JPH03274863A - Image processing method - Google Patents

Abstract

PURPOSE:To make an image into m-ary value satisfactorily without being affected by a noise by performing conversion from an n-ary value to the m-ary value based on the comparison result of a line drawing detecting pattern with an image data value in the prescribed window of an original image. CONSTITUTION:The energy of a noise signal that exists in isolated point fashion is reduced to 1/3 in a window of 3X3, and erroneous decision can be reduced. It means that the influence of the noise can be hard to be received as increasing the size of the window to 5X5 and 7X7. A K1 multiplication circuit 402 outputs a value in which the mean value of the window of 3X3 is multiplied by K as multivalue data. A K1 multiplication circuit 418 outputs a value in which the data of a remarked point itself is multiplied by K. A switch 420 is switched by control data from a line drawing detecting part(maximum value circuit 414). The output of the K1 multiplication circuit 418 is selected when the image seems to be the line drawing, and that of the K1 multiplication circuit 402 is outputted when it seems to be an image part.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to a method for converting an n-value image into an m-value image (n<m).

[Prior Art] Conventionally, an adaptive multi-value quantization method has been known in which a character portion and an image portion of an original image are distinguished and processing suitable for each is performed.

By the way, the conventional adaptive multilevel quantization method has been carried out using a systematic dither method. This is because in the systematic dither method, since the threshold matrix arrangement is known in advance, the threshold matrix size of the W binary image can be estimated by pattern matching with the binary image.

[Problem to be solved by the invention] For this reason, in concentration-preserving binarization methods such as error diffusion, the threshold matrix size is not determined and cannot be estimated by pattern matching. The present invention, which had the drawback that image data cannot be properly restored with multiple values, can reliably determine character parts, especially thin line character parts, in a binary image, and can also prevent noise-induced parts from being emphasized. An object of the present invention is to provide a method for converting n-value image data into m-value data.

[Means for Solving the Problems J] The present invention converts n-value image data into m-value image data (n <
m), the original image is divided into predetermined windows, and the conversion is controlled based on a comparison result between a prepared line drawing detection pattern and image data values within the predetermined windows.

[Operation] In the present invention, since the n value is converted to the m value based on the comparison result between the line drawing detection pattern and the image data value within a predetermined window of the original image, the conversion is performed without being affected by noise.
It is possible to convert images into m-values satisfactorily.

[Embodiment] FIG. 1 is a circuit diagram showing an image area discrimination circuit according to an embodiment of the present invention.

In this embodiment, as shown in FIG.
This uses a thin line detection window.

Further, FIG. 3 shows an example of a pattern for detecting a thin line portion, and here it is assumed that the point of interest is at the position A22. moreover,
In Fig. 4, the broken lines in Fig. 0 show the relationship between the original picture, the window, and the point of interest.
is set.

Then, a pixel of interest 30 is placed in the center of the 3×3 window 301.
3 will be located.

Next, each element of the image area discrimination circuit will be explained based on FIG.

The image memory 401 is made up of, for example, a hard disk device and stores image data.

Shift registers 403-405 form a 3-line buffer memory. Therefore, image memory 401
The data is outputted as AI, A2A3, .

Adder 406 calculates the pattern shown in FIG. 3(A), adder 407 calculates pattern (B), adder 408 calculates pattern (C), and adder 409 calculates pattern (D).
) is a circuit that calculates the pattern of

The summation circuit 415 is a circuit that inputs the addition results of the adders 406 to 409 and calculates the sum of image data within a 3×3 window. Since the pixel data of interest in the input data is four times as large, this is a circuit for obtaining triple data to subtract triple data of the pixel of interest from this.

The 173 circuit 419 converts 3×3 9 pixel data into data values for 3 pixels.

The subtraction circuits 410 to 413 calculate the difference between each pattern component and the window average value, and equivalently obtain the result of calculating the line drawing component.

The maximum value 8 circuit 414 is a circuit that obtains the maximum absolute value of each component (A) to (D), and outputs the final line drawing amount.

In the above configuration, the energy of noise and signals that exist as isolated points is reduced to ⅓ in a 3×3 window, and erroneous determinations due to noise are reduced. As the window size increases to 5×5.7×7, it becomes less susceptible to noise.

The K1 multiplier circuit 402 outputs a value obtained by multiplying the average value of the 3×3 window by , as multi-value data.

The K1 multiplier circuit 418 outputs a value obtained by multiplying the data itself at the point of interest by .

The switch 420 is switched by control data from the line drawing detection section (maximum value circuit 414).

If it is a line drawing, the output of the 1x circuit 418 is selected, and if it is an image part, the output of the 1x circuit 402 is selected.

In addition, in the above embodiment, the line drawing detection method was basically explained using a 3×3 pattern, but the line drawing detection may be performed using any pattern using NXM pixels, and the multilevel matrix size can also be changed to the PXQ matrix size. It may be composed of

Further, although the line drawing detection pattern has been described for the case where the background is white as shown in FIG. 2, even if the background is reversed, it can be handled by preparing a similar pattern. Furthermore, multi-level conversion may be performed by adding weights to the multi-level matrix size.

In this embodiment, if 406 to 409 are constituted by AND circuits and 410 to 414 are constituted by OR gates, it is possible to determine whether it is a line drawing by pattern matching. Furthermore, by increasing the size of the matrix used for such discrimination, the accuracy of discrimination can be further improved.

As explained above, according to this embodiment, line drawing parts such as two characters can be detected stably, and there is an effect that good line drawing parts can be reproduced along the original binary image.

In addition, line image misjudgments due to noise are extremely reduced, and image quality deterioration is eliminated.

Furthermore, image areas close to line drawings, that is, image areas with high-frequency acid content, can be multivalued with a small multivalue matrix size, making them suitable for high-resolution recording; conversely, areas detected as image areas have large multivalue If the data is multivalued using a matrix, the multivalued data will be recorded with high gradations.

[Effects of the Invention] As described above, according to the present invention, image data can be successfully converted from n value to m value without being affected by noise.

[Brief explanation of drawings]

FIG. 1 is a circuit diagram showing an image area discrimination circuit according to an embodiment of the present invention. FIG. 2 is a schematic diagram showing a window for detecting a thin line portion in the same embodiment. FIG. 3 is a schematic diagram showing an example of a pattern for detecting a thin line portion in the same embodiment. FIG. 4 is a schematic diagram showing the relationship between the original picture, the window, and the point of interest in the same embodiment. 401... Image memory, 402.418... K1 times circuit, 403-405... 3 line buffer memory, 406
~409... Adder, 414... Maximum value circuit, 415... Total circuit 415°, 417... Triple circuit, 419... 1/3 circuit, 420... Switch. Figure 1

Claims (2)

[Claims] (1) When converting n-value image data to m-value image data (n<m), the original image is divided into predetermined windows, and based on the comparison result between the prepared line drawing detection pattern and the image data values within the predetermined window. An image processing method, comprising: controlling the conversion. (2) The image processing method according to claim (1), characterized in that multi-value conversion is performed by varying a multi-value conversion window size in accordance with the likeness of a line drawing based on the comparison result.