JPH0522601A - Method and device for emphasizing contour - Google Patents

Abstract

PURPOSE:To attain the contour emphasizing method/device which is free from the deterioration of the contour emphasizing degree and also free from the occurrence of the double and triple contours. CONSTITUTION:The sharp signal S which is converted into an electric signal by an image sensor undergoes the filter processing like the average processing, etc., through a filter processing circuit 2 after such processing operations as the log conversion, the A/D conversion, the shading correction, etc. Thus the signal S undergone the filter processing is obtained and then differentiated in the prescribed direction by a primary differential signal production circuit 4. Thus the primary differential signal S' of a noted picture element is obtained. Then the signal S' undergoes the blind sector processing through a blind sector processing circuit 6 and then undergoes the secondary differentiation in the prescribed direction again through an edge emphasis signal production circuit 8. Thus an edge emphasis signal Se'' is obtained after the average processing of the signal S'. Then the signal S undergone the filter processing is outputted together with a signal S-Se'' which is operated and emphasized by an emphasized signal production circuit 10. A delay circuit 12 functions to control the timing of the averaged sharp signal S in principle.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour emphasizing method and a contour emphasizing apparatus used in an image scanning recording apparatus such as a scanner for plate making.

[0002]

2. Description of the Related Art FIG. 18 shows a conventional contour enhancement device.

The sharp signal S for each pixel read by scanning the original is supplied to the unsharp signal generating circuit 202. Here, the sharp signal S is a digital signal, and its value represents the density of the document. When the light and dark parts of the document are scanned, the level of the sharp signal S is as shown in FIG. 19 (1).
As shown in. The unsharp signal creation circuit 202 averages the sharp signal S such as simple average or weighted average to create an unsharp signal U from the sharp signal S. The unsharp signal U (X15, Y15) corresponds to the sharp signal S (X15, Y15) in the pixel of interest. For example, the sharp signal S (31 × 31 rectangular range around the pixel of interest and the pixel of interest is a sharp signal S ( X = 0, Y = 0), ~, S (X = 3
0, Y = 30) is averaged. Therefore, the unsharp signal U becomes as shown in FIG. The sharp signal S and the unsharp signal U are given to the subtractor 204, and a difference signal (SU) for enhancing the contour is created (see FIG. 19 (3)). This difference signal (S−U) is given to the coefficient unit 206, and an edge emphasis signal k · (S−U) is created by multiplying it by an appropriate coefficient (FIG. 19).
(See (4)). The edge emphasis signal k · (S−U) is generated in correspondence with a portion α1 of the document which sharply changes from black to white and a portion α2 of the document which sharply changes from white to black (FIG. 19 ( See 4) β1 and β2). Sharp signal S
And the edge enhanced signal k · (S−U) are given to the adder 208, and the enhanced signal S + k · (S− is obtained from the sum of both signals.
U) is created (see FIG. 19 (5)).

In this emphasized signal S + k.multidot. (S-U), the density difference is associated with the portion .alpha.1 of the original that sharply changes from black to white and the portion .alpha.2 of the original that sharply changes from white to black. More emphasized. Therefore, the contour of the original image is emphasized, and a reproduced image which is apparently clear can be obtained.

When the original is a film or the like, the sharp signal S may include a noise component due to graininess and the like. Therefore, the sharp signal S
When a noise component is included in the noise component, if the contour enhancement device and the contour enhancement method shown in FIG. 18 are used to enhance the contour, the noise component portion is also emphasized, so that the noise is also conspicuous. It was

In order to prevent the noise from being emphasized, FIG.
0, the contour emphasis device of FIG. 22 is considered.

In the contour emphasizing apparatus of FIG. 20, the sharp signal S is also given to the averaging circuit 210. In the averaging circuit 210, for example, the averaged sharp signal S is created from the sharp signal S in the mask of the target pixel and the 15 × 15 rectangular range around the target pixel. The averaging circuit 210 averages the sharp signal S, such as a simple average or a weighted average, and creates an averaged sharp signal S from the sharp signal S. Therefore, even when the sharp signal S includes noise (see FIG. 21).
(1)), the noise component is averaged, and the averaged sharp signal S contains almost no noise component (see FIG. 21 (2)). The unsharp signal U averaged over a wider range than the sharp signal S also does not include a noise component (see FIG. 21 (3)).

The averaged sharp signal S and unsharp signal U are applied to the subtractor 204 and the difference signal ( S-
U) is generated (see FIG. 21 (4)), and the edge enhancement signal k · ( S− U) is generated by the coefficient unit 206 (see FIG. 21 (5)). The averaged sharp signal S and the edge emphasis signal k · ( S− U) are given to the adder 208, and the emphasized signal S + k · ( S− U) is created from the sum of both signals (FIG. 21). (See (6)).

Also in this emphasized signal S + k.multidot. ( S- U), the density corresponding to the portion .alpha.1 of the manuscript which sharply changes from black to white and the portion .alpha.2 of the manuscript which sharply changes from white to black. The difference is more emphasized (δ in Fig. 21 (6)
See 1 and δ2). Therefore, the contour can be emphasized without emphasizing noise.

In the contour emphasizing device of FIG. 22, the edge emphasizing signal k · (S−U) output from the coefficient unit 206 (FIG. 23).
(See (3)) is provided to the dead zone processing circuit 212. The dead zone processing circuit 212 removes a signal having a level in the range of the dead zone Δ and passes only a signal having a level exceeding the dead zone Δ. The dead zone Δ is set to be slightly larger than the maximum value of the noise component level (see FIG. 23 (1)). The dead zone processing circuit 212 removes the signal in the range of the dead zone Δ from the edge enhanced signal k · (S−U), and the edge enhanced signal k · (S−U) subjected to the dead zone at a level exceeding the dead zone Δ.
Δ is output (see FIG. 23 (4)).

Therefore, the dead zone processing circuit 212 uses the edge emphasis signal k. (S-
U) It does not output Δ and outputs the edge corresponding to the portion α1 of the document that sharply changes from black to white and the portion α2 of the document that sharply changes from white to black (see FIG. 21 (1)). An emphasis signal is output (see FIG. 23 (4)).

The sharp signal S and the dead band processed edge enhanced signal k · (S−U) Δ are given to the adder 208, and the enhanced signal S + k · (S−U) Δ is obtained from the sum of both signals.
Is created (see FIG. 23 (5)).

Also in this emphasized signal S + k · (S−U) Δ, the density corresponding to the portion α1 of the original that sharply changes from black to white and the portion α2 of the original that sharply changes from white to black, respectively. The difference is more emphasized (see δ1 and δ2 in FIG. 23 (5)). Therefore, the contour can be emphasized while leaving the noise as it is.

[0014]

However, FIG. 20 and FIG.
In the contour emphasizing device and the contour emphasizing method shown in FIG. 1, since the emphasizing width of the edge emphasizing signal is widened (see FIG. 21 (5)), the inclination of the emphasizing portion of the emphasizing signal S + k · ( S− U) becomes gentle. Will end up. For this reason, there is a drawback that the degree of contour enhancement decreases.

In the contour emphasizing apparatus and the contour emphasizing method shown in FIGS. 22 and 23, the dead zone processed edge enhancing signal k
So-called zero-cross distortion occurs in (S−U) Δ (see ε1 and ε2 in FIG. 23 (4)). Therefore, there is a step in the contour emphasis portion of the emphasized signal S + k · (S−U) Δ, and 2
There was a defect that contours were tripled and tripled.

The present invention solves the above-mentioned technical problems, prevents noise from being emphasized, does not reduce the degree of contour emphasis, and does not generate double or triple contours. , An outline emphasis device is provided.

[0017]

In order to solve the above technical problems, the present invention has the following configurations.

That is, according to the contour enhancement method of claim 1, the sharp signal representing the density of the pixel of interest obtained by reading the original image and the sharp signal of the peripheral pixels around a predetermined range of the pixel of interest are filtered. , Create a filtered sharp signal in the pixel of interest, directionally differentiate the filtered sharp signal in a predetermined direction, create a first-order differential signal in the target pixel, and again direct the first-order differential signal in the predetermined direction. Differentiate to create a secondary differential signal in the pixel of interest, average the secondary differential signals in each direction to create an edge enhancement signal, calculate the filtered sharp signal and the edge enhancement signal, and calculate the pixel of interest. And creating an enhanced signal in.

According to a second aspect of the present invention, the contour emphasis method filters the sharp signal representing the density of the pixel of interest obtained by reading the original image and the sharp signals of peripheral pixels around a predetermined range of the pixel of interest, and the sharp signal is noted. A filtered sharp signal for a pixel is created, the filtered sharp signal is directionally differentiated in a predetermined direction, and 1
A primary differential signal is created, dead zone processing is performed on the primary differential signal, a dead zone processed primary differential signal in the pixel of interest is created, and the dead zone processed primary differential signal is directional differentiated again in a predetermined direction, A secondary differential signal in the pixel of interest is created, the secondary differential signals in each direction are averaged to create an edge emphasis signal, and a sharp signal or a filtered sharp signal and an edge emphasis signal are calculated to calculate the edge emphasis signal in the pixel of interest. It is characterized by creating an emphasized signal.

A contour enhancement method according to a third aspect of the present invention is the method according to the first or second aspect, wherein the filtered sharp signal is subjected to a direction differentiation in a predetermined direction to obtain a 1 in a target pixel. It is characterized in that it is a second derivative signal.

According to another aspect of the present invention, the contour emphasizing device filters the sharp signal representing the density of the pixel of interest obtained by reading the original image and the sharp signals of peripheral pixels around a predetermined range of the pixel of interest, and the sharp signal is processed. A filter processing circuit that creates a filtered sharp signal in a pixel, a primary differential signal creation circuit that differentially differentiates the filtered sharp signal in a predetermined direction to create a primary differential signal in a pixel of interest, and a primary differential signal An edge enhancement signal creation circuit that creates a secondary differential signal in the target pixel by deriving a differential differentiation of the differential signal in a predetermined direction again, and averages the secondary differential signals in each direction to create an edge enhancement signal, and a filtering process. An enhanced signal generating circuit that calculates an enhanced sharp signal and an edge enhanced signal to create an enhanced signal in the pixel of interest. And butterflies.

According to another aspect of the present invention, the contour emphasizing apparatus filters the sharp signal representing the density of the pixel of interest obtained by reading the original image and the sharp signals of the peripheral pixels around a predetermined range of the pixel of interest to filter the sharp signal. A filter processing circuit that creates a filtered sharp signal in a pixel, a primary differential signal creation circuit that differentially differentiates the filtered sharp signal in a predetermined direction to create a primary differential signal in a pixel of interest, and a primary differential signal A dead-zone processing circuit that performs dead-zone processing on the differential signal to create a dead-zone processed first-order differential signal at the pixel of interest, and a dead-zone processed first-order differential signal that is directionally differentiated again in a predetermined direction to obtain a secondary signal at the pixel of interest. An edge emphasis signal generation circuit for generating a differential signal and averaging secondary differential signals in each direction to generate an edge emphasis signal;
A sharpened signal or a filtered sharpened signal and an edge-enhanced signal are calculated, and an enhanced signal generation circuit for generating an enhanced signal in a target pixel is provided.

According to a sixth aspect of the present invention, in the contour enhancing apparatus according to the fourth or fifth aspect, the primary differential signal generating circuit further averages the directionally differentiated ones in a predetermined direction to obtain a target pixel. It is characterized in that it is a first-order differential signal in.

[0024]

In the contour enhancing method according to the first aspect and the contour enhancing apparatus according to the fourth aspect, when filtering is performed, even if the sharp signal includes a noise component, the filtered sharp signal hardly includes a noise component. Like

When the filtered sharp signal is directional differentiated in a predetermined direction, a point where the change is sharp, that is,
It is possible to obtain a first-order differential signal having a positively convex peak or a negatively convex peak at dark-to-light and light-to-dark transition portions of an image.

When the primary differential signal is directional differentiated again in a predetermined direction, the point where the change is abrupt, that is, the peak of the primary differential signal is scissored, and there is a positive convex peak or a negative convex peak in the vicinity. A secondary differential signal can be obtained, and an edge emphasis signal can be obtained by averaging the secondary differential signals for each direction.

Next, the filtered sharp signal and the edge emphasizing signal are calculated to create the emphasizing signal in the target pixel. The enhanced signal has peaks at both ends of the dark-to-light and light-to-dark transitions of the image. Moreover, the slope of the changing portion is steep and there is no step.

In the contour emphasizing method according to claim 2 and the contour emphasizing device according to claim 5, when the primary differential signal is subjected to the dead zone processing, even if the primary differential signal includes a noise component, this noise component is removed. Will be.

Next, the sharp signal or the filtered sharp signal and the edge enhancement signal are calculated to create the enhanced signal in the target pixel. The enhanced signal has peaks at both ends of the dark-to-light and light-to-dark transitions of the image. Moreover, the slope of the changing portion is steep and there is no step. Also, the noise component is less emphasized.

The contour enhancement method according to claim 3 and the contour enhancement apparatus according to claim 6 are the contour enhancement method according to claim 1 or 2, and those according to claim 4 or 5, wherein the filtered sharp signal is directed. Since the differentiated one is further direction-averaged in a predetermined direction to obtain a primary differential signal in the target pixel, even if the primary differential signal contains a noise component, this noise component will be removed. Therefore, the noise component is less emphasized.

[0031]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of an outline emphasizing apparatus according to an embodiment of the present invention.

This contour emphasizing device comprises a filter processing circuit 2, a primary differential signal generating circuit 4, a dead zone processing circuit 6, an edge emphasizing signal generating circuit 8, an emphasized signal generating circuit 10 and a delay circuit 12.

An image obtained by scanning an original in the main scanning direction X and the sub scanning direction Y is converted into an electric signal by an image sensor, and a sharp signal S is output from the image sensor. The sharp signal S represents the density of the document image in each position of the document in the main scanning direction X and the sub scanning direction Y, that is, the target pixel. The sharp signal S is processed after log conversion, analog / digital conversion, shading correction, etc.
It is given to the filter processing circuit 2 of the contour enhancement device.

-Filter Processing Circuit 2 and Filtered Sharp Signal Generation-The filter processing circuit 2 has, for example, as shown in FIG. 2, 15 rectangular shapes in the main scanning direction X and 15 rectangular shapes in the sub scanning direction Y. Filtering is performed by simply averaging the sharp signals S in the range mask. By this filtering process, the filtered sharp signal S for the pixel of interest is created. As shown in FIG. 2, the sharp signal in the target pixel is S (X = 7, Y = 7) (hereinafter simply referred to as S (7,7)), and the sharp signal in the predetermined range of the target pixel is S (0,7,7).
0), ~, S (7,6), S (7,8), ~, S (14,14)
Filtered sharp signal S (7,
7) is (S (0,0) + ~ + S (7,6) + S (7,7) + S (7,8) + ~
+ S (14,14)) / (15 × 15). A filter processing circuit 2 for creating this filtered sharp signal S
A specific circuit of the above is shown in FIG.

Each time the sharp signal S is sent, the sharp signal S is latched by the latch 101. The line buffer 104 includes 15 line memories 104 _{0 to} 104 ₁₄
Is equipped with. Line memory 104 _{0-104 14} has a memory area of 64K pixels respectively.

Now, the sharp signal S (0,15) is latched by the latch 101.
Suppose it is latched to. At this time, the contents of the line buffer 104 are as shown in FIG.

First, the contents [S (0,0) to S (0,15)] of the address "0" of the line memory are read out and latched in the latch 105 and the latch 102. The latch 105 latches all 15 data, but the latch 102 holds only 14 data excluding the contents of the line memory 104 ₀ .

Next, the sharp signal S (0,15) held in the latch 101 and the 14 pieces of data (S (0,1) to S (0,14)) held in the latch 102 are stored. Together, S (0,15)
To the line memory 104 ₁₄ and S (0,14) to the line memory 104 ₁₃
And S (0,13) are sequentially written into the line memory 104 _{1 2} and so on. This operation is repeated, and the sharp signals S15 in each of the line memories 104 _{0 to} 104 ₁₄ are set to 1
It is read every time and held by the latch 105.

Since the sharp signal S of one line in the sub-scanning direction Y is temporarily latched in the latch 105, the addition of the sharp signal S of each pixel in the sub-scanning direction Y is performed line by line in the data selectors 106 to 109. To be done. The data selector 106 performs parallel / serial conversion and sequentially outputs 15 sharp signals S included in one line one by one. For example, the latch 105 outputs a sharp signal S (14,0), ...,
When S (14,14) is output, the sharp signal S (14,14)
0), to, S (14, 14) are sequentially output to the adder 107 one by one.

In the adder 107 and the latch 108, first,
When 107 receives the sharp signal S (14,0) and outputs it, the latch 108 latches the sharp signal S (14,0) and returns it to the adder 107. Next, the adder 107 outputs the sharp signal S (14,
0) and the sharp signal S (14,1) are received and added, and the latch 108 outputs the sharp signal S (14,0) + S.
Latch (14,1) and return it to the adder 107. Repeat this,
The adder 107 and the latch 108 receive the sharp signal S (14,
0), 〜, S (14,14) are sequentially cumulatively added. When the cumulative addition for one line is completed, the latch 109 latches the cumulative result ΣS14 (ΣS14 = S (14,0) + to + S (14,14)) of the cumulative addition. This operation is repeated, and cumulative results ΣS13, ΣS12, ... For one line are sequentially output from the latch 109. The data selector 106 to the latch 109 are added until the latch 105 outputs the sharp signal S of the next line.

Since the latch 109 successively obtains the accumulated result for one line, the delay 110 to the latch 115 obtain the accumulated result of 15 consecutive accumulated results in the main scanning direction X, and the shifter 116 obtains the average value. . When the latch 109 is latching ΣS15, the latch 109 uses the preceding cumulative result ΣS1.
4, ΣS13, ..., ΣS0 have already been output. At this time,
The adder 113 and the latch 114 cumulatively add the preceding 15 accumulated results ΣS14 to ΣS0. Therefore, the latch 115 calculates the cumulative result Σ (ΣS14-
ΣS0) has already been latched. Therefore, the latch 115
, The sum of the sharp signals S of the 15 × 15 mask is obtained. This sum Σ (ΣS14 to ΣS0) is assigned to shifter 1
At 16, an arbitrary number of bits are shifted in the LSB direction for digit alignment. As a result, the filtered sharp signal S ( S = Σ (ΣS14 to ΣS0) / (15 × 1
5)) is required. The sharp signal S is the filtered sharp signal S (7,7) corresponding to the sharp signal S (7,7) when the pixel of interest is the sharp signal S (7,7).

The delay 110 has 15 storage areas, and the preceding 15 accumulated results ΣS14 to ΣS0 output by the latch 109 in the past are already stored in this storage area. When the latch 109 outputs the cumulative result ΣS15, the delay 110 stores the latest cumulative result ΣS15 and outputs the oldest cumulative result ΣS0. This cumulative result ΣS0 is made negative (−ΣS0) by the code conversion 111. The data selector 112 alternately switches the output “ΣS15” of the latch 109 and the output “−ΣS0” of the code conversion 111 and outputs the output to the adder 113. This "ΣS15" and "-ΣS0" are
Cumulative addition is performed by the adder 113 and the latch 114 (Σ (ΣS14
.About..SIGMA.S0) +. SIGMA.S15-.SIGMA.S0 = .SIGMA. (.SIGMA.S15 to .SIGMA.S1)).
As a result, in the shifter 116, the sharp signal S (8,
The filtered sharp signal S (8,7) when 7) is the target pixel is obtained. This operation is repeated, and filtered sharp signals S (8,7) , S (9,
7) , ... Can be sequentially obtained at high speed.

Here, when the dark and light portions of the original are scanned and the original has no graininess, the sharp signal S does not include a noise component (see FIG. 5A (1)). When scanning the dark part of the document, the level of the sharp signal S is low (see α1 and α2 in FIG. 5A (1)). When scanning the bright part of the document, the level of the sharp signal S is high (see α3 in FIG. 5A (1)). Then, at the edge portion of the boundary between the light portion and the dark portion of the document, the level of the sharp signal S sharply changes at the position corresponding to this edge portion (see α4 and α5 in FIG. 5A (1)). When the sharp signal S that does not include this noise component is filtered, the edge portion α4, 4 of the sharp signal S is slightly smoothed due to the filtering process.
Filtered sharp signal S at the position corresponding to α5
The edge portions β4 and β5 of can be obtained.

On the other hand, when the original has graininess, the sharp signal S includes a noise component due to this graininess (see FIG. 5B (1)). Since the level of this noise component is log conversion, the part α where the level of the sharp signal S is low is α.
It appears large at 1 and α2, and appears small at a portion α3 where the level of the sharp signal S is high. When the sharp signal S including this noise component is filtered, the noise component is averaged due to the filtering process, and the filtered sharp signal S
Almost no noise component is included in (Fig. 5B (2)
See β1, β2, β3). Further, the edge portions β4 and β5 of the filtered sharp signal S can be obtained at the positions corresponding to the edge portions α4 and α5 of the sharp signal S. Therefore, when the filtering process is performed, even if the sharp signal S includes a noise component, the filtered sharp signal S
In S , the influence of noise can be almost eliminated. However, when the level of the noise component is very large, the influence of the noise component remains, and the undulation occurs in the filtered sharp signal S having a low level due to the influence of the noise component (β6 in FIG. 5B (2)). , Β7). The filtered sharp signal S is given to the primary differential signal generation circuit 4 and the delay circuit 12.

-1st order differential signal creating circuit 4, 1st order differential signal S '. Directionally averaged 1st order differential signal S'created-The 1st order differential signal creating circuit 4 is, for example, as shown in FIG. One mask has three rectangular areas in X and three rectangular areas in the sub-scanning direction Y. The filtered sharp signal S included in this mask has a predetermined direction around X = 1, Y = 1, for example, a main scanning direction X, a sub scanning direction Y, and a main scanning direction X and a sub scanning direction Y. Direction differentiation is performed in four diagonal directions (upward to the right and upward to the left). Then, the primary differential signal S ′ in the four-direction target pixel is created. Similarly, the primary differential signal S'included in this mask is directionally averaged in four directions around X = 1 and Y = 1 to generate a primary differential signal S ' . FIG. 7 shows a specific circuit of the primary differential signal creating circuit 4 for creating the primary differential signal S ′ and the primary differential signal S ′ . In FIG. 7, the latch 117 to the adder 135 are four-direction 1
This is the part that creates the second derivative signal S ′, and is the directional averaging circuit 19
The 0, ..., Directional averaging circuit 193 is a part that creates four-direction primary differential signals S ′ .

First, the latch 117 to the adder 135 for producing the four-direction primary differential signal S'will be described. Each time the filtered sharp signal S is sent from the filter processing circuit 2, the filtered sharp signal S is latched by the latch 117. Line buffer 1
20 includes three line memories 120 ₀ to 120 ₂ . Each of the line memories 120 ₀ to 120 ₂ has a storage area for 64K pixels. The averaged sharp signal S latched by the latch 117 in this storage area is stored from the address “0” of the line memory 120 ₀ . Latch 11
The operations of 7, ..., Latch 121 operate in the same manner as the operations of the latches 101-105 of the filter processing circuit 2 described above.
Three line memories of the line buffer 120 120 ₀ to 120
_The difference is that it is done in ₂ .

When the four-direction first-order differential signals S'are created, the filtered sharp signals S are aligned by the latches 117 to 121 so as to be easily differentiated. If the filtered sharp signal S (7,7) is stored at the address "7" of the line memory 120 ₁ (see FIG. 8), another filtered sharp signal S (6,6) ,
, S (8,8) are stored in the line memories 120 _{0 to} 120 ₃ at addresses “6” to “8”, respectively. The latch 121 receives the filtered sharp signals S (6,8) , S
(7,8) , S (8,8) , S (6,7) , S (7,7) , S (8,7) , S (6,
6) , S (7,6) , S (8,6) , ... are latched for every three lines and output.

The data selector 122, ...
It is a part that differentiates each direction. This directional differentiation is performed by subtraction. The data selector 122 is filtered sharp signal S at the line memory 120 ₀ that is output from the latch 121 (6,8), S (6,7 ), S (6,
6) , ... Is output to the line 122a. Also line memory 1
Filtered sharp signal S (7, 20 ₁ , 120 ₂
8) , S (7,7) , S (7,6) , ..., S (8,8) , S (8,7) , S (8,
6) are output to the lines 122b and 122c, respectively.

The delays 123 to 125 each have a storage area for storing two filtered sharp signals S and perform a FIFO operation. The delay 126 and the delay 127 each have a storage area for storing one filtered sharp signal S. A filtered sharp signal S of one line in the sub-scanning direction Y is input to the latch 121.
If (6,6) , S (7,6) , and S (8,6) are latched (see FIG. 8), the delay 123 has filtered sharp signals S (8,8) , S ( 8) . , 7) is stored and the delay 127
The filtered sharp signal S (8,7) is stored in. The delay 124 stores filtered sharp signals S (7,8) and S (7,7) . Delay 125
The filtered sharp signals S (6,8) and S (6,7) are stored in, and the filtered sharp signal S (6,7) is stored in the delay 126. Data selector 1
Filtered sharp signal S (6,6) , S (7,
6) and S (8,6) are output, delay 123, ..., Delay 127 store them and store the oldest filtered sharp signal S (8,8) , S (7,8 ) . ) , S (6,8) , S (6,
7) and S (8,7) are output respectively. The outputs of the delay 123, ..., And the delay 126 are made negative by the code conversion 128, .about., And the code conversion 131, and are given to the adders 132, .about., And 135, respectively. The output of the delay 127 is directly given to the adder 135.

The adder 132 outputs the code conversion 128 ( -S
(8,8) ) and the output ( S (6,6) ) of the data selector 122 via the line 122a are added ( -S (8,8) + S (6,6) ).
This addition result is a downward-sloping subtraction that includes the filtered sharp signal S (7,7) corresponding to the sharp signal S (7,7) when the pixel of interest is the sharp signal. Therefore, it represents a partial differential in the downward-right direction, that is, a primary differential signal S (7,7) ′ in the downward-right direction. Adder 133
Is the output of the code conversion 129 ( -S (7,8) ) and the line 122b.
The output ( S (7,6) ) of the data selector 122 via ( ) is added ( -S (7,8) + S (7,6) ). This addition result is a subtraction that includes the filtered sharp signal S (7,7) when the sharp signal S (7,7) is the target pixel. Therefore,
Partial differentiation in the main scanning direction X, that is, the main scanning direction X
Represents the first-order differential signal S (7,7) ′ in.

The adder 134 outputs the output ( -S
(6,8) ) and the output ( S (8,6) ) of the data selector 122 via the line 122c are added ( -S (6,8) + S (8,6) ).
The result of this addition is the subtraction of the section in the upward right direction that encloses the filtered sharp signal S (7,7) when the sharp signal S (7,7) is the pixel of interest. Therefore,
The partial differential in the upward-sloping direction, that is, the primary differential signal S (7,7) ′ in the upward-sloping direction is shown. The adder 135 is
The output of the code conversion 131 ( -S (6,7) ) and the output of the delay 127 ( S (8,7) ) are added ( -S (6,7) + S (8,7) ). This addition result is the subtraction of the section in the sub-scanning direction Y that includes the filtered sharp signal S (7,7) when the sharp signal S (7,7) is the target pixel. Therefore, the partial differential in the sub-scanning direction Y, that is, the primary differential signal S (7,7) ′ in the sub-scanning direction Y is represented. This operation is repeatedly performed, and the adder 132, ...
The secondary differential signals S (7,7) ', S (7,6)', ... Are sequentially output. The four-direction primary differential signal S ′ is obtained by the directional averaging circuit.
190, ..., respectively, to the direction averaging circuit 193.

[0053] Then, the direction averaging circuit 190 to create a first-order differential signal S ', ~, described direction averaging circuit 193.
The directional averaging circuit 190 uses the first derivative signal S ′ in the downward right direction.
Is averaged in the downward-rightward direction to generate a first-order differential signal S ′ in the downward-rightward direction. Directional averaging circuits 191, 192, 193 are
Similar to the direction averaging circuit 190, a primary differential signal S ′ is created for the main scanning direction X, the downward rightward direction, and the sub-scanning direction Y. Since the directional averaging circuits 190, ..., 193 have substantially the same configuration, a specific circuit of the directional averaging circuit 190 is shown in FIG.
Shown in.

Each time the primary differential signal S ′ in the downward direction is sequentially sent from the adder 132, the primary differential signal S ′ is sequentially latched by the latch 136. Line buffer 139, similar to the line buffer 120 three line memories 139 ₀ ~, and a 139 _2. Line memory 13
9 _0, -, 139 ₂ has a storage area of 64K pixels. The primary differential signal S latched by the latch 136 in the storage area right downward direction 'is stored in the line memory 139 ₀ address "0". Latch 136, ~, Latch 1
The operation of 41 operates in substantially the same manner as the latch 101, ..., Latch 105 of the filter processing circuit 2, and the latch 117, ..., Latch 121 of the first derivative signal generation circuit 4, but the address conversion is performed by the address conversion 140. Different in points.

[0055] right down to when the direction of the first-order differential signal S (7,7) 'has been stored in the line memory 139 ₁ of address "7", the other around the first-order differential signal S (6,6 ) ', ~, S (8,
8) 'are respectively stored in the line memory 139 _{0-139 2} address "6" to the address "8". When the address generator 138 addresses the address “7”, the address conversion 140 decrements “1” to the line memory 139 ₀ and outputs it to the line memory 139 ₁ as it is. 139 for the ₂ output is incremented by "1".

In this case, the latch 141 is the line buffer 1
The first derivative signal S (6,
6) ', S (7,7)' and S (8,8) 'are latched and output. First derivative signal S (6,6) 'output from the latch 141,
S (7,7) 'and S (8,8)' are parallel / serial converted by the data selector 142, and the adder 143 and the latch 144
Are sequentially accumulated by (S (6,6) '+ S (7,7)' + S
(8,8) ′ = ΣS7 ′), the addition result ΣS7 ′ is latched by the latch 145. Since the addition result in the lower right direction is obtained, the shifter 146 shifts an arbitrary number of bits in the LSB direction for digit alignment. As a result, the first-order differential signal S in the right-down direction obtained by further averaging in the down-right direction
(7,7) ' ( S (7,7)' = ΣS7 '/ 3) is obtained. The primary differential signal S (7,7) ′ is latched by the latch 147.

The direction averaging circuits 191, 192, and 193 differ only in that the address conversion is converted by the address conversion 140 into the main scanning direction X, the upward right direction, and the sub-scanning direction Y. Therefore, the direction averaging circuit 191
Is a primary differential signal S (7,7) ' ( S (7,7)' = (S (7,6) '+ S
(7,7) '+ S (7,8)') / 3 = ΣS7 '/ 3) is output. The directional averaging circuit 192 further calculates the right-upward first-order differential signal S (7,7) ′ ( S (7,
7) ' = (S (6,8)' + S (7,7) '+ S (8,6)') / 3 = ΣS7 '
/ 3) is output. The direction averaging circuit 193 further averages the direction in the sub-scanning direction Y in the sub-scanning direction Y to obtain a primary differential signal S (7,7) ' ( S (7,7)' = (S (6,7) '). + S (7,7) '+ S
(8,7) ') / 3 = ΣS7' / 3) is output.

This operation is repeated and the directional averaging circuit
190, ..., From the direction averaging circuit 193, the primary differential signal S in the downward right direction, the main scanning direction X, the upward right direction, and the sub scanning direction Y.
(7,7) ' , S (7,6)' , ... Are sequentially output.

Here, the primary differential signal S'represents the slope of the filtered sharp signal S. Therefore, when the sharp signal S does not include a noise component (see FIG. 5A).
(See (1)) is the sharp signal that has been filtered.
Convex upward at a position corresponding to the edge portion β4 of S (see γ4 in FIG. 5A (3)) and downward at a position corresponding to the edge portion β5 (see FIG. 5A).
It is possible to obtain the first-order differential signal S ′ of A (3) (see γ5). Further, when the directional average of the first-order differential signal S ′ is further obtained, it becomes slightly gentle, but the convex portion γ of the first-order differential signal S ′ is obtained.
It is possible to obtain a direction-averaged first-order differential signal S ′ that is convex upward (see δ4 in FIG. 5A (4)) and downward convex (see δ5 in FIG. 5A (3)) at a position corresponding to 4. The convex portions γ4 and γ5 of the primary differential signal S ′ and the convex portions δ4 and δ5 of the primary differential signal S ′ also correspond to the positions of the edge portions α4 and α5 of the sharp signal S, respectively.

On the other hand, even when the sharp signal S includes a noise component (see (1) in FIG. 5B), it corresponds to the edge portion β4 of the filtered sharp signal S with almost no effect on the noise component. To the position where
B (3) (see γ4) and a downwardly convex (see γ5 in FIG. 5B (3)) first-order differential signal S ′ at a position corresponding to the edge portion β5 can be obtained. Further, when the directional average of the first-order differential signal S ′ is further obtained, it becomes slightly gentle, but the first-order differential signal S ′ is obtained.
Direction average of the upward convex (see δ4 in FIG. 5A (4)) and the downward convex (see δ5 in FIG. 5A (3)) at the position corresponding to the convex portion γ4 of 1
The second derivative signal S ′ can be obtained. However, when the level of the noise component is extremely large, slight levels of convex portions γ6 and γ7 occur at positions corresponding to the undulations β6 and β7 of the sharp signal S. Therefore, this first derivative signal S '
Is further averaged, the convex portions γ6 and γ7 are averaged so that the first-order differential signal S ′ is almost not included, and even if it is included, the level becomes a slight level (FIG. 5B).
(See δ6 and δ7 in (4)). This four-direction primary differential signal S '
Is a lookup table 148a of the dead zone processing circuit 6,
To 148d (see FIG. 10).

-Delay circuit 12-The averaged sharp signal S output from the filter processing circuit 2 is applied to the delay circuit 12. Delay circuit
12 is for adjusting the timing of the averaged sharp signal S and giving it to the emphasized signal generating circuit 10 in principle. It is also used for various processes in the dead zone processing circuit 6 and the emphasized signal generation circuit 10.

FIG. 12 shows a specific circuit of the delay circuit 12. In the line buffer 120 of the primary differential signal generation circuit 4, a delay of two lines of pixels occurs in the main scanning direction X from the input to the output of the averaged sharp signal S of the target pixel (64K ×). 2). In the line buffer 139, a delay of two lines of pixels occurs in the main scanning direction X from the input to the output of the primary differential signal S ′ of the target pixel (64K × 2). Therefore, when the pixel of interest is output from the direction averaging circuits 190 to 190, a delay of 4 lines of pixels occurs in the main scanning direction X (64K × 4). Therefore, the buffer memory 184
Is, four of the line memory 184 _0, ~, it is equipped with a 184 _3.
Each of the line memories 184 ₀ to 184 ₃ has a storage area for 64K pixels.

When the averaged sharp signal S (7,7) is latched in the latch 117 of the first-order differential signal generating circuit 4,
The averaged sharp signal S (7,7) is latched by the latch 182 of the delay circuit 12. Then, as the line memory of the line buffer 120 advances two lines in the main scanning direction X, the line memories 184 ₀ and 184 of the buffer memory 184 are moved.
Go on 2 lines of ₁ . When the primary differential signal S (7,7) 'is stored at the address "7" of the line memory 139 ₀ of the line buffer 139, the sharp signal S (7,7) becomes the address of the line memory 184 ₂ of the buffer memory 184. It is stored in "7". Then, the line memory of the line buffer 139 is set in the main scanning direction X.
2 lines to the line memories 184 ₂ and 184 ₃ of the buffer memory 184. First derivative signal S (7,
When 7) 'is latched in the latch 147 of the direction averaging circuit 190, ..., 193, the sharp signal S (7,7) is latched in the latch 185. The sharp signal S (7,7) is given to the look-up tables 148a to 148d of the dead zone processing circuit 6 (see FIG. 10). Therefore, the look-up tables 148a to 148d of the dead zone processing circuit 6 are simultaneously provided with the sharp signal S (7,7) and the primary differential signal S (7,7) ' of the corresponding pixel. Even in the edge emphasis signal generation circuit 8 described later, 4 times in the main scanning direction X from the input to output of the corresponding pixel of interest.
A line pixel delay occurs (64K × 4: see FIG. 14). Therefore, the buffer memory 188 includes four line memories 188 ₀ to 188 ₃ . Each line memory
Each of 188 ₀ to 188 ₃ has a storage area for 64 K pixels. Like the latches 182 to 185, the latches 186 to 189 form delays of four lines of pixels in the main scanning direction X (64K × 4). Therefore, the emphasized signal generation circuit 10 sends the sharp signal S of the corresponding pixel.
(7,7) and the first derivative signal S (7,7) ' are given at the same time.

[0064] - dead zone processing circuit 6, the dead zone treated primary differential signal S 'delta created - dead zone processing circuit 6, as shown in FIG. 10, the look-up tables 148a, ~, equipped with 148d, 4 directions of dead vs. Perform each process. Look-up table 148a, ~, 1
The averaged sharp signal S and the four-direction primary differential signal S ′ from the delay circuit 12 are applied to 48 d. The sharp signal S and the first derivative signal S ′ are
The same pixel of interest (eg, S (7,7) and S ( 7,7)
7) ' ). The lookup tables 148a to 148d are composed of, for example, a ROM and the like, and are provided with a plurality of tables in which a plurality of characteristic data are stored in advance. The look-up table 148a, ~, 148d as a function of the input and the sharp signal S are averaged primary differential signal S ', select one of the tables of the plurality of tables, of the selected table Output characteristic data. This characteristic data includes a dead zone Δ (see FIGS. 5A (3), (4), FIG. 5B (3), (4)).
Is provided. Primary differential signal S 'level of the case located between the dead zone delta, the signal of level "0", i.e., the dead zone treated primary differential signal S' and outputs the delta. 1
Next the differential signal S 'if the level of which exceeds the dead zone delta, a signal with a portion exceeding, i.e., the dead zone treated primary differential signal S' and outputs the delta.

When the first-order differential signal S ' (see FIG. 5A (4)) when the sharp signal S does not include a noise component is input to the look-up tables 148a to 148d,
The convex portion δ4 of the first-order differential signal S ′ at a level lower than the dead zone Δ,
The part of δ5 is cut. Then, only the high level portion is output (see ε4 and ε5 in FIG. 5A (5)). Therefore, the dead zone processed primary differential signal S ′ Δ has convex portions ε at positions corresponding to the convex portions δ4 and δ5 of the primary differential signal S ′ , respectively.
4 and ε5 signal.

Further, even when the first-order differential signal S ' (see FIG. 5B (4)) when the sharp signal S includes a noise component is input to the lookup tables 148a to 148d, the dead zone is generated. First-order differential signal lower than Δ
The convex portions δ4 and δ5 of S ′ and the undulations δ6 and δ7 are cut. Then, only the high level portion is output (see ε4 and ε5 in FIG. 5B (5)).

Therefore, the dead band processed first-order differential signal
S ′ Δ becomes signals of the convex portions ε4 and ε5 at the positions corresponding to the convex portions δ4 and δ5 of the primary differential signal S ′ , respectively. This eliminates the influence of the noise component even when the level of the noise component is very large. The dead zone treated primary differential signal S 'delta convex portion epsilon] 4, Ipushiron5 is primary differential signal S' of the convex portion [gamma] 4, Ganma5,1 order differential signal protrusion δ4 of S ', .DELTA.5
And the positions of the edge portions α4 and α5 of the sharp signal S respectively (FIGS. 5A (1), (2), (3), FIG. 5B (1),
(See (2) and (3)).

Here, the look-up table 148 shows the sharp signal S of the corresponding pixel together with the first-order differential signal S ′.
a, ..., 148d are given because the input data increases due to both signals, and by using this input data as a function, the convex portion ε of the dead zone processed first-order differential signal S ′ Δ
This is because the shapes of 4 and ε5 can be changed in various ways.
By changing the value of the characteristic data, for example, as shown in FIG.
As shown in (1), the heights and widths of the convex portions ε4 and ε5 of the primary differential signal S ′ Δ can be freely changed, and the shape of the curve can be freely changed. Also, the main scanning direction X
It is also possible to correct the difference (square root of 2) between the distance in the sub-scanning direction Y and the distance in the lower right direction and the lower right direction. Further, as shown in FIGS. 11 (2) and 11 (3), it may be triangular. However, such first-order differential signal after dead zone processing
Various changes in the shapes of the convex portions ε4 and ε5 of S ′ Δ can be sufficiently performed with only the data of the first-order differential signal S ′ . Therefore,
The sharp signal S is not always necessary, and
Basically, the primary differential signal S ′ due to the influence of noise components
It is sufficient if the undulations δ6 and δ7 can be removed. The four-direction dead zone processed first-order differential signal S ′ Δ is given to the edge emphasis signal generation circuit 8.

[0069] - edge enhancement signal generation circuit 8, 2-order differential signal S "· direction averaging secondary differential signal S" Create - a concrete circuit of the edge enhancement signal generation circuit 8 shown in FIG. 13.

First-order differential signal which has been subjected to dead band processing in four directions
S ′ Δ is given to the second-order differentiation / direction averaging circuits 194, 195, 196, and 197 of the edge-enhanced signal generation circuit 8, respectively. Two
The second derivative / direction averaging circuits 194, to 197 are the first derivative signals S ′ that have been subjected to the dead zone processing and are included in the mask of FIG. 6 described above.
.DELTA. Is further differentiated into four directions centering around X = 1 and Y = 1. Then, the secondary differential signal S ″ in the target pixel is created. Similarly, the secondary differential signal S ″ contained in this mask is 4 with X = 1, Y = 1 as the center.
And direction average direction, so as to create a direction averaging secondary differential signal S "which is the direction averaged. The data selectors 170, ~, latch 175, 4-way direction averaging secondary differential signal S ” Is averaged to create an edge enhancement signal.

First, a description will be given of the secondary differential / directional averaging circuits 194, ..., 197 for creating the directional averaged secondary differential signal S ″ . The secondary differential / directional averaging circuits 194, ..., 197 are substantially the same. The second derivative / direction averaging circuit 19
A specific circuit of 4 is shown in FIG. In FIG. 14, the latch 149, ..., The adder 157 is the part that creates the second order differential signal S ″, and the latch 158, ..., The latch 169 is the part that creates the directional averaged second order differential signal S ″ .

[0072] 1 of the look-up is output from the table 148a was right down the direction of the dead zone processed order differential signal S 'Δ
Are sequentially transmitted, the insensitive pair processed first-order differential signal S ′ Δ is sequentially latched in the latch 149. Like the line buffer 139, the latch 149 has three 6
Line memories 152 ₀ , ..., Having a storage area for 4K pixels
And a 152 _2. The latches 149, ..., Latch 154 operate in the same manner as the latches 136, ..., Latch 141 of the directional averaging circuit 190 of the primary differential signal generation circuit 4.

[0073] right-down when the first-order differential signal of the direction of the dead zone the processed S 'delta is stored in the line memory 152 ₁ of address "7" is first derivative signal of the other around the dead zone treated S (6,6) is 'Δ, ~, S (8,8 )' Δ, respectively stored in the line memory 152 _{0-152 2} address "6" to the address "8". When the address generator 151 addresses the address “7”, the address conversion 153
The line memory 152 ₀ is decremented by “1”, the line memory 152 ₁ is output as it is, and the line memory 152 ₂ is incremented by “1” and output.

In this case, the latch 154 is the line buffer 1
The first-order differential signal S (6,6) ' Δ, S (7,7)' Δ, S (8,8) ' Δ output from 52, which has been subjected to the dead zone processing in the downward right direction, is set to 3
Latch and output. The primary differential signal S (6,6) ' Δ, S (7,7)' Δ, S (8,8) ' Δ output from the latch 141
Is the primary differential signal S (6,
6) ' Δ, S (7,7)' Δ, S (8,8) ' Δ first derivative signal of Δ
S (6,6) ' Δ and S (8,8)' Δ are selected. First derivative signal
S (6,6) ′ Δ is directly given to the adder 157. The primary differential signal S (8,8) ′ Δ is negative ( −S (8,
8) ′ Δ) and given to the adder 157. The adder 157 is
Output of data selector 155 ( S (6,6) ' Δ) and code conversion 15
The output of 6 ( -S (8,8) ' Δ) is added ( S (6,6)' Δ
-S (8,8) ' Δ). The result of this addition is the sharp signal S (7,
When the pixel 7) is the target pixel, the subtraction is performed in the downward-sloping direction including the corresponding first-order differential signal S (7,7) 'Δ. Therefore, the partial differential in the downward-right direction, that is, the secondary differential signal S (7,7) "(S (7,7)" = in the downward-right direction,
S (6,6) 'Δ-S (8,8)' Δ). This operation is repeated, and the second derivative signal S (7,
6) ”, S (7,5)”, ... Are sequentially output.

Averaged second derivative signal in the downward right direction
In the latches 158, ..., Latch 169 that generate S ″ , the latch 158 outputs the secondary differential signals S (7,6) ″, S (7,5) ″, ... The line buffer 161 includes three line memories 161.
_0, ~, it is equipped with a 161 _2. Each line memory 161 ₀ , ~, 1
61 ₂ has a storage area of 64K pixels. Have been right-down direction of the secondary differential signal S "is latched by the latch 158 in the storage area. Latch 158 is stored in the line memory 161 _0, ~, operation of the latch 169, the primary differential signal generating circuit 4 Latch 136 of direction averaging circuit 190, ~, latch
Works the same as 141.

[0076] If the address generator 160 is addressing the address "7", the address conversion 162, to the line memory 161 ₀ decremented "1", the line memory
It is output as it is to 161 ₁ and line memory 161 ₂
Is incremented by 1 and output. Therefore, it is addressed in a downward right direction. Right-down direction of the secondary differential signal S "when are stored in the line memory 161 ₁ of address" 7 "is the other surrounding secondary differential signal S (6,6)", ~, S (8 , 8) ”is the line memory 161 ₀ ~
Respectively stored in the 161 _second address "6" to the address "8".

In this case, the latch 163 is the line buffer 1
The second derivative signal S (6,
6) ", S (7,7)" and S (8,8) "are latched and output. The secondary differential signal S (6,6)" output from the latch 163,
The data selector 164 performs parallel / serial conversion on S (7,7) "and S (8,8)" to adder 165 and latch 166.
Is sequentially accumulated by (S (6,6) "+ S (7,7)" + S
(8,8) "= ΣS7"), the addition result ΣS7 "is latched in the latch 167. Since the addition result in the downward right direction is obtained, the shifter 168 shifts an arbitrary number of bits in the LSB direction for digit alignment. As a result, the second-order differential signal S in the downward-rightward direction averaged in the downward-rightward direction is further obtained.
(7,7) " ( S (7,7)" = ΣS7 "/ 3) is obtained. This secondary differential signal S (7,7)" is latched by the latch 169.

The second derivative / direction averaging circuit 195, ..., 1
The only difference in 97 is that the addressing is converted into the main scanning direction X, the upward right direction and the sub-scanning direction Y by the address conversions 153 and 162. Therefore, the secondary differential / direction averaging circuit 195 creates the secondary differential signal S (7,7) ″ in the main scanning direction X and further averages the direction in the main scanning direction X to obtain the secondary differential signal in the main scanning direction X. S (7,7) ” ( S (7,7)” ＝
(S (7,6) ”+ S (7,7)” + S (7,8) ”) / 3 = ΣS7” /
3) is output. The second order differential / direction averaging circuit 196 creates a second order differential signal S (7,7) ”in the right upward direction, and further performs a direction average in the right upward direction.
S (7,7) " ( S (7,7)" = (S (6,8) "+ S (7,7)" + S
Outputs (8,6) ”) / 3 = ΣS7 ″ / 3). The secondary differential / direction averaging circuit 197 creates a secondary differential signal S (7,7) ″ in the sub-scanning direction Y, and further averages the secondary differential signal S (7,7) ″ in the sub-scanning direction Y. S (7,7) ” ( S (7,7)”
= (S (6,7) ”+ S (7,7)” + S (8,7) ”) / 3 = ΣS7”
/ 3) is output.

This operation is repeated, and the second-order differentiation / direction averaging circuit 194, ... From the second-order differentiation / direction averaging circuit 197, the direction average of the downward rightward direction, the main scanning direction X, the upward rightward direction, and the sub-scanning direction Y. Second derivative signal S (7,7) " , S (7,6)" ,
... are sequentially output.

-Creation of Edge Enhancement Signal Se " -The data selector 170 to the latch 175 will be described (FIG. 1).
3). The four-direction directional averaged second-order differential signal S (7,7) ” output from the latch 169 of the second-order differential / direction averaging circuit 194, ...
Parallel / serial conversion is performed at 70, and cumulative addition is performed by the adder 171 and the latch 172 ( S (7,7) ” + S (7,7)” +
S (7,7) ” + S (7,7)” = Σ S (7,7) ” ), addition result ΣS
7 "is latched in the latch 167. Since the addition result of each direction component is obtained, the shifter 174 shifts an arbitrary number of bits in the LSB direction for digit alignment. As a result, the edge emphasis signal having no direction component is obtained. Se (7,7) ” ( S
e (7,7) ” = ΣS (7,7)” / 4) is obtained. The edge emphasis signal Se (7,7) ″ is latched by the latch 175.
This operation is repeatedly performed, and the edge enhancement signals Se (7,7) " , Se (7,6)" , ... Which have no directional component are sequentially output from the latch 175.

Here, the secondary differential signal S ″ represents the slope of the dead zone processed primary differential signal S ′ Δ.
When the sharp signal S does not include a noise component (see FIG. 5A (1)), the dead zone-processed first-order differential signal S ′ Δ is projected upward with a narrow width at a position corresponding to the convex portion ε4. And a second-order differential signal S ″ of a steep concave and convex shape (see ζ4 in FIG. 5A (6)) can be obtained. Further, at a position corresponding to the convex portion ε5, a downward convex and upward with a narrow width can be obtained. It is possible to obtain a secondary differential signal S ″ of a convex and steep unevenness (see ζ5 in FIG. 5A (6)). Further averaging the quadratic differential signal S ″,
Somewhat gentle, but the uneven portion ζ4 of the secondary differential signal S ″
The sharp unevenness with a narrow width at the position corresponding to (Fig. 5A (7) η4
), The direction-averaged second-order differential signal of a steep unevenness with a narrow width (see η5 in FIG. 5A (7)) at a position corresponding to the unevenness ζ5
S " can be obtained.

On the other hand, even when the sharp signal S includes a noise component (see FIG. 5B (1)), the dead zone-processed primary differential signal S ′ Δ is narrow at a position corresponding to the convex portion ε4. A steep unevenness that is convex upward and downward in width (Fig. 5B).
It is possible to obtain a second derivative signal S ″ of (6) in (6). Further, a steep unevenness having a narrow width and a convex shape with a narrow width at a position corresponding to the convex portion ε5 (FIG. 5B (6)). The second derivative signal S ″ of (see ζ5) can be obtained. This second derivative signal S "
Further, when the direction average is further obtained, it becomes slightly gentler, but it corresponds to the uneven portion ζ5 which is steep unevenness with a narrow width (see η4 in FIG. 5B (7)) at a position corresponding to the uneven portion ζ4 of the second derivative signal S ″. It is possible to obtain the direction-averaged second-order differential signal S ″ of steep irregularities (see η5 in FIG. 5B (7)) having a narrow width at the position.

The uneven portions ζ4 and ζ5 of the secondary differential signal S ″ and the uneven portions η4 and η5 of the directional averaged secondary differential signal S ″ are the convex portions ε4 and ε5 of the dead zone processed primary differential signal S ′ Δ. The positions of the convex portions γ4 and γ5 of the primary differential signal S ′ , the convex portions δ4 and δ5 of the primary differential signal S ′ , and the edge portions α4 and α5 of the sharp signal S are also corresponded (FIG. 5A (1), (2), (3), (4), see FIG. 5B (1), (2), (3), (4)). Here, the reason why the secondary differential signal S ″ is further averaged is to eliminate this effect even when the secondary differential signal S ″ is still affected by the noise component. The edge emphasizing signal Se ″ generated by the directional averaged second derivative signal S ″ is supplied to the emphasizing signal generating circuit 10.

-Enhanced signal generation circuit 10, enhanced signal generation-The enhanced signal generation circuit 10 is provided with a lookup table 176, ..., a latch 181, and a subtractor 199 as shown in FIG. The processed sharp signal S and the edge emphasis signal Se ″ are calculated. The filtered sharp signal S is supplied to the look-up tables 176 and 177 and the subtractor 199. The edge emphasis signal Se ″ is looked up. And the subtractor 199. Incidentally, the sharp signal S and the edge enhancement signal Se "is of the same pixel of interest (e.g., S (7, 7) and S e (7,7)").

The look-up tables 176 and 177 are composed of, for example, a ROM, and are provided with a plurality of tables in which a plurality of characteristic data are stored in advance. The look-up table 176 selects one of the plurality of tables by using the averaged sharp signal S and the edge emphasis signal Se ″ as input functions, and outputs the characteristic data of the selected table. For example, this characteristic data is obtained by comparing the convex portion and the concave portion η4 on the concave portion η4 of the edge emphasis signal Se ″.
With respect to the upward convex portion of 5 (see FIGS. 5A (7) and 5B (7)), it is designed to be very large on the negative side.
The convex portion below the concave and convex portion η4 and the convex portion below the concave and convex portion η5 of the edge enhancement signal Se ″ are set to be slightly larger than the positive side. The output comparison result is the edge enhancement signal.
Invert Se " ( -Se" ) and invert the edge enhancement signal S
The negative part of e ″ is made large and the positive part is made small. This is because the margin of the dynamic range of the sharp signal S when scanning a bright original portion is small and a dark original portion is scanned. This is because the margin of the dynamic range is large.

The look-up table 177 is for creating a halftone dot signal, and compares the averaged sharp signal S with the characteristic data as a function of the input. Comparison result output from lookup table 176, edge enhancement signal
Se ″ ( −Se ″ ) and the comparison result output from the look-up table 177, the averaged sharp signal S is latched by latches 178 and 179, respectively, and added by the adder 180 ( S − Se ″). ). The addition result ( S - Se " ) output from the adder 180, that is, the emphasized signal ( S - Se" ).
Are latched in the latch 181.

Here, when the sharp signal S does not include a noise component (see FIG. 5A (1)), the edge portion β4 (see FIG. 5A (2)) and edge of the averaged sharp signal S are detected. Emphasis has been made on the steep irregularities (see θ4 in FIG. 5A (8)) of a large convex downward and a small convex upward with a narrow width at a position corresponding to the concave / convex portion η4 (see FIG. 5A (7)) of the emphasis signal Se ″ . A signal ( S - Se " ) can be obtained. Also, the edge part β5
And a steep unevenness with a small upward convex and a large downward convex with a narrow width at a position corresponding to the uneven portion η5 (see θ5 in FIG. 5A (8)).
Of the enhanced signal ( S - Se " ) can be obtained.

On the other hand, even when the sharp signal S includes a noise component (see FIG. 5B (1)), the edge portion β4 (see FIG. 5B (2)) of the averaged sharp signal S and the edge emphasis are performed. Concavo-convex part η4 of signal Se ” (see Fig. 5B (7))
It is possible to obtain an emphasized signal ( S − Se ″ ) of steep irregularities (see θ4 in FIG. 5B (8)) of a large convex downward and a small convex upward with a narrow width at a position corresponding to the edge. Part β5
And a steep unevenness with a small width upward and a large height downward with a narrow width at a position corresponding to the uneven portion η5 (see θ5 in FIG. 5B (8)).
Of the enhanced signal ( S - Se " ) is obtained. Therefore, the contour is enhanced with a narrow width. The uneven portions θ4 and θ5 of the enhanced signal ( S - Se" ) are quadratic. Concavo-convex portions ζ4 and ζ5 of the differential signal S ″ , convex portions ε4 and ε5 of the dead zone processed first-order differential signal S ′ Δ, convex portions γ4 and γ5 of the first-order differential signal S ′, and convex portions of the first-order differential signal S ′ The positions of δ4, δ5 and the edge portions α4, α5 of the sharp signal S are also corresponded (FIGS. 5A (1), (3), (4), (5), (6), FIG. 5B (1),
(3), (4), (5), (6)).

The sharp signal S of the corresponding pixel together with the edge emphasis signal Se ″ is used for the lookup table 176.
The input data is increased by both signals, and the shapes of the concavo-convex portions θ4 and θ5 of the emphasized signal ( S − Se ″ ) can be variously changed by using the input data as a function. By changing the value of the characteristic data, for example, as shown in FIG. 16, the heights and widths of the concavo-convex portions θ4 and θ5 of the emphasized signal ( S − Se ″ ) can be freely changed, and The shape of the curve can be changed freely. In addition, the difference between the main scanning direction X and the sub-scanning direction Y and the distance between the right lowering direction and the right lowering direction (2
You can also modify the square root of.

[0090] On the other hand, the subtracter 199, simply subtract is performed-enhanced signal (S - Se ") is created in this-enhanced signal. (S - Se" in), a look-up table 176
Although it is not possible to perform the processing performed in step 1, outline enhancement can be performed with a narrow width.

As another embodiment of the present invention, FIG.
As shown in (1), the sharp signal S may be given to the emphasized signal generating circuit 10 via the delay circuit 12 as it is without averaging, and the emphasized signal (S− Se ″ ) may be generated. In this case, the delay circuit 12 may increase the delay by 7 lines in the main scanning direction.
As shown in FIG. 17 (2), the primary differential signal S ′ output from the primary differential signal generating circuit 4 may be directly applied to the edge emphasis signal generating circuit 8.

Further, although the averaging is performed using a 15 × 15 mask and a 3 × 3 mask, averaging may be performed using masks of other sizes and shapes. Further, although the simple averaging is performed, the weighted averaging may be performed.

Although the directional differentiation is performed in four directions, the directional differentiation may be performed in two directions.

Further, in the primary differential signal creating circuit 4, the primary differential signal S'is further averaged to create the primary differential signal S ' . However, the primary differential signal creating circuit is not averaged here. You may make it output 4th-order differential signal S '.

[0095]

According to the contour enhancing method of the first aspect and the contour enhancing apparatus of the fourth aspect, the sharp signal and the sharp signals of the peripheral pixels around the predetermined range of the target pixel are filtered to filter the target pixel. A processed sharp signal is created, the filtered sharp signal is directionally differentiated in a predetermined direction, a primary differential signal in the pixel of interest is created, and the primary differential signal is again differentiated in a predetermined direction to be focused. A secondary differential signal in the pixel is created, the secondary differential signal in each direction is averaged to create an edge enhancement signal, and the filtered sharp signal and the edge enhancement signal are calculated to obtain the enhanced signal in the pixel of interest. I am trying to create it.

Therefore, the enhanced signal has peaks at both ends of the dark-to-light and light-to-dark transitions of the image. Moreover, since the slope of the changing portion is steep and there is no step, noise is prevented from being emphasized, and the degree of contour emphasis is not lowered, and a double or triple contour may be formed. Absent.

In the contour emphasizing method according to claim 2 and the contour emphasizing device according to claim 5, the sharp signal and the sharp signal of the peripheral pixels around the predetermined range of the target pixel are filtered, and the target pixel is filtered. A sharp signal is created, the filtered sharp signal is directionally differentiated in a predetermined direction to create a first-order differential signal in the pixel of interest, the first-order differential signal is subjected to dead zone processing, and the dead zone-processed primary in the target pixel is performed. A differential signal is created, the first-order differential signal that has been subjected to the dead zone processing is directionally differentiated again in a predetermined direction to create a second-order differential signal in the target pixel, and the second-order differential signal in each direction is averaged to obtain an edge emphasis signal. To calculate the sharpened signal or the filtered sharpened signal and the edge enhanced signal to create the enhanced signal in the pixel of interest. There.

Therefore, even if a noise component is included in the secondary differential signal, this noise component is removed and the noise resistance is improved.

The contour emphasizing method according to claim 3 and the contour emphasizing apparatus according to claim 6 are the contour emphasizing method according to claim 1 or 2, and the contour emphasizing apparatus according to claim 4 or 5, wherein the filtered sharp signal is directed. The differentiated signals are further subjected to direction average in a predetermined direction to obtain a primary differential signal in the pixel of interest.

Therefore, even if a noise component is included in the first-order differential signal, this noise component is removed and the noise resistance is improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a contour enhancement device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a mask for averaging.

FIG. 3 is a diagram showing a specific circuit of a filter processing circuit 2.

FIG. 4 is a diagram showing a line buffer 104.

5 is a diagram showing waveforms of respective signals output from the filter processing circuit 2, the first-order differential signal generation circuit 4, the dead zone processing circuit 6, the edge emphasis signal generation circuit 8, and the emphasized signal generation circuit 10. FIG.

FIG. 6 is a diagram showing a mask for directional differentiation / direction averaging.

FIG. 7 is a diagram showing a specific circuit of the primary differential signal generation circuit 4.

FIG. 8 is a diagram showing a line buffer 120.

9 is a diagram showing a specific circuit of a direction averaging circuit 190. FIG.

FIG. 10 is a diagram showing a specific circuit of a dead zone processing circuit 6.

FIG. 11 is a diagram showing another waveform of the primary differential signal S ′ Δ that has been subjected to the dead zone processing.

FIG. 12 is a diagram showing a specific circuit of a delay circuit 12.

FIG. 13 is a diagram showing a specific circuit of an edge enhancement signal creation circuit 8.

FIG. 14 is a diagram showing a specific circuit of a second derivative / direction averaging circuit 194.

FIG. 15 is a diagram showing a specific circuit of the emphasized signal generation circuit 10.

FIG. 16 is a diagram showing another waveform of the emphasized signal ( S − Se ″ ).

FIG. 17 is a diagram showing a contour emphasizing device according to another embodiment of the present invention.

FIG. 18 is a diagram showing a conventional contour enhancement device.

FIG. 19 is a waveform chart of each signal for explaining the operation of FIG. 18.

FIG. 20 is a diagram showing another conventional contour enhancement device.

21 is a waveform chart of each signal for explaining the operation of FIG. 20. FIG.

FIG. 22 is a diagram showing another conventional contour enhancement device.

23 is a waveform diagram of each signal for explaining the operation of FIG.

[Explanation of symbols]

2 ... Filter processing circuit 4 ... Primary differential signal creation circuit 6 ... Dead band processing circuit 8 ... Edge enhancement signal creation circuit 10 ... Enhanced signal creation circuit 12 ... Delay circuit S ... Sharp signal S Filtered sharp signal S S ', S '... primary differential signal S' delta ... dead zone treated primary differential signal S ", S" ... 2-order differential signal Se "... the edge enhancement signal (S - Se") ...-enhanced signal

Claims (6)

[Claims] 1. A filtered sharp signal for a pixel of interest, which is obtained by filtering a sharp signal representing a density of a pixel of interest obtained by reading an original image and a sharp signal of peripheral pixels around a predetermined range of the pixel of interest. Is generated, and the filtered sharp signal is directional differentiated in a predetermined direction to create a first-order differential signal in the pixel of interest. A differential signal is created, the secondary differential signal in each direction is averaged to create an edge-enhanced signal, and the filtered sharp signal and the edge-enhanced signal are calculated to create the enhanced signal in the pixel of interest. A characteristic edge enhancement method. 2. A sharp signal representing the density of a pixel of interest obtained by reading an original image and a sharp signal of peripheral pixels around a predetermined range of the pixel of interest are filtered to obtain a filtered sharp signal for the pixel of interest. Is generated, the filtered sharp signal is directionally differentiated in a predetermined direction to create a first-order differential signal at the target pixel, the first-order differential signal is subjected to dead zone processing, and the first-order differential signal at the target pixel subjected to dead zone processing is generated. , The dead-zone processed first-order differential signal is directionally differentiated again in a predetermined direction to create a second-order differential signal in the pixel of interest, and the second-order differential signal in each direction is averaged to create an edge emphasis signal. Then, the sharp signal or the filtered sharp signal and the edge enhancement signal are operated to create the enhanced signal in the pixel of interest. Contour enhancement method. 3. A first-order differential signal in a pixel of interest, which is obtained by further directionally averaging a directionally differentiated version of the filtered sharp signal in a predetermined direction. Contour enhancement method. 4. A sharp signal representing the density of a pixel of interest obtained by reading an original image and a sharp signal of peripheral pixels around a predetermined range of the pixel of interest are filtered to obtain a filtered sharp signal for the pixel of interest. And a first-order differential signal creating circuit that creates a first-order differential signal in the pixel of interest by differentiating the filtered sharp signal in a predetermined direction. Edge-enhanced signal creating circuit that creates a second-order differential signal in the target pixel by averaging the second-order differentiated signals in each direction, and creates an edge-enhanced signal by filtering the sharpened signal and the edge-enhanced signal. And a emphasized signal generating circuit for calculating an emphasized signal in a pixel of interest, and . 5. A filtered sharp signal for a pixel of interest, which is obtained by filtering a sharp signal representing a density of a pixel of interest obtained by reading an original image and a sharp signal of peripheral pixels around a predetermined range of the pixel of interest. And a first-order differential signal creation circuit that creates a first-order differential signal in the target pixel by directional differentiating the filtered sharp signal in a predetermined direction, and performs a dead zone process on the first-order differential signal, A dead zone processing circuit that creates a dead zone processed primary differential signal in the pixel of interest, and the dead band processed primary differential signal is directionally differentiated again in a predetermined direction to create a secondary differential signal in the pixel of interest. Edge enhancement signal creation circuit that creates an edge enhancement signal by averaging the second derivative signal in the direction, and a sharp signal or filtered sharp An outline emphasis apparatus, comprising: an emphasis signal generation circuit that calculates a signal and an edge emphasis signal to generate an emphasis signal in a target pixel. 6. The primary differential signal generating circuit further averages the directionally differentiated ones in a predetermined direction to obtain a primary differential signal in the pixel of interest. 5 contour enhancement device.