JPH0140545B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for identifying whether a multivalued digitized image is a halftone image or another image (character image, line image, etc.).

When processing or transmitting an image, it is often convenient to identify in advance whether the image should be treated as a halftone image. Particularly when dealing with document images that partially include photographs, it is desirable to distinguish between halftone image areas and other image areas within the image.

Various halftone image identification methods have already been proposed for this purpose. A typical example is the Showa 54
``Parallel field segmentation method for free-format documents'' in the proceedings of the 20th National Conference of the Information Processing Society of Japan (pages 453 to 454) and JP-A-Sho.
This is the "Image area signal generation method" of patent publication No. 55-100549.

However, the former method requires the intervention of human judgment, such as checking the processing results and repeating the processing if dissatisfied, and real-time processing cannot be expected. Another problem is that a buffer memory for at least one screen is required.

Although the latter method is only disclosed for processing binary digital images, it can basically be extended and applied to multi-value digital images as well. However, when applied to multivalued digital images, creating and maintaining mask patterns becomes a significant burden. Furthermore, since this method does not process an image by dividing it into meshes, it is extremely difficult to simultaneously achieve both a reduction in processing time per pixel and an improvement in identification accuracy. In other words, in order to shorten the processing time, it is necessary to reduce the number of scanning lines to be input into the buffer, and in doing so, the mask must also be made smaller, which reduces identification accuracy. On the other hand, in order to improve accuracy, it is necessary to increase the number of scanning lines entered into the buffer and to make the mask larger, but this reduces the processing speed and increases the burden of creating and maintaining the mask pattern.

Therefore, an object of the present invention is to provide a halftone image identification method that enables real-time processing without human intervention, requires only a small buffer memory capacity, and has high identification accuracy. be.

According to the present invention, a multivalued digital image is divided into meshes of several pixels. For example, the first
As shown in the figure, a multivalued digital image is divided into a 4×4 pixel mesh of 4 pixels in the main scanning direction and 4 pixels in the sub-scanning direction. Then, dither processing and simple binarization processing are performed for each mesh. Since dither processing is well known as a process for expressing a halftone image using binary black and white pixels, its details will be omitted. Further, the simple binarization process is also a well-known process in which the density value of a pixel is compared with a specific threshold value to determine whether the pixel is a white pixel or a black pixel, so its details will be omitted.

Now, in each mesh, the number N(D) of pixels determined to be black pixels by dither processing, and the number N(D) of pixels determined to be black pixels by dither processing and white pixels by simple binarization processing. Find D-B). Then, it is determined that N(D-B)/N(D)≧C,
A mesh for which a positive result is obtained is determined to be a halftone image area. The above C is a constant, and when the main and sub-scanning line density of a multilevel digital image is, for example, 8 lines/mm,
C can be determined at around 0.2.

Note that the size of the mesh is preferably made to match the size of a threshold matrix (dither matrix) used for dither processing.

Next, an example of an apparatus for implementing the halftone image identification method of the present invention will be explained with reference to FIG.

The image data read by a scanner (not shown) and converted into multi-value digital data is input to the input line in pixel units.
are input sequentially. This input image data is serially input into a shift register IB1 as an input buffer, and sequentially shifted to the right in synchronization with the pixel transfer clock. The image data serially output from this shift register IB1 is input into the next stage shift register IB2. shift register IB
There are a total of eight shift registers, 1 to IB8, and the output of each stage's shift register is sequentially sent to the shift register below it.
The image data output from the shift register IB8 at the bottom stage is sent to the output line _OUT1 .

Each of the shift registers IB1 to IB8 has a capacity that can accommodate one scanning line of image data, and a total of eight scanning lines of image data are temporarily stored. Therefore, the input line
Image data input to IN is output from output line OUT ₁ after 8 scanning lines.

Note that this apparatus assumes that the input image is divided into 4×4 meshes and processed.

The image data input to the input line IN is sent to the dither processing circuit B and the dither/simple binarization processing circuit D.
is also entered. The dither processing circuit B compares the value of each input pixel (density value) with the value of the corresponding element of the 4×4 size dither matrix, and if the former value is greater than the latter value, “ 1” (black pixel)
otherwise outputs “0” (white pixel). This output is input to counter BC, and the number of "1"s is counted. The counter BC has the same number of circuits as the number of meshes on the four scanning lines, and the number N(D) of pixels determined to be black pixels by dither processing in the mesh corresponding to each circuit is determined.

The dither/simple binarization processing circuit D performs dither processing and simple binarization processing for each input pixel,
When a pixel is determined to be a black pixel by dither processing and a white pixel by simple binarization processing, "1" is output corresponding to that pixel. Specifically, this circuit D provides an input pixel to the same circuit as the dither processing circuit B and a comparison circuit for simple binarization processing, and generates a signal obtained by negating the output of the former circuit and the output of the latter circuit. It can be configured as hardware to calculate the logical product of , or it can be configured as firmware that implements the above logic.

Note that it is also possible to configure the dither processing circuit in the dither/simple binarization circuit D to be shared with the dither processing circuit B.

The output of the dither/simple binarization processing circuit D is input to a counter clutch DC, and the number of "1"s is counted. This counter touch DC also has the same number of circuits as the number of meshes on the 4 scanning lines.
Each circuit has a circuit D in its corresponding mesh.
The number N(D-B) of "1"s output from is determined.

The judgment circuit P calculates the ratio N(D-
B)/N(D) and the constant C are compared, and N
If (D-B)/N(D)≧C, the mesh is determined to be a halftone image area and "1" is output, otherwise "0" is output. This determination circuit P can be easily constructed using combinational logic, a fixed memory (ROM), or a semi-permanent memory (PROM).

The output of the determination circuit P is held in the output buffer OB corresponding to the mesh. This output buffer OB
has a capacity capable of holding judgment outputs for four scanning lines.

The contents of the output buffer OB, that is, the halftone image identification results of each mesh, are output to the output line when the image data in the corresponding mesh is sent to the output line OUT ₁ .
Sent to OUT ₂ .

Note that the above-mentioned halftone image identification process is executed in parallel with the image data sending operation. In other words, image data for four scan lines is output as a pixel by pixel.
When transmission starts from OUT ₁ , the identification processing for all meshes on the four scanning lines has been completed and the results have been stored in the output buffer OB. Then, in parallel with outputting the image data for the four scan lines and the determination result, halftone image identification processing is performed on the image data for the next four scan lines. Therefore, if the purpose is only to perform halftone image identification processing, the capacity of the input buffer for 4 scanning lines is sufficient (4 x 4
size).

As described above, the present invention is capable of automatically performing halftone image identification on a multivalued digital image in real-time processing without requiring human intervention, and requires only a small buffer memory. It is possible to solve all the problems of the prior art, such as not being able to use this method, and its effects are remarkable.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an example of a pixel arrangement and mesh division of a multivalued digital image, and FIG. 2 is a block diagram showing an example of an apparatus for carrying out the present invention. IN...Input line for image data, OUT ₁ ...Output line for image data, OUT ₂ ...Output line for halftone image identification results, IB1 to IB8...Shift register (input buffer), B...Dither processing circuit , D... dither/simple binarization processing circuit, BC, DC... counter, P... judgment circuit, OB... output buffer.

Claims (1)

[Claims] 1 Divide the multivalued digital image into meshes containing multiple pixels, perform dither processing and simple binarization processing on the pixels of each mesh, and calculate the number N(D) of pixels determined to be black pixels by dither processing, The number N(D-B) of pixels that are determined to be black pixels in dither processing and white pixels in simple binarization processing is determined for each mesh, and this number of pixels N(D-B) and N( A method for identifying a halftone image, the method comprising: identifying whether each mesh is in a halftone image area or not based on the ratio of the mesh to D).