JPH0546749B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to an image processing device having an image identification function.

[Prior art]

Conventionally, in this type of digital image processing device, the identification accuracy of the read image was insufficient, and a series of image processing was performed on the entire surface of the document image even if the identification was incorrect. It has been difficult to faithfully reproduce all partial image areas for a document containing a mixture of the three types. In particular, it is difficult to distinguish between the halftone image area and the character area.
The disadvantage is that the halftone dot image area undergoes processing for the text area, resulting in significant degradation of the reproduced halftone image.

〔the purpose〕

The present invention has been made in view of the above-mentioned problems, and it is possible to solve the problems described above with respect to an image represented by binary image data for each pixel. Image tone identification that enables accurate identification with a simple configuration when extracting parameters according to two-dimensional spatial frequency characteristics and identifying the image tone of an image within the two-dimensional block based on the extracted parameters. The purpose is to provide a method.

〔Example〕

The present invention will be explained below based on the drawings.

First, the image tone identification algorithm according to the present invention will be described in order of each step.

(Step 1) The original image is read with a scanner consisting of, for example, a CCD, and the read image is digitized to obtain a digital image. This digital image is divided into blocks of N×M pixels (here, N is the number of pixels in the main scanning direction, and M is the number of pixels in the sub-scanning direction). Here, when the scanner scans at 16 pixels/mm, the best value for each of N and M is 8. That is, here, the image is divided into blocks of 8×8 pixels.

(Step 2) The pixel density in each divided block is
Binarize with a constant threshold to create a binary block pattern. On this binary block pattern, the total sum S of the number of inversions of adjacent pixels (binary data) is determined.

For example, in the N×M=8×8 image block shown in FIG. 1, each pixel density value ranges from 0 (white) to 15
(black), the binary block pattern obtained by binarizing at level 7 (fixed threshold 7) is as shown in FIG. Here, when calculating the number of times of inversion (for example, the number of times the pattern changes from 0 to 1 or 1 to 0) in the first main scanning line l1 (top column), 2 is obtained.

When the sum S of the number of inversions on all sub-scanning lines and main scanning lines is calculated in the above manner, S=27 is obtained in the figure. Note that the possible values of S are in the range of 0 to M(N-1)+N(M-1).

(Step 3) Find the average value of pixel density within each divided block. In the example in Figure 1, the average concentration value≒5
becomes.

(Step 4) The preset values P1 and P2 are compared with the sum S obtained within each block, and images are identified for each block based on the following criteria.

(1) S≦P1... Photographic (halftone) image area (2) P1<S≦P2...Character (line drawing) image area (3) S＞P2 ...halftone image area Here, it is assumed that P1<P2.

The above identification standard is based on the statistical property that the two-dimensional spatial frequency of a document decreases in the order of halftone dots > text > photographs, and P1 and P2 are respectively P1 and P2 in this example. =8~10,P2=
A value of about 20 to 24 is practical. Note that in this example, one fixed threshold was used to obtain a binary block pattern from each block, and the total number of reversals S was obtained. However, for example, two types of binary block patterns may be obtained using two fixed thresholds. Alternatively, the sum of the number of inversions may be determined, and the image area may be identified based on the results of these calculations. In any case, it is sufficient to obtain a value that allows two-dimensional determination of the ups and downs of image density.

(Step 5) Each block is converted into a binary image signal based on the result identified in (Step 4).

(5-1) If it is determined to be the photographic image area, each pixel within the block is dithered. Here, dither processing is, for example, by comparing each threshold value of an 8 x 8 threshold matrix with the density data of each pixel in a block, to generate a binary signal "1",
Say what you get “0”.

(5-2) If it is determined that it is in the character area, a binary block pattern as shown in FIG. 2 is output as is.

(5-3) If it is determined to be the halftone image area, the density of each pixel in the block is replaced with the average density ("5" in this example) obtained in (step 3), and dither processing is performed. . That is, the processing method executed when it is determined that the area is a halftone image area is the so-called density pattern method, and the dither matrix used here is preferably a dot concentration type. This is because by calculating the density average value for each block and applying a dot-concentrated dither to this average value, it is possible to create halftone image originals with high spatial frequencies and binarization processing (individuals possessed by the dither matrix). interference (generally occurs as moiré fringes)
Noise can be suppressed.

Using the algorithm described above, the processing in block units (steps 1 to 5) is repeated sequentially, and a single original document is binarized in near real time, using a binary printer such as a laser beam printer (LBP). Image reconstruction can be performed.

Next, an example of an image processing device according to the present invention will be described. The image processing apparatus in this embodiment performs the above-described image processing by computer processing, particularly by software processing by a microcomputer dedicated to image processing having a multi-microprocessor structure. An example of the configuration in this case is shown in the block diagram of FIG. The original image is read by the scanner SC, and the read image signal is A/D converted by an A/D converter (not shown) and temporarily stored in the image memory GM. The image data in the image memory GM is then software-processed by the image processing dedicated microprocessor GP according to the image processing algorithm described above and converted into a binary signal, which is then output to the laser beam printer PR. be done. SP
is a microprocessor for controlling the entire system, SM is a program memory that stores its control contents, and GM is an image memory used in the image processing process.

Figure 4 shows the microprocessor in Figure 3.
This is a flowchart showing the steps of image processing performed by GP. To explain the diagram, in (step 1 to step 2) pixel data is extracted from the image memory GM and 8×
A block of 8 pixels is formed and the process proceeds to step 3. In step 3, each pixel data in the block is binarized using a fixed threshold as described above, and the total sum S of the number of inversions is determined. (Step 4,
In step 5), images are identified based on the determined sum S. “NO” in (Step 4),
When it is determined “NO” in (step 5), the block is determined to be a halftone image area,
In step 6, the average density value of the block is determined and each pixel data within the block is replaced with the average density value. Then, in step 9, dither processing is performed on the block whose density has been smoothed. When it is determined as "NO" in (Step 4) and "YES" in (Step 5), it is determined that the block is in the character image area, and the process moves to (Step 7). (Step 7), the binary data obtained in Step 3 is extracted as is. If it is determined as “YES” in (Step 4), it is determined that the block is a photographic image area,
In step 8, dither processing is performed on the block. (Step 10) Then (Steps 7-9)
The binary data obtained for each block is stored in the image memory.
Store in GM.

In this example, the image memory GM is assumed to be capable of storing at least 16 lines of multilevel density level pixel data and 16 lines of binary data, as shown in FIG. Note that one line here refers to, for example, one main scanning line of a document read by a scanner. The memory areas a ₁ and a ₂ are used to store pixel data sequentially read by the scanner SC and to take out the data in units of 8×8 pixels. That is, for example, pixel data from the scanner SC is sequentially written in the line direction in memory area _a1 , and pixel data is extracted in blocks of 8×8 pixels from memory area _a2 by the microprocessor GP. It is. For example, memory area _b1 has an 8x8
Binary data is written for each block of pixels, and the already written pixel data (binary data) is sequentially read out in the line direction from the memory area _b2 . This read binary data is then output as a reproduced image by the printer PR. In this way, in this example, by performing parallel processing of pixel data using multiple microcomputers, it is possible to simultaneously write multi-level density level pixel data and read out block-by-block multi-level density level pixel data. Since writing of each block of data and reading of binary data can be performed simultaneously, and the parameter S can be calculated two-dimensionally at high speed, the read pixel data can be processed and output in almost real time. In addition, Sukiyana
Control of SC and PR is microprocessor
This is performed by the SP in synchronization with the operation of the microprocessor GP.

[Other Examples]

As a modification of the above embodiment, in particular, when creating a binarized block pattern in (step 2),
The value conversion method will be described. In the above-mentioned embodiment, complete binarization was performed using a constant threshold value 7, but in order to improve the identification accuracy of photographic image areas, changes in density were observed in the main scanning direction (or sub-scanning direction). If the density difference between adjacent pixels exceeds a predetermined value ΔL, a method of inverting and binarizing (for example, converting binary data from "0" to "1") may be used. In other words, the density change in the photographic image area is 5→7→
When 6→6→8, if ΔL=7, the binarization pattern becomes 0-0-0-0-0 and the number of inversions becomes 0. Normally, strong density changes are hardly observed in photographic image areas, so by using this method, more accurate identification of photographic image areas can be performed.

Alternatively, it is also effective to use a method of two-dimensionally determining the difference between the density of the pixel of interest and the average value of the density of eight adjacent pixels, and performing inversion and binarization if this difference is greater than or equal to ΔL.

In addition, in this example, three types of binarization processing were applied to the image data by switching according to the identification results, but the same image data was processed in parallel by the three types of processing methods described above. However, by configuring the system to select one of the three types of binarized image data based on the identification result, faster processing can be performed.

〔effect〕

As described above, according to the present invention, for an image represented by binary image data for each pixel, the image data can be adjusted according to the two-dimensional spatial frequency characteristics within a two-dimensional block of a predetermined size made up of a plurality of pixels. Extract the parameters and perform the above 2 based on the extracted parameters.
When identifying the tone of an image within a dimensional block, 2
Parameters corresponding to the two-dimensional spatial frequency characteristics are extracted from the number of times the value of the value image data is inverted, so it is possible to identify the image tone with a simple configuration, and it is also possible to identify multiple directions that are different from each other within a two-dimensional block. Since the number of inversions for a plurality of lines is calculated for each line, appropriate parameters can be extracted according to the two-dimensional spatial frequency characteristics, and accurate image identification is possible. In particular, it is possible to accurately identify, for example, a halftone image and a character (line) image.

[Brief explanation of the drawing]

FIG. 1 is a density distribution diagram when a document according to the present invention is divided into 8×8 image blocks, FIG. 2 is a diagram showing an example of the binarized block pattern of FIG. 1,
FIG. 3 is a diagram showing an example of the configuration of an image processing device according to the present invention, FIG. 4 is a flowchart showing the procedure of image processing executed by the microprocessor GP, and FIG. 5 is a diagram showing the image memory GM. It is. N×M...Number of pixel divisions in the main scanning and sub-scanning directions of the original image, S...Total number of inversions, SC...Scanner,
GM...image memory, GP, SP...microprocessor, PR...LBP printer, SM...program memory.

Claims (1)

[Claims] For an image represented by binary image data for each pixel, parameters are extracted according to two-dimensional spatial frequency characteristics within a two-dimensional block of a predetermined size made up of a plurality of pixels. , an image tone identification method for identifying the image tone of an image within the two-dimensional block based on the extracted parameters, wherein the values of binary image data of adjacent pixels within the two-dimensional block are inverted. number of times first
calculating the number of times the values of binary image data of adjacent pixels are inverted for multiple lines in a second direction different from the first direction in the two-dimensional block; extracting a parameter corresponding to a two-dimensional spatial frequency characteristic from the number of inversions found in the first and second directions, and representing the binary image data in the two-dimensional block based on the extracted parameter; An image tone identification method characterized by identifying the image tone of an image.