JPS63308471A - Picture signal binarization device - Google Patents

Abstract

PURPOSE:To accurately correct an unevenness in a sesitivity for every dot and to prevent a binarization error by deciding a threshold for every dot of respective main scanning lines based on the peak value of a main and sub- scanning directions. CONSTITUTION:At the time of scanning an the picture of an original, the picture signal level of the main scanning direction is traced by a peak value detecting means 21 to detect the peak value of the main scanning direction for every main scanning line, the picture signal level of the sub-scanning direction is traced to detect the peak value of the sub-scanning direction for every dot and a threshold value deciding means 28 decides the threshold for every dot of the respective main scanning lines based on the peak values of these main and sub-scanning directions. In this case, the peak values of the main and the sub-scanning directions are changed according to the fluctuation in time of the first density or the quantity of light of an original, threshold decided based on the peak value is changed according to the fluctuation in time of the first density or the quantity of the light of the original and the binarization error can be prevented.

Description

[Detailed Description of the Invention] [Field of Application for the Purpose of the Invention (Industry-1)] The present invention provides a method for binarizing an image signal read by a one-dimensional photoelectric converter by comparing it with a threshold value. The present invention relates to an image signal binarization device.

(Prior Art) Conventionally, for example, in a facsimile machine, as shown in FIG. The configuration is such that an image signal is obtained by forming an image on the image sensor 4, and a binary signal of black and white is obtained by comparing this image No. 18 with a threshold value. In this case, the image sensor 4 has a large number of semiconductor photoelectric conversion elements arranged in a straight line, and output signals from each photoelectric conversion element (dot) are sequentially extracted from one side to the other in the main scanning direction ( The entire image of the original 1 is read by scanning in the direction in which the photoelectric conversion elements are arranged and moving the original 1 or the image sensor unit 5 in the sub-scanning direction (the direction perpendicular to the main scanning direction).

With such a configuration, two distortions are superimposed on the image signal output from the image sensor 4: shooting distortion caused by a decrease in the amount of peripheral light from the light source 2, and uneven sensitivity of each dot caused by manufacturing variations. Therefore, even when scanning a white original with equal reflectance, the output signal level of each dot is not constant, as shown in Figure 3, and the output level at both ends of the main scanning line drops due to shooting distortion in particular. becomes noticeable. Even if such an image signal is compared with a certain threshold value, it cannot be binarized accurately (a binarization error occurs), so the following two methods have been considered as countermeasures. One is, for example, Japanese Patent Application Laid-Open No. 61-12
As shown in Publication No. 175, both bright (white) and dark (black) reference signal data are read in advance from a standard manuscript, and the image signal is corrected by normalization calculation based on the reference signal data. Then, the corrected image signal is compared with a certain threshold value and binarized.

Another method, as shown in FIG. 4, is to obtain the bright and dark reference signals 6 and 7 in advance using a standard manuscript, obtain the dynamic range D of the signal for each dot, and then calculate the dynamic range D of the signal for each dot. By setting the 80% value as the threshold value 8, the threshold value 8 is made to correspond to the distortion of the image signal.

(Problems to be Solved by the Invention) Incidentally, the background density (ground color) may vary from document to document, or the amount of light from the light source may vary over time due to ripples in the power supply, etc., but the above-mentioned conventional two methods In all methods for preventing conversion errors, the threshold value is fixed regardless of changes in the background density of the original or temporal fluctuations in the amount of light. Binarization errors sometimes occurred when the signal output level fluctuated.

The present invention attempts to solve these problems, and therefore, its purpose is to change the threshold value to follow changes in the ground density of the original, temporal changes in the amount of light, etc. , prevent binarization errors as much as possible! 1- To provide an image signal binarization device that can be used.

[Structure of the Invention] (Means for Solving the Problems) The image signal binarization device of the present invention compares the image signal obtained by reading image information with a one-dimensional photoelectric converter with a threshold value. main scanning direction peak value detection for tracking the image signal level of the one-dimensional photoelectric converter in the main scanning direction and detecting the peak value in the main scanning direction for each main scanning line; sub-scanning direction peak value detection means for tracking the image signal level of the one-dimensional photoelectric converter in the sub-scanning direction and detecting a peak value in the sub-scanning direction for each dot; threshold determining means for determining the threshold value for each time based on the peak value in the main scanning direction and the peak value in the sub-scanning direction.

(Function) When scanning an image of a document, the image signal level in the main scanning direction is tracked to detect the peak value in the main scanning direction for each main scanning line, and the image signal level in the sub-scanning direction 7 is also tracked. to detect the peak value in the sub-scanning direction for each dot,
A threshold value is determined for each dot in each main scanning line based on these peak values in the main and reverse scanning directions. In this case, since the peak values in the main and rotational opening directions change in accordance with temporal fluctuations in the background density and light intensity of the original, the threshold value determined based on the peak value is It changes in accordance with temporal fluctuations in density and light intensity.

(Example) Hereinafter, one embodiment of the present invention will be described with reference to FIG.

However, a one-dimensional photoelectric converter for reading the image of the original,
The structures of the light source, rod lens, etc. are the same as those shown in FIG. 2 described above, so their explanation will be omitted. FIG. 1 shows a circuit block diagram of an image signal binarization device. In the figure, 11 is an input terminal to which an analog image signal υ output from a one-dimensional photoelectric converter is input, and 12 is an input terminal for the image signal. An A/D converter that converts the signal υ into a digital value V1, 13 is an image signal Vi
A normalization circuit that normalizes the
15.3 latches 1B, 17.18 and 2 line memories 19. [Consists of 0. In this case, the method of normalization calculation is as follows, assuming that the image signal Vi(n) of the n-th dot of the one-dimensional photoelectric converter is normalized to obtain a signal of size Vo(n). 1) It is expressed by the formula.

Here, A (n) is the bright (white) reference signal of the n-th dot, B (n) is the dark (black) reference signal of the n-th dot, and K is a constant. A(n),
The data of B (n) is obtained in advance as follows. First, with the light source turned off (completely dark), a signal is extracted from each dot of the one-dimensional photoelectric converter and stored in the line memory 19 as the dark fundamental signal B (n), then the light source is turned on and the same process is performed. The signal extracted from the bright reference signal A (n
). The bright reference signal A (n) may be read by making the surface of the roller that feeds the document pure white and reading the white background on the surface of the roller, or by scanning a standard document with a white background. In the normalization circuit 13, the latch 16 outputs the dark base q signal B (n) taken out from the line memory 19 to the EP-ROM 14. Historically, the EP-ROM 14 stores the level difference (A(n) - B(n)) between the bright reference signal A(n) and the dark reference signal B(n).
)) and sends it to line memory 2.
0, and a signal corresponding to the level difference (V 1 (n) - B (n)) between the image signal Vi (n) and the dark reference signal B (n) that will be input during subsequent scanning of the original. Output to latch 17. In this case, the EP-ROM 14 has inputs A(n) and B(n) in advance.

Vj(n) and output A (n) −B (n), V
1(n) -B(n) is stored as a tabulated data group, and a value corresponding to the input signal is output from the data group, and this is used to store the data in the EP-ROM 14. It has a subtraction function. on the other hand,
The EP-ROMI 5 has division and multiplication functions, and outputs the output signal of the latch 17 (V j (n) - B (n)) and the output signal of the latch 18 (A (n) - B (n)). From this, Vo(n) in equation (1) is determined and output, and the image signal is normalized by this.

On the other hand, 21 is a peak value detection circuit which serves both as a peak value detection means in the main scanning direction and a peak value detection means in the sub-scanning direction, and is composed of three latches 22, 23, 24, an EP-ROM 25, and a line memory 26. . In this embodiment, the peak value in the main scanning direction is detected as follows. Now, in one main scanning line, from the 1st dot to the nth dot
If the peak value up to 1] is Vp(n), then V
p(n) = Vo(n) (1-exp(-tar)
10Vp(n -1) ... (2) Here, V
o(n) is the inn doratro image signal (normalization circuit 13
output signal), t is the driving time of 1 de soto, and τ is the time constant. By detecting the peak value with a time constant in this manner, the influence of instantaneous fluctuations in the image signal level is eliminated. In this case, the latch 22 transfers the image signal Vo(n) from the normalization circuit 13 to the EP-RO
It is output to M25, and the latch 23 is the (n-1)th
The peak value Vp (n −1) up to de soto IT+ is expressed as E
It is output to the P-ROM 25. And EP-
The ROM25 contains pme human power V o(n), V p(n
-1) and the output Vp(n) is stored as a tabulated data group, and a value corresponding to the human input signal from the data group is output to the latch 23.27.

On the other hand, the peak value detection circuit 21 also detects the peak value in the sub-scanning direction. That is, the nth dot 1 in the main scanning line from the first row 11 to the fth row 1".
Let Vpn (I) be the peak value in the sub-scanning direction of , then Vpn (I) = Vo (n) [1-exp(-t
ar) 10Vpn ((-1)...(3)
Here, V o (n) is the %n-th image signal, t is the driving time of 1+scanning line, and τ is the time constant.

In this case, the line memory 26 stores the peak value in the sub-scanning direction output for each dot from the EP-ROM 25 for one main scanning line, and the stored contents are updated every main scanning line. . Further, the latch 24 outputs the peak values Vpn (, (-
1) is output to the EP-ROM 25.

On the other hand, 28 is a threshold value determining circuit serving as a threshold value determining means, and is composed of two latches 27, 29 and an EP-ROM 2O. In this embodiment, the threshold value Vth of the n-th dot in the main scanning line of the C-th row ((, n) is the peak value vp in the main scanning direction of the main scanning line of the (l-1)th row. , maX((-1) and 2 from the first line
J1. It is obtained from the peak value Vpn(() in the sub-scanning direction of the n-th dot in the main-scanning line up to the row) using the following equation (4).

Vth (1!, n) = Ct ・VLma
x(1!-1)+c2 .Vpn(I) (4) Here, CI+C2 is a constant. In this case, the latch 27 sets the peak value VLmax(, (-1
) to the EP-ROM 30, and the latch 29 outputs the peak value Vpn(, () in the sub-scanning direction input from the EP-ROM 25 to the EP-RoM 3o. , pre-input Vp, m
The correlation between ax(,(-1'), Vl)n(,c) and the output Vth((,n) is stored as a tabulated data group, and from this data group, human signals are A corresponding value is output to the latch 31.

On the other hand, 32 is a comparison circuit, which includes a -2 latch 31, a buffer 33 to which the image signal is input from the W layering circuit 13, and an E
It is composed of a P-ROM 34. In this case, EP
- The ROM 34 compares the image signal inputted from the buffer 33 with the threshold value inputted from the latch 29, binarizes it, and outputs the binarized signal to the register 35. 36
is an output terminal for the binary signal.

Next, the effect of the "-2 configuration" will be explained. When the operation start operation switch (not shown) is turned on, initially,
The operation mode is set to the reference signal reading mode, and the data of both the bright and dark reference signals A (n) and B (n) are read in the following manner. First, with the light source turned off (pure darkness), a signal is extracted from each dot of the one-dimensional photoelectric converter and stored in the line memory 19 as the dark reference signal B(n), and then the light source is turned on to produce pure white. Shine the roller,
The background color is read and used as a bright reference signal A (n). When reading this bright reference signal A (n), the line memory 19
The dark reference signal B (n) sequentially extracted from the latch 16
is input to the EP-ROM 14 via the bright reference signal A (
n). As a result, a signal corresponding to the level difference (A (n) - B (n)) between the bright reference signal A (n) and the dark reference signal B (n) is output from the EP-ROMI 4 for each dot. , this is line memory 20
is memorized. When the reading of the reference signal data as described above is completed, the operation mode is automatically switched to the original scanning mode, and scanning of the original is started. As a result, the original is sequentially fed in the sub-scanning direction by the roller, and the image on the original is read by the one-dimensional photoelectric converter. An analog image signal υ is extracted in time series from each photoelectric conversion element (dot) of this one-dimensional photoelectric converter, and this image signal υ
Each dot is sequentially processed in the pipeline method as follows. First, the image signal υ is sent to the A/D converter 1
2, the data is digitized and input to the EP-ROM 14. At the same time, the dark reference signal B (n) sequentially taken out from the line memory 19 is input to the EP-ROM 14 via the latch 16, and the digital image signal Vi (n
) compared to

This causes the bright reference signal A to be transmitted from EP-ROMI 4.
(n) and the image signal V1(n) (A (
A signal corresponding to n)-V1(n)) is output to the EP-ROM 15 via the latch 17. At the same time, the data (A (n) - B (n)) sequentially retrieved from the line memory 20 is manually input to the EP-ROM 15 via the latch 18, and based on the input value, the data (1)
) The image signal V o (n) normalized by the equation is output to the buffer 33 and latch 22 .

Then, the image signal Vo(n) is sent from the latch 22 to EP-
The peak value in the main scanning direction h is detected manually in the ROM 25 in the following manner. First, for each main scanning line, the image signal V o (1) of the first dot is converted to EP as it is.
- input from ROM 25 to latch 23, peak value V
p(1). Next, when the second dotted image signal Vo(2) is input to the EP-ROM 25, this image signal Vo(2) and the peak value Vp(1) output from the latch 23 are combined. Based on the equation (2) above, the peak value V p (2
) is calculated, and whether this peak value V p(2) or the latch 23
is entered and newly retained. This operation is repeated for each dot, and the output of the EP-ROM 25 at the final dot of each main scanning line becomes the peak value VIA, maX in the main scanning direction of each main scanning line, which is then output to the latch 27. be done.

Regarding the peak value in the sub-scanning direction, first, the image signal for each dot in the first main scanning line is stored as a peak value in the line memory 26 for one main scanning line. Next, move to the main scanning line of the second row 1,
The image signal of the first dot 1-1-1 is stored in the EP-ROM25.
At the same time, the peak value of the first dot 11 is taken out from the line memory 26 and inputted to the EP-ROM 25 via the latch 24, and based on the input value, the above equation (3) is used. The peak value in the sub-scanning direction of the first dot in the main-scanning line up to the second line is determined,
This peak value is newly stored in the line memory 26 and output to the latch 29. Such an operation is repeated for each dot, and the peak value Vpn of the sub-scanning force and direction is detected for each dot.

Then, for each dot of each main scanning line, in the EP-ROM 30 of the peak value detection circuit 21, the peak value Vp, max in the main scanning direction input manually from the latch 27 and the sub-scanning direction input from the latch 29 are stored. Based on the peak value Vpn of the EP-R
OM, 30 is input. In this EP-ROM30,
An image signal is inputted from the buffer 33, and this image signal is compared with a threshold value vth and binarized.

By the way, the background density of the original is dark (light reflectance is low).
If this happens, the overall image signal level will decrease accordingly, and the peak values in both the main and sub-scanning directions will become smaller. Further, if the amount of light from the light source changes due to ripples in the power supply, the image signal level changes accordingly, and the peak value in the main scanning direction changes accordingly. Considering this background, in this embodiment, when scanning an image of a document, the image signal level in the main scanning direction is tracked and the peak value in the main scanning direction is calculated for each main scanning line. At the same time, the image signal level in the sub-scanning direction is tracked to detect the peak value in the sub-scanning direction for each dot, and based on these peak values in both the main and sub-scanning directions, each dot in each main scanning line is detected. Since the threshold value is determined based on the timing, the threshold value can be changed to follow temporal fluctuations in the background density and light intensity of the original, and can be changed to follow temporal fluctuations in the background density and light intensity of the original. Value conversion errors can be prevented as much as possible.Moreover, since the peak value in the sub-scanning direction is detected for each dot and this peak value in the sub-scanning direction is added to the threshold determining factor, the sensitivity of each dot can be reduced as much as possible. The unevenness is corrected by means other than the normalization calculation described above, and the sensitivity unevenness for each dot can be corrected with high accuracy.

It should be noted that the present invention can be modified in various ways without departing from the gist, for example, other mathematical expressions may be used in place of the above-mentioned equation (4) as the function for determining the threshold value.

[Effects of the Invention] = 17 - As is clear from the above description, the present invention detects the peak value in the main scanning direction for each main scanning line, and also detects the peak value in the sub scanning direction for each dot in each main scanning line. The peak values are detected and the threshold values are determined for each dot in each main scanning line based on these peak values in both the main and sub scanning directions. This has the excellent effect of being able to be changed to follow temporal fluctuations, etc., and also being able to accurately correct unevenness in sensitivity for each dot, thereby preventing binarization errors as much as possible.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the present invention, Figs. 2 to 4 show conventional examples, Fig. 2 is a perspective view of an image sensor unit, and Fig. 3 is a block diagram showing one main scanning FIG. 4 is a diagram showing the image signal level before normalization for a line, and is a diagram for explaining a method for determining a threshold value. In the drawing, 13 is a normalization circuit, and 14 is an EP-ROM 121.
25 is a peak value detection circuit (main scanning direction peak value detection means and sub-scanning direction peak value detection means); 25 is an EP-ROM;
28 is a threshold determining circuit (threshold determining means), 30 is an EP-ROM, 32 is a comparison circuit, and 34 is an EP-ROM.

Claims (1)

[Claims] 1. An image signal obtained by reading image information with a one-dimensional photoelectric converter is binarized by comparing it with a threshold value, in which the image signal in the main scanning direction of the one-dimensional photoelectric converter main scanning direction peak value detection means for tracking the level and detecting the peak value in the main scanning direction for each main scanning line, and tracking the image signal level of the one-dimensional photoelectric converter in the sub scanning direction for each dot. a sub-scanning direction peak value detection means for detecting a peak value in the sub-scanning direction, and a sub-scanning direction peak value detecting means for detecting a peak value in the sub-scanning direction; 1. An image signal binarization device comprising: threshold determining means for determining a threshold value based on