luminance signal along a curve close to logarithmic conversion so as not to be deformed by a high density thereby attaining stable identification even if a picture of a color with high density is inputted. CONSTITUTION:When a color picture is converted into an electric signal by an image pickup element and outputted to a color printer or the like, a signal after coordinate conversion outputted from product sum operating circuits 16, 16' and 16'' is inputted to averaging processing circuits 18, 18' and 18''. Then as to the luminance signal, a subtractor 19 subtracts the averaged signal and the original signal. As a result, the signal gives local fluctuation of the luminance signal. Further, this signal is inputted to a memory 20 for deciding binary dither. The relation between an input address and an output data exists in a shape converting memory 30 as shown in the figure, which density information is expanded at a region corresponding to a high density of the input address. Thus, the color reproduction is attained stably without deformation even at the high density region.

Description

[Detailed Description of the Invention] [Field of Technology to which the Invention Pertains] This invention relates to converting a color image into an electrical signal using an image sensor.
7. This relates to a device for outputting an image to a color display device using this electric 4it, and particularly relates to an input display device that can obtain a good raw image even when a high-density image is input as an input image.

[Technical background of the invention and its problems]

Conventionally, when converting a color image into an electrical signal using an image sensor and outputting it to a color printer or the like, it was often impossible to distinguish high-density colors, and all the signals were converted to black and output.

In particular, when reading data using multiple sensors, it is almost impossible to immediately logarithmically transform the sensor outputs, considering the need to correct the characteristics of the multiple sensors. In this case, the processing is often done digitally, and the high-density color is almost indistinguishable and must be displayed as black.

[Purpose of the invention]

The present invention has been made to eliminate the above-mentioned drawbacks, and is an input method that can stably identify even when a high-density color image is input and output an image with good color reproduction. The purpose is to provide equipment.

[Summary of the invention]

In the present invention, a four-power color image is converted into a negative signal by an image sensor, and this electrical signal is separated and converted into a luminance signal and a color difference signal or a signal equivalent thereto by a matrix circuit consisting of a subtractive calculation circuit. . At this time, the coefficients of the matrix are determined so that the converted signals are uniformly distributed over as wide a range as possible in the color difference space. Next, even if noise is superimposed, averaging processing is performed so that these signals have an SN above a certain level. In this case, it is generally necessary to average more color difference signals. The difference between the averaged intensity signal and the original luminance signal is taken and mixed to obtain a signal with an appropriate signal-to-noise ratio and distortion.

Nonlinear processing is performed on these luminance signals and color difference signals. In particular, the luminance signal is converted using a curve close to logarithmic conversion so as not to be distorted by the density. These converted signals are used to draw a color conversion table, determine the ink information, and send the signal to a color printer for full color display.

〔Effect of the invention〕

In the present invention, a matrix circuit separates brightness signals into brightness signals, color difference signals, or signals corresponding to them, and converts the signals so that they are distributed in the widest possible range (pIj) in the color difference space. It is converted so that the hue can be expressed.Then, the color difference number 46 is generally multiplied by a large coefficient, so the noise component is also emphasized.Therefore, more averaging is performed on this color difference (M number), S of color difference signal
Improve N.

By doing this, the hue can be determined without being affected much even if there is noise.

Next, the luminance signal is transformed using a curve close to logarithmic transformation so as not to be distorted during installation. Color transformation daples are drawn using these transformed signals to determine the interscape.

Therefore, stable color reproduction is possible without distortion even at high (Ir'j) degrees.Also, since a color conversion table is drawn using the luminance signal and this averaged color difference signal,
Since the luminance signal is modulated to a high resolution, it is possible to achieve finer details.

[Embodiments of the invention]

An embodiment of the present invention will be described with reference to the drawings. Figure 1 shows a CCD line sensor 3 with four space-divided noise color filters (for example, Lecture No. 1243 "High-speed color close-contact scanner" at the 1981 Institute of Electronics and Communication Engineers General Conference National Conference), which uses a SELFOC lens l to scan a manuscript into four space-divided noise color filters.
(written by Hosaka). ), the image is formed on an image sensor, photoelectrically converted, and this signal is used to send an electric signal to a color printer 29 for color display. The output magnetic signal of the CCD line sensor 3 is amplified by an amplifier 4,
The AD converter 4 converts it into a digital signal. The signal is divided into three colors, for example, white, yellow, and cyan, corresponding to each color filter, and is stored in the line memory 6.7.
．． 8 are recorded individually. Each CCD from 3 to 5゛
The elements 3, 3, and 3.3 operate in the same manner, are recorded in the line memories 6.7.8, and output as one line of signals.

Here, it is desirable that the resolution of the image formed by the SELFOC lens is about 1/2 of the resolution of the CCD line sensor.

Now, the signal of one line for each name color is then averaged every two pixels by the register 9 and the adder 10. That is, one pixel is recorded in the register 9, and the next data is added to the register 9 at the next time, and the connection of the output of the adder is shifted by 1 bit to the MSB side. In this way, two-pixel averaging is performed to eliminate aliasing noise due to zangling.

Next, normalization is performed to correct shedding for the signals of each pixel of each sensor. Now, when the white level of the standard test pattern is 7 (with M number as hi and black level signal as Il+) and the output of this 1 line signal as I, the standardized output signal IO is as follows: .

This is done for each pixel (this is done by recording the IW in the line memory 11 in advance, and subtracting the subtractor element 12 sequentially based on the output signal from the adder 10 at each pixel). .Then 1/(IB-
The values of IW) are recorded in advance in the line memory 13, and the multiplier 14 sequentially multiplies them i:f: for each pixel. The same process is performed for other color signals as well.

By performing such processing, even if shading occurs in the CCD sensor, an output signal with uniform brightness can be obtained. However, if there is a change in the characteristics of the color filter, it may be output as a subtle color change.

Therefore, the next step was to create a matrix circuit consisting of a product-sum calculation circuit so that it could respond to subtle color changes between each image sensor, and to reduce the memory capacity of the color conversion table. Coordinate transformation is performed so that the coordinates are evenly distributed in the address space of the memory of the color conversion table.

The color filter currently on the CCD sensor is white.

The three colors are yellow and cyan, and the output signals from these are w, y, and c, respectively. Further, the luminance signal is S], the first color difference signal is 82, and the second color difference signal is 83, and each is defined as shown in the following equation. Note that the brightness signal is 5
It may also be 1-W.

Next, coordinate transformation is performed using the following equation. If the converted coordinates are XI + X2 and xa, respectively, the following equation is obtained.

Here, the coordinates after the conversion are determined so as to be uniformly distributed on the color conversion table address. For example, make the distance equal to each color of yellow, cyan, and magenta. At this time, to determine the matrix aij in equation (3), S]+82+83 is measured for each color, and the converted coordinates XI, X2, and xa are determined for the values. Then, aij can be determined by calculating the inverse matrix from these values. Therefore, St+ for each color is calculated for each sensor.
Even if 82+83 are different, the coordinates XI, X after transformation
21X3 can be made to be the same, and the aij at that time can be adjusted. Shinokakatte (2)
, From (3), the following equation is obtained.

The signal flow for the conversion equation created in this way is as follows.

That is, if the normalized W, Y, and C signals are selected by the Marble Grexer 15, and each is multiplied by the coefficient of equation (4) by the product-sum calculation circuit 16 and added, Xi +X2, xa is obtained. circle. At this time, if the coefficients of equation (4) are recorded in advance in the memory 17 for each image sensor, and the coefficients are switched and calculated for each image sensor, variations between the elements can be eliminated. Note that in a CCD sensor, there may be variations from pixel to pixel, and in this case, by changing the coefficient for each pixel, it is possible to suppress the variation in hue from pixel to pixel.

Furthermore, if the coefficient aij is determined so that the coordinate-converted addresses are distributed almost equally on the coordinate space for each color, the memory capacity required for the color conversion table may be small.

Next, the signal whose coordinates have been transformed in this manner is subjected to an averaging process, and after non-linear transformation is performed on the averaged signal, a color conversion table is drawn to determine the amount of can ink. In this embodiment, the determined ink amount value is then binarized using a yen threshold only when the local fluctuation rate of the luminance signal is larger than a predetermined value, and dithered in other cases. ing. These will be explained with reference to FIG.

The signals after coordinate transformation output from the product-sum calculation circuits 16.16' and IJ' are sent to the averaging processing circuits 18.18' and 1g''.
is input. Next, for the luminance signal, the subtracter 19 calculates the difference between the averaged signal and the original signal. As a result, the flaws ′ and ′ give local fluctuations in the luminance signal. Furthermore, this signal is input to the binary dither foot memory 20. The contents of this memory are as shown in Table 1, for example.

Table 1: The range with small local fluctuations, i.e. from -64 to 63, contains codes for dithering and binarizing, while other ranges contain codes for binarizing with a fixed threshold. recorded.

On the other hand, this difference signal and the averaged signal are input to the multiplexer 21, appropriate coefficients are set in the memory 23 for both signals, and sent to the product-sum calculation circuit 22.
Mix both signals. By doing this, it becomes possible to obtain an appropriate signal that is not too blurry. For example, if the signal before averaging is x1 and the signal after averaging is x, the signal from the subtracter 19 will be x - x, and the output of the averaging circuit 18 will be x. So memory 23
If the coefficients of are shifted by a and b, the output y of the product-sum calculation circuit 22 becomes the following equation.

y = a (x - x) + bx (5) where a =
If b=1, y=x, which is the signal before averaging. Furthermore, if a=o and b=1, y=also, which results in an averaged signal. Furthermore, a=1. If b=0, y
== x −x, and it becomes a difference signal, that is, a signal of only local vibrations, or in other words, it becomes something close to a differential waveform. This becomes a low frequency cut signal. In this way, coefficients a, b
By changing , it is possible to freely obtain from an averaged high frequency cut signal to a low frequency cut signal close to the differential waveform. Therefore, coefficients a and b are appropriately selected, and a signal that is not too blurry is input to the nonlinear conversion memory 30, taking into account the SN of the signal. Nonlinear conversion memory 30
In this case, the relationship between the input address and the output data is as shown in FIG. 3, and the relationship is such that the density information is stretched out in the area corresponding to the high density of the input address. By doing this, stable color reproduction is possible without being distorted even in the υ high density region. This signal is input to the color conversion table memories 24, 24', 24''.

In addition, two other color difference signals X2. For X3, the averaging circuits 18' and 18' perform averaging to improve the SN and input the data to the color invasion table memories 24, 247, and 24''.

The address of the color conversion table memory is determined by these three signals, and the amount of ink actually required is determined according to the address. As this color conversion table, for example, the well-known Neugebauer equation may be calculated in advance and recorded in the memories 24, 24', 24''.

In this way, the amounts of yellow and magenta cyan ink actually required are calculated from the memories 24, 24', 24'', and 2f consists of a comparator circuit, and the signals to be compared are multi-blends, 6, 26', 26 ``K'' switches the contents of the dither pattern reference data memories 27, 27', 27' and the fixed range value memories 28, 28', 28'/. The signal necessary for this switching is output from the binary dither publication memory 20. That is, when the local fluctuation rate of the luminance signal Xl is larger than a predetermined value, binarization using a fixed value is selected, and in other cases, dithering is selected. By doing this, it becomes possible to express objects such as character patterns whose luminance values are drastically changed with high resolution by fixed binarization. Furthermore, in a halftone image, the luminance signal does not change significantly, making it possible to express dithered halftones.

The binary signal is sent to a color printer 29 capable of binary expression, such as a thermal transfer printer, to express a full color image.

[Other embodiments of the invention]

In the previous embodiment, three colors of yellow, magenta, and cyan were explained, but the same process can be performed for four colors including black. That is, as in the circuit shown in FIG. 4, a memory 24'' for generating a black signal is provided as a memory for color conversion.Other aspects are the same as in the previous embodiment.

Furthermore, in the previous embodiment, the difference signal is used as it is as an input signal at the input of the binary dither switching memory 20, but in this case, continuity is lost in the contents of the memory. Therefore, instead of inputting it as is, if one bit is added to M'ISB and the result is set as the address of the memory 20, continuity will occur in the contents of the memory.

Furthermore, if the dither reference pattern and fixed threshold value are made common to each color, the circuit as shown in FIG. 5 can be omitted.

Furthermore, in FIG. 6, nonlinear conversion is also performed on the color difference signals. Depending on the application, the memory may still be 24.24
', 24/' may be omitted.

Further, in the previous embodiment, an example of a one-dimensional COD sensor was described as an image sensor, but a two-dimensional sensor may be used. Furthermore, the color filter may be yellow, cyan, or green, or may be red, clean, or blue.

[Brief explanation of drawings]

FIG. 1 and FIG. 2 are diagrams showing one embodiment of the present invention. Figure 3 is a diagram explaining the nonlinear transformation, Figure 4, Figure 5,
FIG. 6 is a diagram showing another embodiment. 181. SELFOC lens array, 2...manuscript, 3
...COD sensor, 4...amplifier, 5...A/
Di converter, 6, 7.8... line memory, 9.19.
...Subtractor, 10...Adder, 11, 13. i7・
... Coefficient memory, 14... Multiplier, 16.22...
Product-sum calculation circuit, 20... Binary dither switching data memory, 24... Color conversion table memory, 27... Dithering turn reference memory, 2B... Fixed binary threshold value, 29
...Color printer, 30.31,32...Nonlinear conversion memory. Agent: Patent attorney Noriyuki Chika (and 1 other person) Figure 1 Figure 2 Figure 3 Is it a person with a cuff? A4 diagram Figure 5

Claims (2)

[Claims] (1) In a device that converts a color image into an electric signal using a color image sensor and outputs the image to a color display device using this electric signal, the signal obtained from the image sensor is converted into a luminance signal and a color difference signal or their equivalents. After separating and converting them into signals, averaging processing is performed on one or both of them so that the SN of both is approximately equal to a certain level or higher, and after further non-linear conversion is performed, a color conversion table is created. An input display device characterized in that signal processing is performed by subtracting a signal. (2) The input display device according to claim 1, wherein non-linear conversion is performed only on a luminance signal or a signal equivalent thereto.