JPH03273758A - Color picture reader - Google Patents

Abstract

PURPOSE:To obtain a read signal faithful to a source picture with high resolution by outputting the color separation signal of a noted picture element and the color separation signal of the surrounding picture element while being combined corresponding to the relative state. CONSTITUTION:A data arranging part 17 composed of a latch circuit combines picture element data to be serially sent from an A/D converter 16 as one couple for each RGB and forms the noted picture element D1 and the preceding and following picture elements D0 and D2. A select signal generating circuit 20 exerts the respective processing of detecting an edge, ON picture element, thin line, contour position and density upon the respective picture element data D0-D2 and generates a select signal A to which the detected results are effected. A data sending circuit 60 is equipped with a memory to store the several pairs of data trains to be obtained by rearranging the respective BRG data of the picture element data D0-D2 in the internal part and sends a pair of data corresponding to the select signal A from the inside of the memory. Thus, the pair of data closest to the arrangement pattern of original color is read out from the memory.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a color image reading device, and more particularly to a color image reading device that separates and reads picture elements in a dot-sequential manner.

[Prior Art] A conventional color image reading device reads picture elements by color-separating them into, for example, RGB signals. This color separation method mainly includes a three-color illumination sequential switching method, a three-line color sensor method, and a point-sequential color sensor method.

In the three-color illumination sequential switching method, one line of monochrome sensor and RGBB color illumination are provided, and the illumination is changed three times for each line to read an image. However, this method does not increase the image reading speed.

The 3-line color sensor system includes RGB filters and three line sensors respectively, and there are two types: a close-contact type and a reduced type. Although the close-contact type allows each line sensor to be placed without any gaps, it is currently very expensive and has a significant impact on equipment cost. On the other hand, in the reduced type, each RGB signal is captured with a time delay because there is a gap. Therefore, a buffer memory is required, and if there is vibration in the scanning mechanism or the like, the time delay varies, causing image deterioration.

The dot-sequential color sensor method uses an RGB filter provided dot-sequentially for picture elements and a one-line sensor, and according to this method, RGB color information can be extracted in one line. However, due to manufacturing constraints, the total number of pixels in one line is limited, and there is a problem that sufficient resolution cannot be obtained.

In other words, since the RGB filters are adjacent to each other, the pixel pitch needs to be about 20 microns, and since the length that can be taken out from the wafer is fixed, the total number of pixels is 2500.
It only becomes. Since this is approximately 840 picture elements, the reading density is only 4 picture elements per millimeter for the width of an A4 document. So-called 4pel (100dp
i).

One way to solve this problem and increase the reading density is to arrange a plurality of sensors in a staggered manner.

However, this method not only increases costs, but also has many issues, such as correction of sensitivity unevenness in the joints and mounting accuracy of each sensor.

[Problems to be Solved by the Invention] The present invention eliminates the above-mentioned drawbacks of the prior art, and its purpose is to provide a color image reading device that provides a high-resolution reading signal that is faithful to the original image. It is about providing.

[Means and effects for solving the problems] In order to achieve the above object, the color image reading device of the present invention is a color image reading device that reads picture elements by dot-sequential color separation.
a state detection means for detecting a related state between the pixel of interest and its surrounding pixels by comparing a color separation signal of the pixel of interest with a color separation signal of surrounding pixels; and a color separation signal of the pixel of interest. The outline of the present invention is to include a data output means for outputting a combination of color separation signals of a picture element and its surrounding picture elements in accordance with the related state detected by the state detection means.

As a result, even if there is an edge on the picture element of interest, for example, the image can be faithfully read based on the information on the picture elements before and after the edge.

Preferably, the state detecting means includes an edge detecting means for detecting the presence of an edge on the picture element of interest.

Preferably, the state detection means includes on-picture element detection means for detecting that an edge exists on the boundary of the picture element of interest.

Preferably, the state detection means includes a boundary position detection means for detecting the position of an edge on the picture element of interest.

Preferably, one aspect of the boundary position detection means is to detect the position of an edge by comparing differences between color separation signals of the same type between picture elements.

Preferably, the data output means outputs a combination consisting only of color separation signals of picture elements on both sides of the picture element of interest.

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

FIG. 6 is a sectional view showing the reading mechanism section of the color image reading device of the embodiment. In the figure, a first mirror unit 3 on which a fluorescent lamp 1 and a mirror 2 are placed, and a second mirror unit 6 on which mirrors 4 and 5 are placed are driven by a sub-scanning motor unit M to move at a 2:1 speed. Reference numeral 10 represents an original table glass, and 11 represents an original pressing plate. As described above, the reflected image of the document 7 illuminated by the fluorescent lamp 1 is reflected by the mirrors 2, 4.5 and the lens 8.
The image is formed on a color sensor (COD) 9 arranged in a line in the direction perpendicular to the paper surface.

FIG. 7 is a diagram showing details of the light receiving surface of the CCD 9. In the figure, on the light receiving surface of the CCD 9, a red filter (R) 12 and a green filter (G) are arranged for each pixel (dot sequentially) on the light receiving surface.
13 and a blue filter (B) 14 are repeatedly provided. One pixel pitch is 20 microns, and the total number of effective pixels is 2550 pixels. The lens 8 projects an image of 0.083 mm on the original onto one pixel on the CCD 9.

That is, a 4 pel image on the original is read by 12 pel pixels B, R, and G on the CCD.

FIG. 14 is a diagram schematically showing the relationship between the original color projected onto the CCD 9 and the Tsuchiyama force of the CCD 9. Below, CCD9
Riki Tsuchiyama (pixel data) of 5 bits (32 gradations) will be explained with the dark side corresponding to level 31 and the bright side corresponding to level O.

When the original color is read by the CCD 9 as shown in FIG. 14, since the resolution of the lens 8 is actually limited, effects between picture elements appear in the Tsuchiyama force signal. There is no problem in the range of the same color, but in the boundary between white and black or black and white, the influence between picture elements cannot be ignored. For example, if the influence of the color directly above the pixel is 80%, and the influence of the adjacent color is 20%, then the Tsuchiyama force R at the boundary between white and black is the R=O of white and the influence of black. With R=30, R=OX0.8+30X0.2=6
become. Furthermore, Chikara Tsuchiyama G becomes G=O'X0.2+30X0.8=24 due to white's G=O and black's G=30. Furthermore, at the next boundary between black and white, B=30X0
．． 8+OXO, 2=24, R=30X0.2+OXO,
It will be 86. Therefore, if the read data is directly outputted to an external printer or the like without any processing being applied to such a main output signal, it is no longer possible to obtain a reproduced image that is faithful to the color of the original. The solution to this problem will be explained in detail below according to the present embodiment.

FIG. 1 is a block diagram of a signal processing section of a color image reading apparatus according to an embodiment. In the figure, the read signal from the CCD 9 is amplified by an amplifier 15 and then sent to an A/D converter 16.
Converted to bit pixel data. 17 is a data alignment section mainly composed of a latch circuit, and an A/D converter 16
The pixel data serially sent from the BRG is divided into one set for each BRG, and the pixel of interest Di (Bl, R1, Gl) and the pixels before and after it DO (BO, RO, Go), D2 (B2°R2,
G2) is formed. Reference numeral 20 denotes a select signal generation circuit, which generates each pixel data Do, which is sent from the data alignment section 17.
Edge, on-pixel, thin line, boundary position, and shading detection processing, which will be described later, is performed on Di and D2, and a selection signal A reflecting the detection results is generated. 60 is a data sending circuit, inside which picture element data Do, Di, D2 are stored.
It is equipped with a memory that stores several sets of data strings obtained by rearranging each BRG data of 1, and sends out one set of data according to the select signal A from the memory. With this configuration, the arrangement pattern of the original color is sequentially detected from the main output signal of the CCD 9, and according to the detection result, the set of data closest to the arrangement pattern of the original color is read out from the memory, and the set of data is read out from the memory at the timing of reproducing the pixel of interest. Sequentially sent to the outside.

FIG. 2 is a circuit diagram showing details of the select signal generation circuit 20. In the figure, 30 is a basic signal generation section, comparators 41 to 46, which will be described later, and an edge detection section 31. On picture element detection unit 32° boundary position detection unit 35. Thin line detection section 3
3 and a density detection section 34. The basic signal generating section 30 detects two arrangement patterns of document colors using the above-mentioned sections, and generates basic signals 81 to S8 indicating the detection results. Reference numeral 50 denotes a condition setting section, which outputs an appropriate selection signal A1 to read out the set of data closest to the arrangement pattern of the original color from the memory according to the generation status of the basic signals 81 to S8.
~A6 is generated.

<Basic signal generation unit 30> Comparators 41 to 46 calculate the difference between predetermined pixel data and output logic 1 when the result is equal to or greater than a set value (for example, 10).

The edge detecting unit 31 detects an edge on the pixel of interest Di or the pixel of interest D1.
This circuit detects that there is an edge of the original color on the boundary of the document, and outputs logic 1 as the edge detection signal S1 when there is an edge.

The on-picture element detection unit 32 is a circuit that detects that the original color has not changed at least on the pixel of interest Di, and when it detects the state of the on-picture element, it outputs logic l to the on-picture element detection signal S2.
Output.

The thin line detection unit 33 is a circuit that detects the presence of all or part of a thin line of the original color on the pixel of interest D1, and outputs logic 1 as a thin line detection signal S3 when a thin line is detected.

The boundary position detection unit 35 detects the pixel RO where the value of the center pixel R1 of the pixel of interest D1 is in the pixels on both sides thereof.

This circuit detects which of the values of R2 is closer to. When the value of R1 is close to RO, it outputs logic 1 to the boundary position detection signal S4, and when the value of R1 is close to R2, it outputs logic 1 to the boundary position detection signal S5.
Outputs logic 1 to

The shading detection circuit 34 is a circuit that detects in what positional relationship such a thin line exists when viewed from the pixel of interest D1 when a thin line due to a manufactured part or a shallow part is detected,
The details of the shade detection circuit 34 will be explained below with reference to FIG.

FIG. 3 is a circuit diagram showing details of the shade detection circuit 34.

In the figure, the shade detection circuit 34 is connected to the manufacturing part detection section 70. It is composed of a manufacturing unit detection section 72 and a priority order setting section 75.

In the manufacturing unit detection unit 70, 701 to 707 are comparators, each of which receives a pixel signal RO.

When GO, Bl, R1, Gl, B2, or R2 is greater than or equal to a set value (for example, 22), a logic 1 is output. 711-7
15 is an AND circuit, each of which connects three consecutive
Outputs the logical AND of the two comparator outputs. This allows the location of the manufacturing part to be determined.

In the manufacturing unit detection unit 72, 721 to 727 are comparators, each of which receives a pixel signal RO.

When GO, Bl, R1, Gl, B2, or R2 is less than the set value 5 (for example, 10), a logic 1 is output. 73
1 to 735 are AND circuits, each of which outputs the logical AND of three consecutive comparator outputs as shown.

This allows the location of the shallow part to be determined. Then, the manufacturing part detection section 70
and each corresponding AND circuit (for example, 71
1 and 731), they are logically ORed by OR circuits 741 to 745, respectively, and input to the priority setting unit 75.

The priority setting unit 75 prioritizes the outputs of the OR circuits 741 to 745. For example, when the output of the OR circuit 741 (picture element DI of interest) is logic 1 (when there is a fabricated part or a shallow part)
In this case, the output of the OR circuit 742 is invalidated. Further, when the output of the OR circuit 741 is logic O, the output of the OR circuit 742 has the next priority, but when the output of the AND circuit 752 becomes logic 1, the output of the OR circuit 743 is invalidated.

The output of the OR circuit 741 becomes the shade detection signal s6. AN
The outputs of the D circuits 752 and 755 are passed through an OR circuit 762 to become a shade detection signal s7.

The outputs of the AND circuits 754 and 751 are passed through an OR circuit 761 to become a grayscale detection signal S8.

<Condition setting unit 50> In FIG.
1.82). The edge detection signal S1 is the on-picture element detection signal s
2 and the output S61 of the AND circuit 80 are both logic 0 (AND circuit 84). Grayness detection signal S7. S
8 further becomes valid when the output SIO of the AND circuit 84 is logic 10 (AND circuits 81 and 82). Boundary position detection signal S4. S5 indicates that the signal SIO is logic 1 and the thin line detection signal S
It becomes effective when 3 is logic O (AND circuit 85.86)
. The select signal AI is a signal Sll obtained by inverting the signal S10, and the select signal A2 is the output signal S41 of the AND circuit 85. Select signal A3 is AND circuit 8
7, and it can be seen that the select signal A2 has priority over the select signal A3. The select signal A4 is the output signal S71 of the AND circuit 81, and the select signal A5 is the output signal S81 of the AND circuit 82. When the select signals A1 to A5 are all logic O and the edge detection signal S1 is logic 1, the select signal 0 becomes logic 1 (NAND circuit 89).

The operation of the condition setting section 50 will be made clearer by referring to specific examples below.

<Data exit circuit 60> FIG. 4 is a block diagram of the data exit circuit 60.

In the figure, 67 is a selector, and the picture element data Do,
Each pixel data of DI, D2 (BO, RO, Go), (B
l, R1, Gl).

(B2.G2.R2) are distributed to each memory 61 to 66 in an arrangement pattern as shown in the figure. That is, the memory 61 stores a set of picture element data (DI, Di, Dl). The memory 62 stores a set of picture element data (Do, Do, D2). The memory 63 stores a set of picture element data (Do, D2.D2). Here, the picture element data D3 to D6 will be explained.

FIGS. 5A and 5B are diagrams for explaining picture element data D3 to D6. In the figure, three consecutive pixels (Go.

9 Take the bare d3 of Bl, R1) and rearrange it in BRG order to obtain the picture element D3 (Bl, R1, Go). below,
Similarly, take the pairs d4 to d6 as shown in the figure, and set each pair to B.
The pixels are rearranged in RG order to obtain picture elements D4 to D6. thus,
Picture elements (windows) whose phases are shifted one pixel at a time are obtained in the order of Do-D5→D3→Di-D4→D6→D2.

Returning to FIG. 4, the memory 64 stores a set of picture element data (D3.D
3. D6) is stored. The memory 65 stores a set of picture element data (
D5. D4. D4) is stored. The memory 66 stores a set of picture element data (D3.DI.

D4) is stored. Then, the selector 68 selects the attention picture element D.
At the timing of reproducing and outputting 1, select signals A1 to A
6 (relationships in Table 69), any one of the picture element data sets in the memories 61 to 66 is selected and output to the outside.

0 FIGS. 8 to 13 are diagrams schematically showing the relationship between the original color projected onto the CCDQ and the pixel of interest D1. Here, for convenience of explanation, 100 is a black area of the document, and 200 is a white area. The picture element of interest is DI (Bl, R1, Gl)
, and the picture elements before and after it are Do (BO, RO, Go)
, D2 (B2.R2, G2). The operation of each detection means will be explained below.

<Edge Detection> In the example of FIG. 8, there is no edge on the picture element DI of interest or on the boundary of the picture element of interest D1. Therefore, since both 1B1-B21≧10 and IGI-GOI≧10 are not satisfied, the edge detection signal S1 becomes logic O. In the examples shown in Figures 9 to 13, on the picture element of interest DI (Figures 9, 10, and 13) or on its boundary (
There is an edge in Figure 11.12). Therefore IBl-B2
+≧io or IGI-GOI≧10 is satisfied, and the edge detection signal S1 becomes logic 1.

Now, consider a colored original image. For example, 1 in Figure 10
If 00 is black and 200 is yellow, both have blue (B
) is not included, so IBI-B2+≧10 is not satisfied. However, yellow 200 includes green (G1), so I
GI-GO+≧10 is satisfied. Therefore, edges are detected at the boundary between black and yellow. On the other hand, 100 is black and 2
Considering the case where 00 is red, red 200 has green (G1
) is not included, so IGI-GO≧10 is no longer satisfied. Therefore, no edges are detected at the boundary between black and red. In this way, image quality is improved by accurately detecting and reproducing such boundaries in document color boundaries that are generally dark and have low contrast, such as red and black, blue and black, red and blue, etc. In this embodiment, such a boundary is not detected because it does not have much influence on the image quality.

On-picture element detection> In the example of FIG. 11, an edge is detected, but the edge is not on the target picture element D1. When reproducing such a picture element, it is desired to output the contents of the noticed picture element D1 as is. The on picture element detection circuit 32 is a circuit that detects such a state.

In the example of Figure 1, 1G1-GO+≧10 and IRI-R
Since both do not satisfy 21≧10, the ON picture element detection signal S2
becomes logic 1. Conversely, the same thing can be said if we consider the case where the area up to GO is white and the area after Bl is black. In this case 1131-5
Since both IRI-R21≧10 and IRI-R21≧10 are not satisfied, the on-picture element detection signal S2 becomes logic 1.

3 Thin Line Detection> In the examples shown in Figures 8 to 10, the area around the pixel of interest D1 is covered with a wide range of black 100 and a wide range of white 200, so generally I
Both BI-BOI≧10 and IGIG2+≧10 are not satisfied. Therefore, the thin line detection signal S3 becomes logic O. However, in the example of Figure 12.13, the target picture element D1 is covered by all or part of the narrow black 100, so IBL
-B01≧10 or IGI-G2+≧10 is satisfied.

Therefore, the thin line detection signal S3 becomes logic 1.

In the example of FIG. 11, there is a wide black 1 area near the picture element D1
00 and a wide white 200, but IGI-G2
Since +≧10 is satisfied, the thin line detection signal S3 becomes logic 1. However, at the same time, the on-picture element detection signal S2 becomes logic 1, so the signal S10 in FIG.
1.82).

<Boundary position detection> An edge is detected, it is not a thin line, and the pixel of interest D1
In some cases, the color may change (that is, the picture element is not on) in the middle of the process. In this case, it is desirable to reflect the color of the neighboring picture element when reproducing the picture element Di of interest. For example, attention picture element D
If there is a boundary between skin color and black on D1, at least the signals at both ends of the set of three signals output when reproducing the picture element of interest D1 are set to skin color and black, respectively. The boundary position detection circuit 35 determines which color to use for the middle signal.

That is, the boundary position detection circuit 35 is a circuit that detects which of the center pixels RO and R2 of the picture elements DO, D2 on both sides of which the value of the center pixel R1 of the picture element of interest D1 is closer. 9th
In the illustrated example, there are black 100 up to B1. Therefore l R1-
R21≧10 is satisfied, and the boundary position detection signal S4 is logic O.
become. On the other hand, since R1-R21≧10 is not satisfied, the boundary position detection signal S5 becomes logic 1. Therefore, in the example of FIG. 9, the color of R1 is close to the picture element B2 side. Further, in the example of FIG. 10, there are black 100 up to R1. Therefore, R1-R21≧
10 is not satisfied, the boundary position detection signal S4 becomes logic 1. Also, since l R1-R21≧10 is satisfied, the boundary position detection signal S5 becomes logic O. Therefore, in the example of FIG. 10, the color of R1 is close to the picture element DO side. Incidentally, when the boundary position detection signals S4 and 85 are both logic 1, the 84 side is given priority in the AND circuit 87 of FIG.

<Shade Detection> In the example of FIG. 13, there is a thin black line 100 spanning the picture element D1 of interest and the picture element Do before it.

In such a case, if the complete black line 100 is reproduced by the picture element DO or D1, reproduction faithful to the original image cannot be obtained. Therefore, in this embodiment, the picture element d3 (Go, Bl, R1) shifted by one pixel as shown in FIG.
, that is, B3 (Bl.

R1, GO). The shading detection circuit 34 in FIG. 3 selects a picture element (manufacturing part) in which all three pixels have a large signal, or a signal in all three pixels has a small signal, from among each set of three consecutive pixels as shown in FIG. 5(A). This is a circuit that searches for picture elements (shallow parts).

In the example of FIG. 13, the set of d3 (Go, Bl, R1) is large, so the output of the AND circuit 712 of FIG. 3 becomes logic 1. Also, in this example, d6(Gl.

B2. Since the R2) set is small, the output of the AND circuit element 735 in FIG. 3 becomes logic 1. Therefore, OR circuit 742
The outputs of and 745 both become logic 1, and both pass through the priority order setting section 75 while remaining valid, and the shade detection signal S7 becomes logic 1. This gray level detection signal S7 satisfies the AND circuit 81 in FIG. 2 and sets the selection signal A4 to logic 1. As a result, the picture element of interest D1 in FIG. 13 is reproduced by the data set (B3.B3.B6) read from the memory 64. That is, as is clear from FIG. 5(A), the first two pixels of the target picture element D1 are reproduced twice by the data D3 of the phase d3, and the remaining one pixel of the target picture element D1 is reproduced twice by the phase d3. It is reproduced once by the data D6 of d6. This reproduces thin lines that are faithful to the original.

FIG. 15 is a diagram illustrating the relationship between an example of original color and reproduced color. In the figure, 300 is a color signal field that represents the basic colors on the original and their original BRG components, 301 is a document color field that schematically represents the arrangement of the original colors, and 302 is an arrangement field for the reading pixels of the CCD 9. , 303 is the picture element number column sequentially attached to the picture elements, 304 is C
A column showing the Tsuchiyama power signal of CD9, 305 a column showing the presence or absence of edge detection, 306 a column showing the presence or absence of on-pixel detection, 3
07 is a column indicating the presence or absence of thin line detection, 308 is a column indicating the state of boundary position detection, 309 is a column indicating the state of grayscale detection, 3
10 is a column showing a selection signal generated for each pixel of interest, 311 is a column showing a set of output data selected by the selection signal, and 312 is a column showing an arrangement of reproduced colors.

To explain the Riki Tsuchiyama column 304, at the boundary between skin color and black in picture element No. 3, B = 16 for skin color and B = 30 for black.
Therefore, B=16xO, 8+30x0.2=18.8
D19. Furthermore, since R=10 for skin color and R=30 for black, R=10X0.2+30X0.826.

In this way, a color output signal with a component different from the original original color appears. The processing of this embodiment will be described below in accordance with the pixel numbers.

<Pixel No. 2> The edge detection column 305 is as follows: IBI-B21=+3O-191=3<101GI-G
01=+18-181=O<10, so "none".

When an edge is not detected, other detections are effectively disabled (indicated by / in the figure), and the select signal A1 becomes logic 1 preferentially, so the memory 61 is selected and three sets of data (DI, DI , Di) are sent out. Therefore, the reproduction of this picture element is faithful to the original original color.

Picture element No. 3> Edge detection column 305 is IBI-B2 +=+ 19-301=11≧101G
“Yes” because I-GO+=+3O-181=12≧10
It is.

The on-picture element detection field 306 is "absent" because there is an edge on the picture element of interest.

The thin line detection column 307 is IBI-B21=+3O-161=3<101GI-G
2+=+3O-301=O<10, so "none". That is, a thin line (black or white) that can be called a thin line does not cover the pixel of interest.

The boundary position detection field 308 is as follows: IRI-R01=+26-101=16≧10IRI-
Close to "R2" because R2+=+26-301=O<101.

The shading detection field 309 is substantially invalidated by "none" in the thin line detection field 307.

As a result of the above, the select signal A3 becomes logic 1, and the memory 6
3 is selected and three sets of data (DO9D2.D2) are sent. Therefore, even if the color output signal of the CCD 9 is affected by the boundary between flesh color and black, the reproduced output after processing is faithful to the original original color.

Pixel element No. 0.4> The edge detection field 305 is IBI-B21=+3O-271=3<101GI-G
O+=+3O-301=O<10, so "none". Therefore, the select signal At becomes logic 1, and the memory 61
is selected and three sets of data (Di, Di, Di) are sent out.

2 <Pixel No., 5> The edge detection column 305 is IBI-B21=+3O-161=11≧101GI-
GO+=+18-301=12≧10, so "yes".

The on-picture element detection field 306 is "absent" because there is an edge on the picture element of interest.

The thin line detection column 307 is as follows: IBI-B21=+3O-301=3<101GI-G
Since 2+=+18-281=10≧10, it is “present”. That is, a part of the skin-colored thin line overlaps the pixel of interest.

The boundary position detection field 308 is substantially invalidated by "existence" in the thin line detection field 307.

The shading detection column 309 indicates RO≧22, Go in FIG.
≧22. Since both satisfy Bl≧22, this indicates that the manufacturing part was detected at d5 in FIG. 5(A), and "D5"
It is. Also, R1≦10, Gl≦10. Since both satisfy B2≦10, this indicates that the shallow portion was detected at d4 in FIG. 5(A), which is "D4". As a result, the priority setting section 75 outputs a signal S8.

As a result, the select signal A5 becomes logic 1, and three sets of data (D5.D4., D4) from the memory 65 are sent out.

Furthermore, looking at the playback output, the BRG of D5 is (27, 30,
30), which is the original black (30, 30, 30
) is somewhat different. This is because the boundaries between the original colors are slightly affected by the optical performance, but this is not a problem since it is actually reproduced as almost black. Similarly, D4 also has a slightly different skin color from the original. However, this is also almost a skin color. Although there are some differences as described above, for example, the size of one pixel on printing is (1/12) millimeter, so it has almost no effect.

<Pixel No. 6> In the same way as Pixel No. 5, a thin line was detected and the manufacturing department
D6" is detected, the select signal becomes A4, and three sets of data (D3.D3.D6) are sent out.

<Pixel No. 0.7> A thin line is detected, but since the manufacturing part is detected at "Dl", the density detection signal S6 is satisfied, and the AND circuit 80 in FIG.
satisfy. This invalidates AND circuit 84 and outputs select signal A1.

Therefore, the memory 61 is selected and three sets of data (DI, D
I, Di) are sent.

<Picture element No. 8> 5 On picture element detection field 306 is IBI-B2+=+19-161=3<10IRI-R
Since 2+=+1O-101=O<10, it becomes "present".

When an on-picture element is detected, the AND circuit 84 in FIG. 2 is blocked, and the select signal A1 becomes logic 1. Therefore, three sets of data (Di, Di, Di) are sent out.

As a result of the above, the original original color 301 was reproduced extremely faithfully as shown in reproduced color 3-2.

FIG. 16 is a diagram showing reproduced colors when no processing is performed for the original colors shown in FIG. 15. In case of no processing, select signal A1
, that is, three sets of data (Dl.

DI, Di). Picture element N in Figure 16
In 0.3, as a result of the CCD 9 reading the boundary image, Riki Tsuchiyama's BRG is (19, 26, 30), but since this is output as is, 6 dark blue is actually reproduced. In the same way, picture element No. 5 becomes yellow-colored, picture element No. 6 becomes shrimp-based, and is reproduced as a completely different color. Furthermore, since the presence or absence of edges, the location of edges, etc. are not detected, the unit of reproduction is of course the unit of picture element, and the texture of the reproduced image becomes coarse. On the other hand, when processing as shown in Fig. 15, 3 of 1 pixel
The output data can be changed in units of 1/10th the size, that is, pixels, and the original can be accurately reproduced.

FIG. 17 is a diagram illustrating the relationship between another example of original colors and reproduced colors. In this example, the thin line is finer than that in Fig. 15, that is, 3
This figure shows how a document with thin black and white lines of pixel width (approximately 0.25 mm) and two-pixel width (approximately 0.17 mm) is processed. In the figure, edges are detected in all picture elements No. 2 to No. 8, and thin lines are detected. Also, picture element No. 0.2~N
At 0.6, a shallow part or a shallow part is detected, and predetermined processing is performed for each. However, if the line width of the original is less than three pixels, the color of the original will not be reproduced exactly. For example, in picture element N0.3, a shallow part (white) is detected at the phase of d5, but the following data d4, that is, D4 is (B, R
.

G) = (6, 24, 24) and is not black. This is blue.

Furthermore, for picture elements No. 7 and No. 8, the original color is a continuous thin line with a width of 2 pixels, and there is no large or small portion for any of the three pixels, so no shading portion is detected in the shading detection. At this time, in FIG. 2, the signal S6. S7. B
8 both become logic O, and eventually the output A of the NAND circuit 89
6 becomes logic l. This results in three sets of data (D3.
Di, D4) is sent out. This data array pattern is a pattern that outputs three consecutive pixels centered on each pixel of the pixel of interest, and is suitable for reflecting the state of picture elements No. 0.7, No. 8, and No. 8. It's a pattern.

As a result, even if black and white cannot be clearly reproduced, they can be reproduced with a considerable difference in density, and the contours are accurately reproduced.

In this way, especially in originals with continuous thin lines, black may be reproduced as blue or green, and white as yellow or light blue, but the boundaries between shades are reproduced correctly and the outline is constructed in pixel units. This allows you to obtain detailed images.

FIG. 18 is a diagram showing reproduced colors when no processing is performed for the original colors shown in FIG. 17. Looking at the reproduction state of FIG. 18, not only are the line widths and positions incorrect, but the white lines that were on the document for picture elements No. 7 to 8 have disappeared in the reproduction. Therefore, this embodiment has the effect of providing extremely good reproducibility even for originals with a series of fairly fine lines (approximately 0.17 mm).

Next, as another example, we will discuss a case where the contrast is low.

FIG. 19 is a diagram illustrating the relationship between the original colors of white and light blue and the reproduced colors. For picture element No. 3, IBI-B2+=+1-71=6<10IRI-R2+
Since =+24-301=6<10, on-picture element detection is "present". By the way, if you look at the original color, there is a border between white and light blue on picture element No. 3, so normally this pixel of interest would not be in the ON picture element state, but because the color of B2 is light blue. As a result, B2=7 is small, and as a result, on-pixel detection is
``Yes''. This kind of case can be dealt with by increasing the judgment conditions, but conversely, if the two data are close, it means that the colors are also similar, so this part should be output unprocessed. However, there is no problem with the color. Therefore, no special measures are taken in this embodiment. Therefore, an ON picture element is detected in picture element No. 3, the select signal becomes A1, and three pieces of data of the target picture element D1 are sent out as they are. However, light blue is originally a color with a small B value, so even if white is bright in the B part of the picture element of interest and its value decreases, there is almost no color effect. However, the light blue reproduction pixel position is shifted by one pixel. The same applies to picture element No. 5.

However, picture element No. 6. In the case of picture element No. 8, the ON picture element is not detected, and the boundary position is detected using normal processing, and the boundary is correctly reproduced.

In this way, if the contrast of the original is low, an error of one pixel may occur depending on the case, but an error of one pixel, or 0.083 mm, in the low contrast area will hardly affect the actual image. No impact.

In the explanation so far, three sets of pixel data are output when reproducing one pixel, but the present invention is not limited to this. For example, it is easy to make two sets, and four or five sets have the effect of reproducing boundaries even more precisely.

FIG. 20 is a diagram illustrating a case where five sets of pixel data are output for one picture element. In picture element No. 0.1, black is applied to the boundary of pixel B, but in picture elements No. 2 and 4, black is applied to the middle of pixel G or pixel R. Therefore, Tsuchiyama force of CCD9 appears smaller in the case of picture elements No. 2 and 4 than in the case of picture element No. 0.1. For example, the G of picture element No. 2 is 16. Therefore, in the example shown in the figure, when G=16, only one set of black signals is sent, and when B=24
At the time of , two sets of black signals are sent out. In this way, subtle differences in the positions of the boundaries cause differences in the raw white power of the CCD 9, and this can be used to consider five sets of outputs.

As explained above (according to this embodiment, boundaries are formed pixel by pixel, that is, at a pitch of 1/12 mm in areas with high contrast, and even in areas with low contrast, such as boundaries between white and light blue, depending on the case) The error is only one pixel, and the original original color is reproduced with great fidelity.

Furthermore, even for thin lines (0.17 mm or less) that are less than one picture element, the shading can be clearly distinguished and reproduced. That is, the image is reproduced at a pixel pitch that is one-third of the pitch of one picture element, and a reproduced image that is fine and sharp is obtained.

For example, the dot sequential sensor used in this embodiment can normally read an A4 document at only 4 pels (100 dpi) due to the number of picture elements, but according to the present invention, it can read an A4 document at 12 pels (3 pels).
00dpi) and can be read and played back.

Although the above embodiments have been described with respect to a scanner that reads an image of a document, the present invention can also be applied to an imaging device such as a video camera.

[Effects of the Invention] As described above, according to the present invention, since each pixel of the dot sequential sensor 4 is effectively utilized, the original can be read and reproduced at a pitch finer than the pixel pitch, resulting in detailed reproduction that is faithful to the original. We can provide images.

Further, since it can be implemented with a simple circuit configuration and does not require complicated optical adjustment, the reading resolution can be increased without a large increase in cost.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the signal processing section of the color image reading device of the embodiment, FIG. 2 is a circuit diagram showing details of the select signal generation circuit 20, and FIG. 3 is a circuit diagram showing details of the shade detection circuit 34. , FIG. 4 is a block configuration diagram of the data exit circuit 60, FIGS. 5(A) and 5(B) are diagrams for explaining picture element data D3 to D6, and FIG. 6 is a reading mechanism of the color image reading device of the embodiment. FIG. 7 is a diagram showing details of the light receiving surface of the CCD 9, and FIGS.
Figure 14 is a diagram schematically showing the relationship between the original color projected onto CCDe and Tsuchiyama force of CCD9, and Figure 15 is the relationship between an example of original color and reproduced color. FIG. 16 is a diagram showing the reproduced color in the case of no processing for the original color in FIG. 15. FIG. 17 is a diagram explaining the relationship between other examples of original color and reproduced color. Figure 18 is a diagram showing the reproduced color in the case of no processing for the original color in Figure 17, Figure 19 is a diagram explaining the relationship between the reproduced color and the original color consisting of white and light blue, and Figure 20 is FIG. 6 is a diagram illustrating a case where five sets of pixel data are output for one picture element. In the figure, 9...CCD, 12...Red filter, 13
...Green filter, 14...Blue filter, 17.
...Data alignment section, 20...Select signal generation circuit,
30... Basic signal generation section, 31... Edge detection section,
32...On picture element detection section, 33...Thin line detection section, 3
4... Concentration detection section, 35... Boundary position detection section, 50
...Condition setting section, 60...Data sending circuit, 61-
66...Data storage memory, 67.68...Selector, 70...Setting part detection unit, 72...Shallow part detection unit.

Claims (6)

[Claims] (1) In a color image reading device that separates and reads picture elements dot-sequentially, the picture element of interest and its surrounding picture elements are identified by comparing the color separation signals of the picture element of interest and the color separation signals of its surrounding picture elements. and a data output means that outputs a combination of a color separation signal of the pixel of interest and a color separation signal of surrounding pixels in accordance with the related state detected by the state detection means. A color image reading device comprising: (2) The color image reading device according to claim 1, wherein the state detecting means includes edge detecting means for detecting that an edge exists on the picture element of interest. (3) The color image reading device according to claim 1, wherein the state detecting means includes on-picture element detecting means for detecting that an edge exists on a boundary of a picture element of interest. (4) The color image reading device according to claim 1, wherein the state detecting means includes a boundary position detecting means for detecting the position of an edge on the picture element of interest. (5) The color image reading device according to claim 4, wherein the boundary position detecting means detects the position of the edge by comparing differences between color separation signals of the same type between picture elements. (6) The color image reading device according to claim 1, wherein the data output means outputs a combination consisting only of color separation signals of picture elements on both sides of the picture element of interest.