JPH0683379B2 - Image reader - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an image reading device for reading color information of an image other than black and white.

[Prior Art] When an image is input to a facsimile device, an image processing device, or the like, it is often convenient if the color information of the image can be identified.

For example, in the case where a transmission document transmitted by a facsimile machine has a red letter, a seal, or an underline in red, if this red information can be output, more effective image information transmission becomes possible.

By the way, there have been proposed many methods for discriminating colors in this way, but most of them are for separating one color by one filter or the like.

However, in such a conventional method, for example, in consideration of the case where a red filter is used to separate light and an image is formed on a line image sensor in order to determine red, a color including a red component to some extent, for example, a magenta image is read. Sometimes the output level of the line image sensor exceeds a predetermined threshold level and is discriminated as red. Therefore, the color other than the color visually discriminated as red is also discriminated as red. there were.

Further, the background color of a general document is white, and therefore, the output level of the line image sensor when the background is read becomes the highest. Therefore, when the red image signal is formed only from the received light signal of the red component as described above,
Since the output level of the line image sensor when reading the red part is close to the level of the white part, the part where the red color of the image is unclear or faint is judged to be white instead of being judged to be red. There were some inconveniences.

Furthermore, the color range that is visually distinguished as red is wide,
This visual characteristic is different from the spectral characteristic when the light is dispersed by a single filter. Therefore, the color visually discriminated as red is not discriminated as red by the device, and an appropriate red image signal cannot be output. There was also.

[Purpose] The present invention has been made in order to eliminate the above-mentioned inconveniences of the related art, and a predetermined color component other than the black and white component
An object of the present invention is to provide an image reading device that can reliably read.

[Structure] In order to achieve this object, the present invention separates an image into a discriminant color and its complementary color, photoelectrically converts it by a line image sensor, and compares the output signals of these line image sensors for each pixel. It is determined whether or not the pixel is of a discrimination color, and for the pixel for which the discrimination is made, a color corresponding to the discrimination color is obtained by performing a predetermined calculation on the image signal of the discrimination color and the image signal of the complementary color. An image signal is formed, the other pixels are judged to be black and white pixels, and the complementary color image signal is output as a black and white image signal. The black pixels that appear pseudo around the discrimination color when it is positioned are corrected to white pixels to improve the image quality.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows an optical system of an image reading apparatus according to an embodiment of the present invention. In this embodiment, red is the discrimination color.

In FIG. 1, an image on a read line of a read document 1 is illuminated by a light source 2, image light reflected from the read line is reflected by a mirror 3 and guided to a lens 4, and further, a half mirror or a prism. Luminous flux splitting device 5
Is branched into two by.

The image light branched in this way is spectrally divided through a red filter 6 that transmits a red component and a cyan filter 7 that transmits a cyan component, and then imaged on the line image sensors 8 and 9.

The red filter 6 has a characteristic of transmitting light having a wavelength of, for example, 580 nm as a transmission limit wavelength, and has a characteristic of transmitting light having a wavelength longer than that. Further, the cyan filter 7 has a transmission limit wavelength of, for example, 620 nm, and transmits light having a wavelength shorter than that. It has the property of transmitting light.

When the image is split into the red component and the cyan component which is the complementary color to the red component in this way, all the colors are divided into these two components and appear as predetermined levels in the output signals of the line image sensors 8 and 9.

Then, when the levels of the red component and the cyan component of the color samples of commercially available red writing utensils and the like were measured, it was found that these color samples were located in the shaded area in FIG. In FIG. 2, the output level of the line image sensors 8 and 9 when the white portion of the background of the document having the highest light receiving level is read is set to 1, and the line image sensor 8 when the actual image is read is set. The output levels of 9 are normalized.

That is, the red color appearing in the original image satisfies the relationship represented by the following formula.

Kc / Kr <α, Kr> β where Kc is the normalized cyan component level (that is, the output level of the line image sensor 9), and Kr is the normalized red component level (that is, the level of the line image sensor 8). ). Also, according to the actual measurement, α = 0.4,
β = 0.4. It should be noted that this value is obtained when the red filter 6 and the cyan filter 7 having the above-mentioned characteristics are used, and when the filters having different characteristics are used, this numerical value may change.

From the above, the line image sensor 8, 9
It is possible to determine whether or not the output levels of the above (normalized ones) satisfy the relationship of the above expression, and to determine the pixel satisfying the relationship as a red pixel (that is, a red pixel).

The output signal from the line image sensor 9 may be used as it is as the monochrome image signal for determining the monochrome pixels other than the red pixels.

By the way, when the red portion is singly located on the white background, as shown in FIG. 3, in the portion where the image changes from white to red and from red to white, the line image receiving the cyan component is received. The light receiving level of the sensor 9 changes smoothly.

Therefore, for example, when three red pixels are located between white pixels, the output levels (normalized) of the line image sensors 8 and 9 at this time are as shown in FIG.
It changes as shown in (b).

That is, the output level of the line image sensor 8 is such that the level of the red pixel is slightly smaller than the level of the white pixel, and the output level of the line image sensor 9 is the level of the red pixel similar to that of the black pixel. , The level of the white pixel adjacent thereto becomes smaller than the levels of the other white pixels, and further becomes smaller than the threshold level SL for identifying black and white.

In this case, the determination state of the red pixel based on the above-described determination reference is as shown in FIG. 7B, and thus the red image signal RD representing red and non-red is as shown in FIG. The black-and-white image signal BW representing black and white is as shown in FIG.

As a result, a black pixel that does not exist in the actual image, that is, a pseudo black pixel, is output as a read image signal by one pixel adjacent to the red image. Therefore, when the read image is recorded, the recorded image is inferior.

In order to eliminate such an inconvenience, when there is one pixel which is determined as a black pixel adjacent to the pixel which is determined as a red pixel, the black pixel is determined as a pseudo black pixel, and the pseudo black pixel is determined. It suffices to execute a correction process for correcting the white pixel into a white pixel.

FIG. 5 implements the principles of the invention described above. 1 illustrates an image reading apparatus according to an embodiment of the present invention. In the figure, the line image sensor 8, which is not directly related to the present invention,
The drive control system and the control system such as the sub-scanning mechanism of 9 are omitted.

In the figure, the image signals P1 and P2 output from the line image sensors 8 and 9 are output to the white waveform memory by the switches 11 and 12.
It is added to either 13, 14 or the normalization circuit 15, 16.

The switches 11, 12 are switched by a control means (not shown) so as to select the white waveform memories 13, 14 when reading a reference white image (for example, a white original). As a result, the image signals P1 and P2 output from the line image sensors 8 and 9 at this time are stored in the white waveform memories 13 and 14.

Further, the switching devices 11 and 12 are switched by the control means so as to select the normalization circuits 15 and 16 when the image of the read document 1 is read.

The normalization circuits 15 and 16 perform level conversion on the basis of the image signals P1 and P2 obtained when the read document 1 is read and the signals added from the white waveform memories 13 and 14 in synchronization with the image signals P1 and P2. In addition, it has a function of converting into a corresponding digital signal, and the normalization circuits 15 and 16 absorb the difference in transmittance between the red filter 6 and the cyan filter 7.

That is, the switch 11, the white waveform memory 13 and the normalization circuit
Image signal P1 output from the line image sensor 8 by 15
Shading correction circuit SD1 for shading correction
In addition, the switch 12, the white waveform memory 14 and the normalization circuit
Image signal P2 output from the line image sensor 9 by 16
Shading correction circuit SD2 for shading correction
Are formed respectively.

The output signals R and C of the normalization circuits 15 and 16 are
Added to 20.

The color calculation circuit 20 executes the processing shown in FIG. 6 to determine the color of each pixel and form the red image signal R1 and the monochrome image signal B1.

That is, first, the ratio K of the input signals R and C is calculated (process 101), and it is determined whether or not the ratio K is larger than a predetermined value α (decision 102).

If the result of this judgment 102 is YES, then it is judged whether or not the signal R is larger than a predetermined value β (decision 103). The predetermined values α and β in this case are both 0.4 as described above.

When the result of this determination 103 is YES, since the pixel is a red pixel, the color calculation circuit 20 calculates the red image signal R1 based on the following equation (process 104).

R1 = a.R + b.C Here, R and C are equal to the signals R and C, respectively, and a and b are constants having different signs. The constants a and b
Is set in consideration of white balance and image enhancement. In this way, the color calculation circuit 20 calculates the red image signal R1 by multiplying it by an appropriate coefficient.
The level of R1 can be expanded, and as a result, various red colors can be reliably output.

Further, a value DW representing white is set to the monochrome image signal B1 corresponding to this pixel (process 105), and the red image signal R1 and the monochrome image signal B1 calculated at this time are output.

When the result of decision 102 is NO, and when the result of decision 103 is NO, the pixel is other than a red pixel, that is,
Since it is either a white pixel or a black pixel, a value NR representing non-red is set in the red image signal R1 (process 106),
The signal C is set to the black-and-white image signal B1 (process 107) and is output respectively.

In this way, the red component and the monochrome component of the image on the read document 1 are converted into the red image signal R1 and the monochrome image signal B1 and output.

The red image signal R1 output from the color calculation circuit 20 is a latch circuit.
21 and one input terminal of the NAND circuit 22, the black-and-white image signal B1 is applied to one input terminal of the latch circuit 23 and the NAND circuit 24.

The output R2 of the latch circuit 21 is output to the outside as a red image signal RD, added to the latch circuit 25, and inverted by the inverter 26 into the signal R3, and then the NAND circuit.
It is added to one of the input terminals of 22,24.

The output B2 of the latch circuit 23 is applied to the latch circuit 27 and also to one input terminal of the OR circuit 28, and after being inverted into the signal B3 via the inverter 29, the NAND circuit 2 is output.
It is added to one input end of 2,24.

The output R4 of the latch circuit 25 is applied to one input end of the NAND circuit 24, and the output B4 of the latch circuit 27 is applied to one input end of the NAND circuit 22.

The NAND circuit 22 detects and corrects a pseudo black pixel in a portion that changes from white to red, and its output S2 is a NAND circuit.
It has been added to one input end of 30. Also, the NAND circuit 24
Is to detect and correct a pseudo black pixel in a portion that changes from red to white, and its output S1 is applied to the other input terminal of the NAND circuit 30.

The output C1 of the NAND circuit 30 is added to the other input terminal of the OR circuit 28, and the output of the OR circuit 28 is the black and white image signal.
Output as BW to the outside.

Now, as shown in FIGS. 7A and 7B, two white pixels, one black pixel, three red pixels, one black pixel, and four white pixels are successively arranged. Consider what appears in signals R and C. At this time, as described above, the black pixel sandwiched between the white pixel and the red pixel is a pseudo black pixel.

Therefore, the output R1 of the latch circuit 22, the output R3 of the inverter 26, and the output R4 of the latch circuit 25 are shown in FIG.
The output B2 of the latch circuit 23, the output B3 of the inverter 29, and the output B4 of the latch circuit 27 change as shown in (d) and (e).
Respectively change as shown in FIGS. 4 (f), (g) and (h).

Observing the state change of the signals R1, B3, R3, B4 applied to the NAND circuit 22, the output signal S2 is shown in FIG. 4 (j) because the signal R1 is the logical level L up to the third pixel. As described above, the logic level becomes H, and all the inputs become the logic level H at the fourth pixel, so that the output signal S2 falls to the logic level L, which causes the output C1 of the NAND circuit 30 (the fourth level).
(See Fig. (K)) rises to the logic level H.
The black pixel (logic level L) of the third pixel of B1 is converted into a white pixel (logic level H), and the pseudo black pixel generated in the portion changing from white to red is removed ((1) signal in FIG. 4). See BW).

After that, since one or more of the inputs of the NAND circuit 22 are at the logical level L, the output signal S2 of the NAND circuit 22 maintains the state of the logical level H.

Observing the state changes of the signals B1, R3, R4, B3 applied to the NAND circuit 24, the signal R4 is at the logical level L up to the fifth pixel.
At the 6th and 7th pixels, the signal B3 is at the logic level L.
The output signal S1 is at the logic level H as shown in FIG. 4 (i), and all inputs are at the logic level H at the 8th pixel.
Therefore, the output signal S1 falls to the logic level L, which causes the output C1 of the NAND circuit 30 to rise to the logic level H.
, The seventh black pixel of the signal B1 is converted into a white pixel, and the pseudo black pixel generated in the portion changing from red to white is removed.

Since any one of the inputs of the NAND circuit 24 is at the logical level L thereafter, the output signal S1 of the NAND circuit 24 maintains the logical level H.

In this way, the pseudo black pixels appearing in the black and white image signal B1 are converted into white pixels to form a black and white image signal BW from which the pseudo black pixels have been removed, and this is output to the outside. That is, the latch circuits 21, 23, 25, 27, the NAND circuits 22, 24, 30, the inverters 26, 29, and the OR circuit 28 form a pseudo black pixel removal circuit DBS that removes pseudo black pixels.

Therefore, the red image signal RD and the black and white image signal BW from which the pseudo black pixel is removed are output in parallel to the outside.

By the way, the color calculation circuit 20 discriminates red color according to the levels of the signals R and C, generates the red image signal R1 and the monochrome image signal B1, and the result is unique to the levels of the signals R and C. Therefore, the color calculation circuit 20 can be configured by a ROM (read only memory) having the signals R and C as input addresses, and in that case, the processing speed can be improved.

In addition, the color calculation circuit 20 and the pseudo black pixel removal circuit DBS are set to 1
It can also be combined into one circuit. Furthermore, the pseudo black pixel removing circuit DBS is not limited to the one described above.

In addition, although the red color of the image is discriminated in the above-described embodiment, it is possible to discriminate colors other than this. Further, the means for dividing the image and separating the spectrum is not limited to the above-mentioned means.

[Effect] As described above, according to the present invention, an image is separated into a discrimination color and its complementary color, photoelectrically converted by a line image sensor, and the output signals of these line image sensors are compared pixel by pixel. It is determined whether or not the pixel is of a discrimination color, and for the determined pixel, a color corresponding to the discrimination color is obtained by performing a predetermined calculation on the discrimination color image signal and the complementary color image signal. An image signal is formed, the other pixels are determined to be black and white pixels, the complementary color image signal is output as a black and white image signal, and when the determined color portion is located independently in the white portion of the background, this Correct the black pixels that appear pseudo around the discrimination color to white pixels,
Since the image quality is improved, there is an advantage that a predetermined color component other than the black-and-white component can be reliably read with high quality.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an example of an optical system of an image reading apparatus according to an embodiment of the present invention, FIG. 2 is a graph diagram for explaining the principle of distinguishing red color, and FIG. 3 is a cyan component. 4 is a waveform diagram for explaining the light receiving characteristics of the light emitting device, FIG. 4 is a waveform diagram for explaining the generation of pseudo black pixels, and FIG. 5 is a block diagram showing an example of an apparatus according to an embodiment of the present invention. FIG. 6 is a flowchart showing a processing example of the color calculation circuit, and FIG. 7 is a waveform diagram for explaining the operation of the pseudo black pixel removal circuit. 6 ... red filter, 7 ... cyan filter, 8,9 ... line image sensor, 11,12 ... switch, 13,14 ... white waveform memory, 15,16 ... normalization circuit, 20 ... Color operation circuit, 2
1,23,25,27 …… Latch circuit, 22,24,30 …… Nand circuit,
26,29 …… Inverter, 28 …… OR circuit, SD1, SD2 …… Shading correction circuit, DBS …… Pseudo black pixel removal circuit.

Claims (1)

[Claims] 1. A spectroscopic means for spectrally dividing image light on a reading line into a predetermined first color and a second color which is a complementary color of the first color, and an image of the first color and the second color. Two line image sensors that receive light and output an image signal corresponding to each pixel, and a color determining unit that determines the first color based on a comparison of the image signals output from these line image sensors. For a pixel for which a discrimination output has been made by the color discrimination means, a color image signal corresponding to the first color is calculated from the image signal on the basis of a predetermined calculation, and there is no discrimination output from the color discrimination means. Color separation means for discriminating a pixel as a black-and-white pixel and outputting an image signal corresponding to the second color as a black-and-white image signal, and an area adjacent to the area for the color-image signal is represented as black by the black-and-white image signal. If the pixels that are An image reading apparatus characterized by having a pseudo black pixel correction means for correcting the black pixels to white pixels.