JPH0336880A - Binarizing processing method - Google Patents

Abstract

PURPOSE:To improve the quality of a read picture by revising a threshold level for binarization for a high spatial frequency and a low spatial frequency. CONSTITUTION:Bit dispersion of a picture signal read by an image sensor 20 is corrected by a shading correction circuit 23 via an amplifier 21 and an A/D converter 22. The picture data is transferred to shift register 25 and a difference between the maximum value and the minimum value is decided by a maximum value decision circuit 26, a minimum value decision circuit 27 and a subtraction circuit 28. Then a comparator circuit 29 compares a threshold level of 4 picture elements included in a window with a parameter (t) given externally and in the case of <t, a Ta given externally as a threshold level T is used for the output to the comparator circuit 32. On the other hand, in the case of >=t, a value of T=a is outputted to the comparator circuit 32. The threshold level T and the picture data stored in the register 25 are compared and a binarized output result is obtained from the comparator circuit 32.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a binarization processing method in an image reading device.

[Conventional technology]

Conventionally, in image input devices such as facsimiles and scanners, in order to obtain binary image data from the immersion liquid signal of the document,
Fixed threshold methods, floating threshold methods, etc. have been used.

In the former case, a fixed threshold value is set for the image signal converted into an electrical signal, and for example, if the image signal is larger than the threshold value, 1 is assigned, and if it is smaller than the threshold value, 0 is assigned. In this case, the image quality deteriorates for originals with non-white backgrounds, such as newspapers, so a method has been devised and put into practical use that feeds back the density of the background to the threshold value. This is called the floating threshold method.

[The problem that the invention attempts to solve]

As shown in Figure 2, the dynamic range of an image (at 1m) is smaller in areas with high spatial frequencies such as letters and thin lines than in areas with low spatial frequencies.This is due to the spatial frequency of the reading system. This is due to the characteristics of the image sensor, so increasing the pixel density of the image sensor does not simply improve the resolution performance.

Even if the image signal shown in Figure 2 is binarized using a simple fixed threshold value or by determining the threshold value by detecting the background or overall density level, the image signal as shown in part a will not be generated. The image quality of areas with high spatial frequencies does not improve.

The present invention aims to eliminate such drawbacks, and its purpose is to improve the quality of the read image by changing the threshold for binarization between high and low spatial frequency parts.

[Means to solve the problem]

As shown in Figure 2, fi! ii (Set window 1 for k signal IJ, detect the difference Δ between the maximum value and minimum value of pixels included in the window, and set the threshold value for binarization using the following conditional expression. Determine T.

If Δ<t then T=Ts If Δ≧t then T=aΔ Here, T- is a fixed threshold value applied where the spatial frequency is not if'li <. t is the threshold value of the image signal △
This is a parameter to determine whether C should be detected as a meaningful signal.When the output signal is normalized to 0 to 1, it is 0. It may take a value of 1 to 0.5.

a is a coefficient that determines the threshold in areas with high spatial frequencies, and is usually 0.5, but in some cases it may be 0.25 to 0.
．． In some cases, a value of 75 is better.

〔Example〕

FIG. 1 shows an example in which the number of pixels included in the window is 4, and the image signal read by the image sensor 20 is
It passes through an amplifier 21 and is converted into an 8-bit ffi by an A/D converter 22. The image f-image signal converted into a digital signal is corrected for bit variations in the optical system and image sensor in a shading correction circuit 23. 24 is a memory for calculation. The corrected 8-bit image data for both images is transferred to a four-stage 8-bit shift register 25, and a primary comparator in which the above-mentioned Δ is determined by a maximum value determining circuit 26, a minimum value determining circuit 27, and a subtracting circuit 28. 29 compares the threshold value △ of the four pixels included in this window with t given from the outside, and if it is a △ comb, the selector 31 outputs T, also given from the outside, to the comparator 32. . On the other hand, when Δ≧t, the value obtained by multiplying Δ by a given from the outside by the multiplier 30 is outputted to the comparison ia+32. The comparator 32 compares the image data and converts the output result of binarization into fV! Ru.

〔Effect of the invention〕

According to the present invention, the quality of binarized images of high spatial frequency images, ie, thin lines such as characters and marked lines, is improved.

Furthermore, according to the present invention, a good binarized image can be obtained even if the reading optical system is slightly blurred.

[Brief explanation of drawings]

FIG. 1 is a circuit diagram showing an embodiment of the present invention. FIG. 2 is a diagram showing the concept of an image signal for explaining the present invention. that's all

Claims (1)

[Scope of Claims] (1) A method for converting the density signal of an image into a binary data string, which sets a window that sees a plurality of pixels for spatially continuous pixels, and A binarization processing method characterized by detecting a difference Δ between a maximum value and a minimum value of image data, and determining a threshold T for binarization with respect to the Δ under the following conditions. If the predetermined value t is the fixed threshold value TO for binarization and the coefficient a, then if Δ<t, then T=TO; if Δ≧t, then T=a*Δ (2) The number of pixels included in the above window is 2 ~32, and when the threshold value of the density signal is normalized to 0 to 1, the above t and a are 0.1≦t≦0.5 and 0.25≦a≦0.75, respectively. The binarization processing method according to claim 1, characterized in that: