JPH05250468A - System for reducing binary picture data - Google Patents

Abstract

PURPOSE:To suppress color collapse as much as possible without remarkably deforming the reduced picture. CONSTITUTION:When the remaining Ki being the minority cut down by the result of multiplication so as to make the run length Mi of scanning data 303 of the reduced picture integer moves more than -1 in the negative direction, the Mi is set to zero so that the deviation of the change point of the color between the original picture and the reduced picture can be suppressed within the range of one picture element. Thus, the deformation of the reduced picture can be prevented. In case where the remaining Ki does not exceed -1, the generation of color collapse can be prevented by deciding the run length Mi to value 1 even though the multiplication result is less than 1.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a binary image data reduction system, and more particularly to a system for reducing a binary image or converting a dense image into a coarse image.

[0002]

2. Description of the Related Art As a method of reducing a binary image (hereinafter referred to as an image) handled by an electronic filing apparatus or a facsimile apparatus or converting a dense image into a coarse image (hereinafter collectively referred to as reduction), "decimation" is performed. Known methods include “method”, “logical sum operation method”, “run length conversion method” and the like.

(1) The thinning method is a method of reducing the number of pixels by deleting one pixel for every several pixels of a scanning line (or sub-scanning line) of an image. In this thinning-out method, some pixels are ignored regardless of the contents of the original image, so that "blackout" and "whiteout" occur.

(2) The OR operation method is a method used to eliminate "black crushing" due to reduction, since the original image is often composed of black information on a white background. In this method, instead of simply deleting pixels as in the thinning-out method, a logical sum (black is 1 and white is 0) of the plurality of pixels is calculated in a portion converted from a plurality of pixels to one pixel, and The value is the value of one pixel after conversion. In this logical sum operation method, “white crushed” occurs instead of “black crushed”. Furthermore, the reduced image becomes darker than the original image. (3) The run-length conversion method can prevent "black crushing" and "white crushing" due to the reduction, and the reduction is performed by the following method.

That is, according to this method, the density conversion is performed not in the area of the original image but in the converted area, and the original image is expressed in run length. The run length corresponds to a continuous length of black or white. This run length is multiplied by the reduction ratio to obtain the run length in the reduced image.
In this case, the run length of the i-th run of the original image is L
If i and the run length after reduction (reduction ratio = r) are Mi, then Li and Mi must both be integers.
Here, the expression for rounding down the fractional part of the real number a is int (a) (1.1) The expression for taking out the larger number of the real numbers a and b is

Defining max (a, b) (1.2), the following equations (2.1), (2.2),
From (2.3), an integerized run length Mi of the reduced image is obtained.

K 0 = 0 (2.1) Mi = max (int (r × Li + K _i-1 ), 1) (2.2) Ki = r × Li + K _i-1 -Mi (2.3)

In equations (2.1) to (2.3), i =
1, 2, ..., I _max , and i _max indicates the total number of run lengths of the original image. Further, Ki is the residue below the decimal point due to the run length of the reduced image being integerized (≧ 1) in the reduction conversion, and the change point of the i-th color of the original image scan data and the i of the reduced image scan data. The change point of the th color shows how much the deviation has occurred.

In this run length conversion method, the equation (2.
As can be seen from 2) and (2.3), the residual K _i−1 generated in the previous run length calculation is added to the run length Li of the original image during the next run length calculation, thereby The deviation Ki is set within the range of Ki <r to prevent the inconvenience caused by the accumulation of errors after the decimal point.

Further, in the equation (2.2), the run length Mi of the reduced image is set to be Mi ≧ 1, so that it is possible to prevent color loss or crushing in the reduced image.

In this run length conversion method, however, run length Li = 1 rarely occurs continuously in a normal original image, so there is no problem. However, run length Li = 1 continues in the original image. In such a case, the following problems occur.

That is, in an artificial image created by a computer or the like, white pixels and black pixels often occur alternately. In such a case, run length L in the original image
Since i = 1 occurs continuously, the value of Ki increases in the negative direction. In this case, the run length of the corresponding reduced image is apparently enlarged, which increases the amount of deviation of the color change point between the original image and the reduced image. Therefore, the reduced image is largely deformed from the original image.

[0013]

Conventionally, when the original image in which white pixels and black pixels are alternately generated is reduced, the reduced image is largely deformed from the original image.

The present invention has been made in view of the above point, and in reducing an original image in which white pixels and black pixels are alternately generated, a reduced image is not greatly deformed, and color collapse occurs. Can be kept to a minimum 2
An object is to provide a value image data reduction method.

[0015]

Means and Actions for Solving the Problems
The run length of the original image is multiplied by the reduction ratio, and the decimal part of the multiplication result is discarded to obtain the run length of the reduced image having an integer value, and the run length of the original image is calculated when calculating the run length of the next reduced image. In the binary image data reduction method in which the run length of the original image is sequentially converted into the run length of the reduced image by adding to, 1 pixel of the original image is 1 pixel of the reduced image when the integer part of the multiplication result is zero. A means for determining the run length value of the reduced image to be one pixel so that the reduced image is converted into a pixel; and a negative decimal part value generated by determining the run length value of the reduced image to be one pixel. , A means for determining the run length of the reduced image to be zero so that one pixel of the original image is not converted into a reduced image when the value of the negative decimal part exceeds a predetermined value in the negative direction. Characterized in that it.

In this binary image data reduction system,
When the minority part of the multiplication result exceeds a predetermined value in the negative direction, the run length of the reduced image is determined to be zero, so that the amount of deviation of the color change point between the original image and the reduced image is kept within a certain range. Therefore, the deformation of the reduced image can be prevented. Also,
In a state where the minority part of the multiplication result does not exceed the predetermined value in the negative direction, the run length of the reduced image is determined to be 1 pixel even if the multiplication result is less than 1, and thus it is possible to prevent the occurrence of color collapse. Therefore, even when the reduction process of the original image in which white pixels and black pixels are alternately generated is performed, it is possible to minimize the color collapse in a range in which the reduced image is not significantly deformed.

[0017]

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

First, the principle of the binary image data reduction method according to the embodiment of the present invention will be described. The general flow of this reduction method is the same as that of the run-length conversion method described above, but in the method of the present invention, the residual Ki indicating the image shift in the run-length conversion method is {−1 <Ki <r}. It has been improved to be within the range. This range {-1 <K
Pixels that do not fall within i <r} allow color collapse as pixels that cannot be physically converted.

That is, in the run-length conversion method, the run-length (Li) of the original image is multiplied by the reduction ratio (r) and the decimal part (residual Ki) of the multiplication result is cut off to reduce the run-length of the reduced image having an integer value. (Mi) is obtained, and the minority part (residual Ki) is added to the run length of the original image during the run length calculation of the next reduced image, whereby the residual Ki is contained within the range of {Ki <r}. However, in this run length conversion method, it is always accepted that the residual Ki is {Ki <0} so that the run length (Mi) of the reduced image is always {Mi ≧ 1}.

In this embodiment, in order to reduce an artificial image made by a computer or the like without a large deformation, it is not always accepted that the residual Ki becomes {Ki <0}, but the value of the residual Ki is And the value of the residual Ki is {Ki ≦ −
1}, the run length (Mi) of the reduced image is set to zero so that one pixel of the original image is not converted into the reduced image.

The run length Mi of the reduced image is the run length Li (i = 1 to i _max , i _max of the scan data of the original image.
Is calculated by iteratively calculating the run length total number of the original image by the following equations (5.1) to (5.5) for i = 1 to i _max .

K 0 = 0 (5.1) Mi ′ = max (int (r × Li + K _i−1 ), 1) (5.2) Ki ′ = r × Li + K _i−1 −Mi ′ (5) .3) M i = Mi ′ (−1 <Ki ′), or = 0 (Ki ′ ≦ −1) ... (5.4) Ki = Ki ′ (−1 <Ki ′), or = Ki ′ + 1 ( Ki ′ ≦ −1) (5.5) The calculation contents of equations (5.1) to (5.5) are as follows.

Initial residual K which is the initial value of the shift of the reduced image
0 is zero (5.1). A reduced run length Mi ′ that does not consider the size of the deviation is obtained (5.2). A shift Ki 'is calculated when the reduced run length value is Mi' (5.3).

When Ki ′ is in the range of {−1 <Ki ′} (when the pixel shift in the scanning direction is smaller than the size of one pixel in the reduced image), the reduced run length Mi to be obtained is It is equal to Mi '(5.4). Further, the deviation Ki at that time is also made equal to Ki '(5.5).

When Ki 'is in the range of {Ki'≤-1} (when the pixel shift in the scanning direction is more than one pixel in the reduced image), the reduced image is deformed (reduced). It is determined that the image is significantly deviated in the scanning direction, and Mi = 0 is set, and the run length data is absent (5.
4). Further, at that time, since Mi = 0, the shift of the reduced image is also eliminated, and thus Ki = Ki ′ + 1 is set (5.5). FIG. 1 shows an image processing apparatus that performs image processing by the binary image data reduction method described above.

The image processing apparatus includes a CPU 101 (or DSP: Digital Signal Prossesor), a DMA controller (DMAC) 102, a program memory (P).
ROG MEMO) 103, upper interface (C
MD I / F) 104, image data memory (IMGR)
AM) 105 and image data interface (IMG I / F) 106.

The CPU 101 controls the entire apparatus and performs image reduction processing. DMA controller (DMAC) 1
02 is controlled by the CPU 101, inputs image data from the image data interface (IMG I / F) 106, and stores it in the image data memory (IMG RA).
M) 105, or conversely, the data read from the image data memory (IMG RAM) 105 is output to the image data interface (IMG I / F) 106.

Program memory (PROG MEMO)
A program for controlling the image processing operation by the CPU 101 is stored in 103. The host interface (CMD I / F) 104 inputs an instruction of processing to be executed by the image processing apparatus from the host apparatus, and outputs the execution result and the like to the host apparatus. Image data is stored in the image data memory (IMG RAM) 105. Image data interface (IMG I / F)
Reference numeral 106 inputs / outputs image data to / from external devices such as a printer, a scanner, a memory, a display, and a facsimile connected to the image processing apparatus. Next, a binary image reduction calculation by the CPU 101 will be described with reference to the flowchart in FIG.

First, the CPU 101 puts the run length of the scanning data of the original image into Li as a calculation parameter, and sets K0 = 0 as the initial value of the shift of the reduced image (step S1). Next, the CPU 101 puts an initial value 1 into the variable i indicating the run length number (step S2), and when the variable i exceeds the total run length value i _max of the original image, the calculation ends, otherwise the step A conditional branch is executed so that the calculation process of the reduced run length Mi is continued in S4 to S8 (step S3).
When i ≦ i _max , the CPU 101 reduces the reduced image run length Mi ′ when the size of the shift of the reduced image is ignored.
And the deviation Ki 'at that time,

Mi ′ = max (int (r × Li + K _i−1 ), 1) (5.2) Ki ′ = r × Li + K _i−1 −Mi ′ (5.3) Obtained (step S4).

Here, max (int (r × Li + K
_i-1 ), 1) is obtained by multiplying the run length Li of the original image by the reduction ratio r and adding the residual value K _i-1 in the run length calculation of the previous line to the multiplication result. This is an operation for comparing the integer value obtained by truncating the part with the value 1, and determining which one is larger as the reduced image run length Mi ′ when the size of the shift of the reduced image is ignored. By the calculation, when the integer value is zero, one pixel of the original image is converted into one pixel of the reduced image.

Next, the CPU 101 determines whether the deviation Ki 'obtained in step S4 is -1 or less (step S5). If the deviation Ki 'is not less than -1, Ki
Since the absolute value of ′ is small, Mi ′ obtained in step S4,
Ki 'is directly determined as the values of Mi and Ki (step S6). On the other hand, when the deviation Ki ′ is −1 or less, the absolute value of Ki ′ is large, so Mi = 0 is set, and in order to correct the deviation amount eliminated by this,
Ki = Ki'-1 is set (step S7). After this, C
The PU 101 repeats steps S3 to S7 while incrementing the run length number i by 1 (step S8).
FIG. 3 shows an example of a reduced image obtained by such a reduction method.

Here, the scan data 301 of the original image is set to 3
Scan data 30 of the reduced image reduced by a reduction ratio of / 4
The case of conversion into 3 is illustrated. Scan data 30
In 1, 303, “0” indicates white and “1” indicates black.

The first run length L1 of the scan data 301 of the original image is L1 = 3, and the corresponding run length M1 of the reduced image is obtained by substituting a value into the expression int (r × Li + K _i-1 ). Then, {(3/4) × 3 + 0 = 9/4 =
2 + 1/4}, so M1 = 2 and K1 is K
1 = 1/4.

Similarly, the run length M2 of the reduced image corresponding to the run length L2 = 1 is {(3/4) × 1.
+ 1/4 = 1}, so M2 = 1 and K2 = 0 /
4 = 0.

Further, the run length M3 of the reduced image corresponding to the run length L3 = 4 is {(3/4) × 4 + 0 =
3}, M3 = 3, K3 = 0/4 = 0, and the reduced image run length M4 corresponding to the run length L4 = 1.
Is {(3/4) × 1 + 0 = 0 + (3/4) = 1− (1
/ 4)}, M4 = 1, K4 = -1 / 4, and so on. The run length M5 of the reduced image corresponding to the run length L5 = 1 is {(3/4) × 1- (1/4) = 0 + (2 /
4) = 1- (2/4)}, M5 = 1, K5 = -2 /
4, and the run length M6 of the reduced image corresponding to the run length L6 = 1 is {(3/4) × 1- (2
/ 4) = 0 + (1/4) = 1- (3/4)}, M6
= 1 and K6 = -3 / 4.

After that, when the calculation for obtaining the run length M7 of the reduced image corresponding to the run length L7 = 1 is performed,
Since {(3/4) × 1- (3/4) = 0 = 1-1}, if M7 = 1, then K7 = -4 / 4, so here, M7 = 0, K7 =-(4 /
4) + 1 = 0.

As a result, the run length L7 in the scan data 301 of the original image is not converted, and as a result, between the color change point in the scan data 301 of the original image and the color change point in the scan data 303 of the reduced image. The deviation can be accommodated within 1 pixel.

As described above, in this embodiment, the residual Ki, which is the fractional part truncated from the multiplication result in order to convert the run length Mi of the scan data 303 of the reduced image into an integer, becomes -1 or more in the negative direction. Then, the run length Mi of the reduced image is determined to be zero, so that the amount of deviation of the color change point between the original image and the reduced image is kept within the range of 1 pixel. Therefore, the deformation of the reduced image can be prevented. Further, when the residual Ki does not exceed -1, even if the multiplication result is less than 1, the run length Mi of the reduced image is set to the value 1, so that it is possible to prevent color collapse.

Therefore, even when the reduction processing of the original image in which white pixels and black pixels are alternately generated is performed, it is possible to minimize the color crushing within a range in which the reduced image is not largely deformed.

Here, main scanning data (horizontal direction)
However, if the same process is performed on the sub-scanning data (vertical direction) after this process, any vertical and horizontal reduction can be performed.

[0042]

As described above, according to the present invention, reduction of an original image in which white pixels and black pixels are alternately generated can be performed within a range in which the reduced image is not largely deformed, and color collapse is minimized. It becomes possible to do so.

[Brief description of drawings]

FIG. 1 is a diagram showing a system configuration for realizing a binary image data reduction method according to an embodiment of the present invention.

FIG. 2 is a flowchart for explaining an image reduction operation of the embodiment.

FIG. 3 is a diagram showing an example of a reduced image converted by the image reducing method of the embodiment.

[Explanation of symbols]

101 ... CPU, 102 ... DMA controller, 103
… Program memory, 104… Upper interface,
105 ... Image data memory, 106 ... Image data interface, 301 ... Original image scan data, 303 ...
Scanned data of reduced image, Li ... Run length of original image,
Mi ... Run length of reduced image, Ki ... Residual value indicating amount of deviation between original image and reduced image.

Claims (1)

[Claims] 1. A run length of a reduced image having an integer value is obtained by multiplying a run length of an original image by a reduction scale factor and truncating a decimal part of the multiplication result, and the run length of the next reduced image is calculated by calculating the run length of the reduced image. In the binary image data reduction method in which the run length of the original image is sequentially converted to the run length of the reduced image by adding to the run length of the original image, one pixel of the original image is obtained when the integer part of the multiplication result is zero. Means for determining the run length value of the reduced image to be one pixel so that is converted into one pixel of the reduced image, and a negative decimal number generated by determining the run length value of the reduced image to be one pixel. The value of the part is examined, and when the value of the negative decimal part exceeds a predetermined value in the negative direction, the run length of the reduced image is set to zero so that one pixel of the original image is not converted into the reduced image. Binary image data reduction method characterized by comprising the that means.