JPH03184472A - Picture information binarizing device - Google Patents

Abstract

PURPOSE:To prevent a picture from being degraded and to binarize an attention picture element without fail by calculating error between a forecasted level and the real level of the attention picture element by a forecasting error calculating means and determining a threshold value for binarizing the attention picture element by a threshold value determining means based on this calculated forecasting error. CONSTITUTION:Picture information positioned around the attention picture element read from a memory 12 are supplied to a forecasting value calculation part 14 and at this forecasting value calculation part 14, a forecasting value x' of an attention picture element (x) is calculated. This forecasting value x' is supplied to an error calculation part 15 together with the attention picture element (x) read from the memory 12. At this error calculation part 15, the forecasting error is calculated and this calculated forecasting error is supplied to a threshold value decision part 16. At this threshold value decision part 16, a dither method is selected when the supplied forecasting error is within an allowable value set in advance, and a simple binarizing method is selected when the forecasting error is more than the allowable value set in advance. Then, the threshold value is supplied to a binarizing circuit 17 corresponding to these selected binarizing methods.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a conversion device that converts image information read by, for example, a scanner device into a binary image.

(Prior Art) For example, to binarize image information read by a scanner device, there are a simple binarization method in which the image information is binarized using a fixed threshold value, a dither method in which the threshold value is intentionally changed, and the like. simple 2
The value conversion method converts image information such as print that does not include halftones into 2
The dither method is suitable for converting image information including halftones, such as a photographic image, into binarization.

Therefore, image information including halftones can be digitized using the simple binarization method.
When converting into a value, or when image information that does not include halftones is converted into a binary image using a dither method, etc., the image quality deteriorates, and it is necessary to adopt a binary conversion method that is appropriate for the characteristics of the image information. It is.

Conventionally, whether image information read by a scanner device is an image containing halftones or not is determined by, for example, spatial filtering.

However, when spatial filtering is performed, the amount of information in the image decreases, so particularly in the case of an image including halftones, the image becomes blurred as a whole, resulting in a problem that the image quality deteriorates.

(Problems to be Solved by the Invention) This invention solves the problems related to image region discrimination using spatial filtering, and its purpose is to prevent deterioration of image quality and improve halftones. It is an object of the present invention to provide an image information binarization device capable of binarizing images that include images and images that do not include images.

[Structure of the Invention] (Means for Solving the Problems) In order to solve the above problems, the present invention converts the level of a pixel of interest out of multivalued quantized image information into the level of pixels surrounding the pixel. a prediction means for predicting from the pixel of interest; a prediction error calculation means for calculating a prediction error from the difference between the level predicted by the prediction means and the level of the pixel of interest; and a means for binarizing the pixel of interest according to the obtained prediction error. and a threshold value determining means for determining the threshold value of the threshold value.

(Function) That is, the present invention predicts the level of a pixel of interest in multivalued quantized image information by a prediction means from the levels of surrounding pixels, and compares this predicted level with the actual level of the pixel of interest. Calculate the error by using a prediction error calculation means,
Based on the obtained prediction error, the threshold value determining means determines a threshold value for binarizing the pixel of interest.
This prevents deterioration of image quality and enables accurate binarization of the pixel of interest.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

First, the principle of this invention will be explained using FIGS. 2 to 4.

If a part of multivalued quantized image information is as shown in Figure 2 (, ah, and X, and the pixel of interest as the processing target is the center The level of the pixel of interest X is predicted using the levels of pixels a - h surrounding the pixel of interest X. This predicted value X° is x' m (B+l)+c+d+6+r+g+h)7
It is determined from the average value of the levels a - h of the pixels around the pixel of interest X, as shown by g -.-rt).

Next, a pre-11PI error xd is calculated from the error between the calculated predicted value X' and the actual level of the pixel of interest X as shown in the following equation.

xd −1x−x′ l −(2)
If this prediction error xd is small, there is not much difference between the level of the pixel of interest X and the level of surrounding pixels, so the pixel of interest X is determined to be the background, for example, and the pixel of interest X is binarized by the dither method. If the prediction error is large, the difference between the level of the pixel of interest X and the surrounding pixels is large, so the pixel of interest X is determined to be a thin line, for example, and the pixel of interest X is binarized by the dither method.

Figures 3 and 4 specifically illustrate the above principle.
This shows the case where each pixel is quantized with 16 gradations.

FIG. 3 shows a case where the pixel of interest X is part of a background image.

In this case, the predicted value X° of the pixel of interest X is x' - (5+5+4+14+5+6+5)/8-4
0/8-5, and the pre-#J error is as small as x-x' I J-5-2. Therefore, the pixel of interest X has the same level as the surrounding pixels and is determined to be the background.

FIG. 4 shows a case where a pixel of interest X is part of a thin line g in a multivalued image including a thin line indicated by a - dotted broken line.

In this case, the predicted value X' of the pixel of interest X is x' = (4+4+14+l+5+2+13+3
)/8-48/8-6, and the prediction error is as large as Ix-xl=14-8-8. Therefore, since the pixel of interest X has a large level difference from the surrounding pixels, it is determined that the pixel X is part of the thin line p.

FIG. 1 shows an embodiment of an image information binarization device based on the above principle.

The scanner device 11 reads image information line by line, and the scanner device 11 outputs multivalued image information corresponding to the read image information. This multivalued image information is stored in the memory 12. This memory 12 includes, for example, four line buffers y12a, 12b, 12c, 12
d, and the image information output from the scanner device 11 is transmitted through these line buffers 12a and 12b.
.

12c and 12d. This memory 12 is provided with an address control section 13, and the address control section 13 controls writing and reading of image information to and from the line buffers 12a to 12d. That is, when performing binarization processing on image information stored in the line buffer 12b, for example, image information around the pixel of interest is sequentially read out from the line buffers 12a and 12c adjacent to the line buffer 12b. Furthermore, image information output from the scanner device 11 is written into the remaining line buffer 12d.

The image information located around the pixel of interest read out from the memory 12 is supplied to the predicted value calculation section 14, and the predicted value calculation section 14 calculates the predicted value X' of the pixel of interest X according to equation (1). be done. This predicted value X' is the memory 12
It is supplied to the error calculation unit 15 together with the pixel of interest X read out from the pixel of interest. The error calculation unit 15 calculates a prediction error according to equation (2), and supplies the calculated prediction error to the threshold determination unit 16. In this threshold determination unit 16, if the supplied prediction error is within a preset tolerance value,
If the dither method is selected and the prediction error is greater than a preset tolerance value, the simple binarization method is selected, and the threshold values corresponding to these selected binarization methods are supplied to the binarization circuit 17. Ru. This binarization circuit 17 binarizes image information by a simple binarization method or a dither method. This binarization circuit 17 is supplied with image information of the pixel of interest X from the memory 12, and this image information is binarized in accordance with the supplied threshold.

According to the above embodiment, the level of the pixel of interest is predicted from the pixels surrounding the pixel of interest, and it is determined whether or not the predicted level is close to the level of the pixel of interest. It is determined whether to convert into a value or to binarize using a simple binarization method. Therefore, since the threshold value of the pixel of interest is determined from the relationship with the surrounding pixels, it is possible to set the optimal threshold value according to the characteristics of the pixel. In the case of images such as figures and printed characters, thin lines can be expressed without disappearing.

Moreover, as in the case of spatial filtering processing,
Since the number of pixels does not decrease, deterioration of image quality can be prevented.

Note that in the above embodiment, the level of the pixel of interest is predicted from eight pixels surrounding the pixel of interest, but the present invention is not limited to this, and a method using a Markov model using a pixel immediately before the pixel of interest or a group of consecutive pixels may be used. It is also possible to predict by

Further, the threshold value is not limited to the systematic dither method and a fixed value.

It goes without saying that various other modifications can be made without departing from the gist of the invention.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide an image information binarization device that can prevent deterioration of image quality and accurately binarize a pixel of interest.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIGS. 2 to 4 are diagrams for explaining the invention in detail. 11... Scanner device, 12... Memory, 14...
・Preliminary IJj value calculation unit, 15...Error calculation unit, 16...
- Threshold determination unit, 17...binarization unit.

Claims (1)

[Claims] Prediction means for predicting the level of a pixel of interest from the levels of pixels surrounding the pixel in multivalued quantized image information; Image information characterized by comprising a prediction error calculation means for calculating a prediction error from the difference in levels, and a threshold determination means for determining a threshold for binarizing the pixel of interest according to the calculated prediction error. Binarization device.