JP3584478B2 - Binarization processing apparatus and binarization processing method - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to a binarization processing device and a binarization processing method for binarizing input image data.
[0002]
[Prior art]
For example, in thinning processing of an image signal, a method of performing thinning processing subsequent to binarization processing is general. At this time, in the binarization processing, a pixel block adjacent to a certain pixel (hereinafter, referred to as All or most of the neighboring pixels should be binarized to one value, but some pixels are binarized to the other value due to the influence of noise and the like. There are cases where it is done.
[0003]
As described above, an arbitrary pixel which is binarized to the other value out of all or most of the continuous pixels of the pixel block which should be binarized to one value is an isolated point or a quasi-isolated point. is called.
[0004]
The isolated point or the quasi-isolated point has an adverse effect at the stage of the thinning processing performed after the binarization processing as described above. Therefore, a process of removing the isolated point (hereinafter, referred to as an isolated point removal process) has been considered.
[0005]
[Problems to be solved by the invention]
By the way, as a method conventionally considered as the above-mentioned isolated point removal processing, for example, a pixel which is in contact with a pixel of one value by using one of the binarized values as the other binarized value in the vicinity After that, there is a process of conversely reducing the other value to one value, or conversely, a process of reducing and enlarging.
[0006]
In such a conventional isolation removal process, the binarized image must be inspected again, and an isolated point removal operation must be newly performed, which is troublesome and troublesome.
[0007]
Therefore, the present invention has been made in view of the above-described circumstances, and an object of the present invention is to provide a binarization processing device and a binarization processing method that eliminate the need for isolated point removal processing after binarization processing.
[0008]
[Means for Solving the Problems]
A binarization processing apparatus according to the present invention is a binarization processing apparatus for performing binarization processing on input image data, wherein the input image data is divided into blocks and divided into classes, and class information corresponding to the class division is output. A class dividing unit, a storing unit in which a class-specific pattern obtained by performing learning of an isolated point removing process in advance is stored, and a class-specific pattern corresponding to the class information is read out from the storing unit so that the input is performed. A conversion unit for converting an image, wherein the learning is to obtain the class-specific pattern corresponding to the class information from a multi-valued image for learning and a binary image for learning. Solve.
[0009]
In this case, the class division unit includes a blocking processing unit that blocks input image data, a threshold value determination processing unit that determines a threshold value from the input image data, and a block processing unit that blocks the input image data. A binarization processing unit for performing a binarization process on the input image data using the threshold value determined by the threshold value determination processing unit, and image data binarized by the binarization processing unit And a conversion processing unit that divides the class information into corresponding class information.
[0010]
Further, in the binarization processing method according to the present invention, in the binarization processing method for binarizing input image data, the input image data is divided into blocks and divided into classes, and class information corresponding to the class division is obtained. The input image is read by reading a class-specific pattern corresponding to the class information from a storage unit in which a class-dividing step to be output and a class-specific pattern obtained by performing learning of an isolated point removal process in advance are stored. A conversion step of performing conversion, wherein the learning solves the problem by obtaining the class-specific pattern corresponding to the class information from a multi-valued image for learning and a binary image for learning. I do.
[0011]
In this case, the class division step includes a block processing step of dividing input image data into blocks, an image storage step of storing the input image data for one screen, and an input image of one screen stored in the image storage step. a threshold value determination processing step of determining the data or Rashi threshold, blocked input image data in using a threshold determined by the threshold value determination process step 2 values from the blocking process steps It is preferable to include a binarization processing step of performing a binarization processing, and a conversion processing step of dividing the image data binarized by the binarization processing step into classes and converting the image data into corresponding class information.
[0012]
[Action]
According to the present invention, a class-specific pattern corresponding to class information is read from a storage unit in which a class-specific pattern obtained by previously performing learning of an isolated point removal process is stored, and an input image is converted into a class-specific pattern. . In the learning of the isolated point removal processing, a class-specific pattern corresponding to class information is obtained from a learning multi-value image and a learning binary image.
[0013]
【Example】
Hereinafter, embodiments of a binarization processing device and a binarization processing method according to the present invention will be described with reference to the drawings.
[0014]
In this embodiment, as shown in FIG. 1, the input image input from the input terminal 1 is subjected to a binarization process before the input image is thinned by the thinning processing device 6. It is a value processing device 2. Here, input image data input from the input terminal 1 is converted into a one-dimensional digital signal by, for example, scan conversion.
The one-dimensional digital signal input from the input terminal 1 is supplied to the class division processing circuit 3. The class division processing circuit 3 converts the one-dimensional digital signal into blocks, classifies them, converts them into class information, and supplies them to the class-specific pattern conversion processing circuit 4. In the class-specific pattern conversion processing circuit 4, a representative binary signal that has been subjected to the learning of the isolated point removal processing in advance by the conversion pattern determination processing circuit 5 is stored as a conversion pattern so as to correspond to the class information. I have. The class-specific pattern conversion processing circuit 4 reads a representative binary signal corresponding to the class information from the class-specific conversion pattern. The binarized signal output from the class-specific pattern conversion processing circuit 4 is supplied to a thinning processing device 6.
[0016]
Here, the class division processing circuit 3 will be described with reference to FIG.
As shown in FIG. 2, the class division processing circuit 3 includes a blocking processing unit 11, an image storage unit 14, a threshold value determination processing unit 15, a binarization processing unit 12, and a packing processing unit 13. It is composed of
[0017]
The one-dimensional digital signal input from the input terminal 1 is supplied to the block processing unit 11 and also to the screen storage unit 14.
[0018]
The blocking processing unit 11 blocks the one-dimensional digital signal. At this time, blocks near the block composed of the two-dimensional digital signal are converted into a one-dimensional signal sequence.
[0019]
The image storage unit 14 stores one-dimensional digital signals for one screen. The image data for one screen stored in the image storage unit 14 is supplied to the threshold value determination processing unit 15. The threshold value determination processing unit 15 calculates, for example, an average value of image data for one screen, and performs a determination process using this value as a threshold value.
[0020]
The threshold value determined by the threshold value determination processing unit 15 is supplied to the binarization processing unit 12, and the binary processing unit 12 converts the threshold value into binary image data from the block processing unit 11. It is used when performing the chemical treatment.
[0021]
The binarization processing performed by the binarization processing unit 12 on the block image data from the block processing unit 11 will be described.
For example, if the blocking unit 11 blocks image data of m bits per word for every n words, there are 2 ^mn classes for each block. On the other hand, since the binarization processing unit 12 performs binarization on the image data that has been divided into blocks, only ²ⁿ classes are required for classification.
[0022]
The image data binarized by the binarization processing unit 12 is supplied to the packing processing unit 13. The packing processing unit 13 classifies the image data binarized by the binarization processing unit 12 and converts the image data into corresponding class information. That is, the binarization processing unit 12 converts the image data in a binary pattern of "1/0" for each block into class information such as an index and a class code corresponding to each block. This class information is supplied from the output terminal 16 to the class-specific pattern conversion processing circuit 4 shown in FIG.
[0023]
The class-specific pattern conversion processing circuit 4 has a ROM table 22 as shown in FIG. 3 for storing the conversion patterns learned by the conversion pattern determination processing circuit 5.
[0024]
Here, learning of the conversion pattern in the conversion pattern determination processing circuit 5 will be described.
The conversion pattern determination processing circuit 5 includes, in advance, class information serving as an index from a multi-level signal to be input and a binarized signal when the multi-level signal is binarized, and a representative corresponding to the class information. Such a binary signal is obtained by learning. Here, a typical binary signal is a signal from which isolated points have been removed. That is, the conversion pattern determination processing circuit 5 determines, as a conversion pattern, the class information and a representative binary signal from which the corresponding isolated points have been removed, and stores the conversion information in the ROM table 22 of the class-specific pattern conversion processing circuit 4. Let it be.
[0025]
Therefore, in the class-specific pattern conversion processing circuit 4, as shown in FIG. 3, a typical binary signal corresponding to the class information from the division processing circuit 3 supplied via the input terminal 21 is used as a conversion pattern, 22 and output from an output terminal 23. The conversion pattern output from the output terminal 23 is supplied to the thinning processing device 6.
[0026]
Here, the conversion pattern supplied from the output terminal 23 to the thinning processing device 6 is a conversion pattern for each class from which isolated points have been removed in advance, as described above. Therefore, the thinning processing device 6 is supplied with binarization processing data that does not require the isolated point removal processing after the binarization processing, that is, binary image data.
[0027]
The thinning processing device 6 performs thinning processing on the input binary image data, and outputs a line image from the output terminal 7.
[0028]
It should be noted that the binarization processing device and the binarization processing method according to the present invention are not limited to the above embodiment. For example, in the block division unit 3, the binarization processing is performed on the block signal data. However, instead of this binarization processing, ADRC (adaptive dynamic range coding) processing may be performed to reduce the number of class divisions.
[0029]
【The invention's effect】
The binarization processing device according to the present invention blocks input image data by class dividing means to divide the class into classes, obtains class information corresponding to the class division, and learns isolated point removal processing in accordance with the class information. The conversion means reads the class-specific patterns from the storage means in which the class-specific patterns obtained by performing the above in advance are stored, and converts the input image into the class-specific patterns. In the learning of the isolated point removal processing, a class-specific pattern corresponding to class information is obtained from a learning multi-value image and a learning binary image. Therefore, according to the binarization processing device of the present invention, the isolated point removal processing after the binarization processing becomes unnecessary.
[0030]
Further, in the binarization processing method according to the present invention, the input image data is divided into blocks by a class division step to divide the class, class information corresponding to the class division is obtained, and an isolated point removal processing is performed according to the class information. The conversion step reads the class-specific patterns from the storage means in which the class-specific patterns obtained by performing the learning in advance are stored, and converts the input image into the class-specific patterns. In the learning of the isolated point removal processing, a class-specific pattern corresponding to class information is obtained from a learning multi-value image and a learning binary image. Therefore, according to the binarization processing method of the present invention, the isolated point removal processing after the binarization processing becomes unnecessary.
[Brief description of the drawings]
FIG. 1 is a block circuit diagram showing a schematic configuration of an embodiment of a binarization processing device and a binarization processing method according to the present invention.
FIG. 2 is a block circuit diagram showing a configuration of a class division circuit according to the embodiment shown in FIG. 1;
FIG. 3 is a diagram illustrating a binarization process of a class-specific pattern conversion processing unit according to the embodiment shown in FIG. 1;
[Explanation of symbols]
2 binarization processing device 3 class division processing circuit 4 class-specific pattern conversion processing circuit 5 conversion pattern determination processing circuit 6 Thinning processing device 11 Blocking processing unit 12 Binary processing unit 13 Packing processing unit 14 Screen storage unit 15 Threshold determination processing unit

Claims (4)

In a binarization processing device that binarizes input image data,
Class dividing means for dividing the input image data into blocks and class dividing, and outputting class information corresponding to the class division;
Storage means for storing a class-specific pattern obtained by previously performing learning of the isolated point removal processing;
And a converting means for converting the input image class-specific pattern corresponding to the upper chrysanthemum Las information by reading from the storage means,
The binarization processing device, wherein the learning is to obtain the class-specific pattern corresponding to the class information from a learning multi-valued image and a learning binary image. The class dividing means includes a blocking processing unit configured to block input image data, a threshold determination processing unit configured to determine a threshold value from the input image data, and a blocked input from the blocking processing unit. A binarization processing unit that performs binarization processing on the image data using the threshold value determined by the threshold value determination processing unit, and class-divides the image data binarized by the binarization processing unit 2. The binarization processing device according to claim 1, further comprising a conversion processing unit that converts the data into corresponding class information. In a binarization processing method for binarizing input image data,
A class dividing step of dividing the input image data into blocks and class dividing, and outputting class information corresponding to the class division;
From the storage means by class pattern obtained by performing the learning of the isolated point removal processing in advance are stored, a conversion step of converting the input image by reading a class-specific pattern corresponding to the class information Has,
The above-mentioned learning is a binarization processing method, wherein the class-specific pattern corresponding to the class information is obtained from a learning multi-valued image and a learning binary image. The class division step includes a block processing step of dividing input image data into blocks, an image storage step of storing the input image data for one screen, and an input image data of one screen stored in the image storage step. A threshold value determining process for determining a threshold value; and a binarizing process using the threshold value determined in the threshold value determining process process on the input image data blocked from the blocking process step. 4. The image processing apparatus according to claim 3, further comprising: a binarization processing step of applying; and a conversion processing step of classifying the image data binarized by the binarization processing step and converting the image data into corresponding class information. Binarization processing method.

Images