JPH0352383A - Picture binarizing system - Google Patents

Abstract

PURPOSE:To improve the picture quality of the entire binarizing picture by discriminating the degree of a change in the gradation from the size of gradation change to select the threshold level pattern used for organizing dither processing thereby balancing the gradation and the resolution in response to the level of the gradation change. CONSTITUTION:The system is provided with a gradation change extraction means 4 calculating a change in gradation of a multi-gradation picture, a gradation change discrimination means 5 discriminating the degree of a change in the gradation depending on the level of gradation change calculated by the gradation change extraction means 4 and a dither pattern switching means 6 switching the threshold level pattern adaptively used for the organic dither processing based on the result of discrimination by the gradation change discrimination means 5. Then the dither threshold level matrix emphasizing the resolution and the dither threshold level matrix emphasizing the gradation are used depending on the situation. Thus, a binary picture satisfying both the resolution and gradation is obtained and the picture quality as the entire binary picture is improved.

Description

[Detailed Description of the Invention] [Summary] An image binarization method that converts a multi-level image into a binary image with good resolution and gradation in order to output a multi-level image to a binary image output device. The purpose of this technology is to improve the image quality of the entire binarized image by making it possible to change the balance between gradation and resolution according to the magnitude of the gradation change. In an image processing system that binarizes a multi-tone image using a systematic dither method in order to output the image to a binary image output device, tone change degree extraction calculates the degree of change in tone of a multi-tone image. means and
A gradation change degree determining means for determining the degree of gradation change from the value of the gradation change degree calculated by the first gradation change degree extraction means; and dither pattern switching means for adaptively switching the threshold pattern used for digital dithering. [Industrial Application Field] The present invention relates to an image binarization method for converting a multi-level image into a binary image with good resolution and gradation in order to output a multi-level image to a binary image output device. Regarding.

[Conventional technology]

Among the devices that output images, there are many that can only display binary values of light and dark. These include dot printers that print black dots of a fixed size, and liquid crystal displays and plasma displays that have display cells of a fixed size arranged in a matrix. In order to output multivalued images with halftones such as photographs from these output devices, image binarization processing is required to convert multivalued image data into binary data.

Conventionally, a systematic dither method has been frequently used as an image binarization method for converting a multi-gradation image into a binary image.

Figure 6 shows an example of the configuration of a system that implements the systematic dithering method. Centering on CPU61, which is an example of a microprocessor,
The external bus is connected to a ROM 62, a RAM 63, an external storage device 64 such as a disk device, and an image output device 65 such as a thermal printer that outputs image data converted into a binary image.

First, a multivalued original image is stored in the external storage device 64,
The original image data is processed by the CPU 61 and converted into a binary image. Here, the ROM 6 2 includes a BIO S (basic input out) for executing the CPU 6 1.
tputsystem), etc., and the RAM 63 stores an operating system, a systematic dither method processing program, processing data, and the like. Then, the converted binary image is output to the image output device 65.

The image binarization process is executed according to the operation flowchart shown in FIG.

The CPU 61 first reads multivalued image data from the external storage device 64 and stores it in the RAM 62 (S71). Next, the multivalued images stored in the RAM 62 are sequentially dithered (S72). Systematic dithering will be described later. Finally, the image binarized by dithering is output to the image output device 65 (S73).

The dither method compares the gradation of the original image with a threshold value for each pixel, and if the gradation is larger than the threshold value, it is turned on (
For example, the value is "1", and if it is smaller, it is turned off (for example, "0"). At this time, the threshold value is set to take a random value for each pixel. This randomization method visually improves image quality by removing the spatial correlation of pseudo-contour noise that occurs when images are quantized using low bits. There are several methods for dithering, depending on the rules used to randomize the threshold values, but systematic dithering is currently widely used as it has relatively good image quality and simple processing. The threshold value of the systematic dither method is an nXn matrix, and this threshold matrix (dither matrix) is compared with the nXn pixels of the original image in order from the corner of the original image, and if each pixel is larger than the threshold value, then It is binarized so that it is on, and if it is small, it is off. Each element of the nXn dither matrix is an integer from O to n2-1, and each element is set to be a different integer. This makes it possible to express n2+1 gradations. In other words, a 2×2 dither matrix has 4 thresholds and can express 5 gradations, and
A 4×4 dither matrix has 16 threshold values and can express 17 gradations.

Various methods of dithering the dither matrix have been studied. n
Since the Xn dither matrix is repeatedly compared with the n×n pixels of the original image, a background texture is inevitably generated due to the periodicity of this repetition. A dither matrix that makes this background pattern as inconspicuous as possible is optimal.

FIG. 8 is an explanatory diagram of a commonly used dither threshold matrix.

FIG. 5A shows a Bayer type threshold matrix.

This is a threshold matrix in which the lowest frequency in the spatial frequency spectrum of the background pattern is set at a high level so as not to make the background pattern stand out. Usually 4
Either a x4 ((a-1) in the same figure) or a 2×2 ((a-2) in the same figure) threshold matrix is used fixedly.

On the other hand, the figure (b) shows a spiral threshold matrix.
The threshold value in the center of the matrix is set to take a small value, and the value increases toward the outside in a spiral pattern. The figure (b-1) is 4x4, the figure (b-2) is 2
This is a ×2 spiral matrix. In the case of this threshold matrix, the central part of the matrix is binarized relatively black even if the gradation of the original image is low.

In the conventional method, one of the threshold matrices, for example the matrix in FIG. 8, is selected and fixedly applied to the original image for binarization. That is, if a 4×4 Bayer type threshold matrix is selected, that threshold matrix is fixedly applied to the entire one multivalued original image.

Since the number of gradations that can be expressed is determined by the order n of the threshold pattern, n2 1, generally speaking, the larger the threshold matrix, the more differences in gradation can be expressed. However, the resolution decreases because the repeating pattern of the threshold increases. In the case of a small threshold matrix such as 2×2, the gradation expression ability is low, but the repeating pattern is small, so the resolution is good. When selecting a threshold matrix, it was necessary to take such characteristics into consideration and select it according to the application. In other words, a large threshold matrix is used for natural images with relatively small gradation changes, and a small threshold matrix is used for line drawings and character images. Characters and lines can basically be expressed as binary images, so there is no need to increase the number of gradation levels, and a smaller number of gradation levels allows pixels of the same gradation to be encoded into different binary expressions. This is because the missing dots caused by this can be suppressed.

On the other hand, in order to suppress the decrease in resolution as much as possible, a method is also used in which processing such as contour enhancement is performed on the original image before applying the systematic dithering method.

[Problem to be solved by the invention]

However, the conventional method still has the problem that the image quality of the binarized image is poor in terms of resolution and gradation depending on the type of the original image. The problem here is particularly in images that have a mixture of areas with gradual gradation changes, such as natural images and facial images, and areas with rapid gradation changes, such as characters and lines. This is dithering. Generally, images with gradual gradation changes, such as natural images or facial images, look more natural visually when expressed with a large number of gradations. That is, it is better to set the value of n in the threshold matrix to a certain degree.

However, if there are characters or lines in part of the image,
Even if you perform processing such as edge enhancement and then binarize using systematic dithering, parts of the characters and lines will still be missing. This occurs because the size of the threshold matrix is too large to represent fine details such as characters. Therefore, suppose that the size of the threshold matrix is reduced to, for example, 2×2. This will eliminate missing characters and lines, but the naturalness and quality of the image as a whole will deteriorate.

The present invention aims to improve the image quality of the entire binarized image by making it possible to change the balance between gradation and resolution depending on the magnitude of gradation change. [Means for Solving the Problems] FIG. 1 is a functional block diagram of the present invention. The present invention
A multivalued image inputted by a multivalued image input device 1 is binarized by an image processing system 2 using a method based on a systematic dither method,
It is assumed that the image is output to the binary image output device 3.

The gradation change degree extraction means 4 extracts the multi-valued image inputted by the multi-valued image input device 1 as a partial image for each appropriate number of pixels (for example, 4×4 pixels), and extracts the gradation change within that range as a partial image. Calculate the degree of gradation change representing magnitude.

The gradation change degree determining means 5 compares, for each partial image, a predetermined threshold value regarding the gradation change degree with the gradation change degree extracted by the previous gradation change degree extraction means 4, and Determine whether the tone change in a partial image is large or small. In the dither pattern switching step 6, if it is determined that the Ill tone change is large based on the result of the former tone change degree determining means 5, the dither pattern switching step 6 performs dithering that prioritizes resolution over tone quality for that partial image when it is determined that the Ill tone change is large based on the result of the former tone change degree determining means 5. Applying a threshold pattern, while
If it is determined that the gradation change is small, image binarization is performed using a systematic dither method using a dither threshold pattern that prioritizes gradation over resolution, and the binary image is converted into a binary image. Output to the image output device 3.

[For production]

First, from the multi-value image input device 1, a multi-value image, for example 25 pixels per pixel
The original image with six gradations is read into the image processing system 2. The read original image data is input to the gradation change degree extraction stage 4 for each partial image of an appropriate number of pixels (for example, 4×4 pixels).

The gradation change degree extraction means 4 calculates the gradation changes of adjacent pixels each time the gradation data of a partial image is input, and accumulates these gradation changes to determine the gradation level within the partial image. Calculate the degree of change. The calculated degree of tone change of the partial image is sent to the degree of tone change determination means 5.

Upon receiving the input of the degree of gradation change of the partial image, the 1000-step gradation change degree judgment 5 compares the degree of gradation change with a preset threshold value for the degree of gradation change, and compares the result. It is sent to the dither pattern switching means 6.

The dither pattern switching means 6 receives the determination result from the gradation change degree determination means 5, selects an appropriate dither threshold pattern according to the determination result, and executes image binarization processing. In other words, if the degree of gradation change of the partial image is greater than the threshold, it is determined that the partial image contains elements that require high resolution, such as characters or lines, and the gradation is A low-order dither threshold matrix with emphasis on resolution is applied to the partial image, and the partial image is binarized. On the other hand, if the degree of gradation change of the partial image is smaller than the threshold value, a dither threshold matrix with a large order that emphasizes gradation rather than resolution is applied to binarize the partial image. . The binarized image is output to the binary image output device 3.

〔Example〕

Examples will be described below with reference to FIGS. 2 to 5.

FIG. 2 is a system configuration diagram of an embodiment of the present invention. This embodiment can be implemented using a computer system such as a personal computer or a workstation. In other words, the system is built around the CPU 2 1, which is an example of a microprocessor, and the ROM 22 and RAM are connected to the external bus of the CPU 2 1.
This is a computer system in which 23 memory groups and a hard disk 24 are connected. Also, an image digitizer 2 for importing multivalued images into the computer system.
5, and a binary thermal printer 26 for outputting a binary image.
Connect to the external bus of CPU21. A TV camera 27 for capturing images is connected to the image digitizer 25.

The ROM22 contains the CPU2, such as the BIOS.
The RAM 23 stores basic and fixed system programs and data that are essential for executing the processing of step 1. In addition to the OS, the RAM 23 also stores processing programs and data for the image binarization method of the present invention. The hard disk 24 stores image data collected as multivalued digital data by the image digitizer 25, image data after binarization, and the like.

Next, the operation of this system will be explained.

The user uses the TV camera 27 to take an original image such as a photograph that he or she wants to binarize. The NTSC signal of the photographed original image is input to the image digitizer 25 and encoded into multi-valued image data that can be processed by a computer system.

That is, the image digitizer 25 converts the NTSC signal into
For example, it is encoded using the NETEC-22H method or the like and converted into multivalued image data. With this encoding, for example, l pixels 8
Bit (256 gradation), 512×512 i# element/original image multivalued image data is obtained. The encoded multivalued image data is then temporarily stored in the hard disk 24.

When the multivalued image data is stored in the hard disk 24,
The CPU 2 1 sequentially binarizes the multivalued image data read from the hard disk 24 in accordance with the processing programs stored in the RAM 23 and ROM 22 using a method based on the systematic dither method, and stores the binarized image data in the RAM.
23 or hard disk 24. Finally, the CPU
2 1 reads binary image data from the RAM 23 or hard disk 24 and outputs it to the binary thermal printer 26 . Through the above operations, it becomes possible to output the halftone image captured by the TV camera 27 to the binary thermal printer 26, which is a binary image output device.

Next, the operation of the image binarization processing program in one embodiment will be explained along the operation flowchart of FIG.

First, the original image data encoded into multi-values, for example l screen 5
Data of 12 x 512 pixels, 8 bits per pixel is read from the image digitizer 25 to the hard disk 24 (Sl
-1), and then read from the hard disk 24 into the RAM 23 for further binarization processing (St-2).

Partial images of an appropriate size, for example, 4×4 pixels, are sequentially extracted from the multivalued image data read into the RAM 23, and a scale V indicating the degree of gradation change in the partial images is calculated (S2). In this embodiment, the degree of change in tone V is the sum of differences in tone between adjacent pixels in a partial image. 4×4
A method for calculating the degree of gradation change V in a partial region of a pixel will be explained with reference to FIG.

The gradation of a total of 16 pixels (4×4) is expressed as g▲J (1%J=1,
2, 3, and 4), the degree of gradation change V in this partial image is as follows. The first term of this equation accumulates the tone differences between horizontally adjacent pixels of the 4×4 pixels, and the second term accumulates the tone differences between vertically adjacent pixels. Using this formula, the gradation differences of all adjacent pixels included in the 4×4 pixel partial image are summed, and the gradation change degree V is determined. When the partial image is, for example, a part of a face image, the difference in gradation between adjacent pixels is small, and ■ is considered to take a small value. On the other hand, if the partial image includes a part of a line or character, the gradation of each pixel in the partial image will differ greatly, and the value of ■ will become large.

Once the gradation change degree V of the partial image is calculated, the value of ■ is compared with the threshold value t (S3). This threshold value t is
A value is set in advance as a reference value for determining whether the degree of gradation change V is large, that is, whether the gradation changes significantly in a partial image.

According to the result of this determination (S3), dithering using different dither threshold patterns (dithering 1, dithering 2)
) is executed (S4 and S5).

That is, if the degree of gradation change V in the partial image is larger than the threshold value t in S3 (YES), the threshold pattern of FIG. If the degree of change V is smaller than the threshold t (NO), dithering is performed using the threshold pattern (6) in the same figure (S5
).

The threshold pattern in FIG. 5(a) is a 2×2 Baye
In order to apply the r-type threshold pattern 51 to a partial image of 4 x 4 pixels, four pieces are arranged to form 4 x 4 pixels, and it is applied when the degree of gradation change V is large (S4). According to this threshold pattern, it is possible to perform dithering that takes advantage of the high resolution characteristic of the 2×2 threshold pattern.
This prevents deterioration in image quality due to the loss of fine image elements such as lines.

On the other hand, the threshold pattern in Figure 5(b) is a normal 4x4
A Bayer type threshold matrix is applied to a partial image in which the degree of gradation change ■ is not large (S5).

Although the resolution is lower than the 2×2 threshold pattern, it is a threshold pattern with excellent gradation characteristics, and is applied to image parts where gradation changes are relatively small, such as facial images and natural images. As a result, a binary image with visually high quality can be obtained.

When the dithering in S4 and S5 is completed and a binary image is obtained, the binary image is output from the binary thermal printer 26 (S
6), end the process.

〔Effect of the invention〕

According to the degree of gradation change within the original copy, the present invention
By selectively applying a dither threshold matrix that emphasizes resolution and a dither threshold matrix that emphasizes gradation, it is possible to obtain a binary image that satisfies both resolution and gradation. Improve the overall image quality.

[Brief explanation of drawings]

Fig. 1 is a functional block diagram of the present invention, Fig. 2 is a system configuration diagram of one embodiment, Fig. 3 is an operation flowchart of one embodiment, Fig. 4 is an explanatory diagram of the scale of the degree of gradation change, Figure 5 is a diagram showing the Bayer type threshold matrix used in one embodiment, Figure 6 is a system configuration diagram (conventional system), Figure 7 is an operation flowchart of the conventional system, and Figure 8 is a diagram of the threshold matrix. It is an explanatory diagram. 1... Multivalued image input device, 2... Image processing threshold value, 3... Binary image output device, 4... Gradation change degree extraction means, 5... Gradation change degree determination means,

Claims (1)

[Claims] 1) In order to output the multi-gradation image input from the multi-level image input device (1) to the binary image output device (3), the multi-gradation image is converted into a binary image by a systematic dither method. In the image processing system (2), the gradation change degree extraction means (4) calculates the degree of gradation change of a multi-tone image, and the gradation change degree calculated by the former gradation change degree extraction means (4). gradation change degree determining means (5) for determining the degree of gradation change from the magnitude of the degree value;
) A dither pattern switching means (6) for adaptively switching a threshold pattern used for systematic dithering based on the determination result of (6). 2) The early dither pattern switching means (6) selects a threshold pattern having a characteristic of emphasizing gradation when the early gradation change determining means (5) determines that the gradation change is large; 2. The image binarization method according to claim 1, further comprising selecting a threshold pattern having a characteristic that emphasizes resolution rather than gradation when it is determined that the change in gradation is small.