JPH083827B2 - Character image processing method - Google Patents

Description

The present invention relates to an optical reader (OCR), etc.
The present invention relates to a character image processing method that is preferably implemented.

2. Description of the Related Art An optical reader (OCR) has been used as a kind of input means for a computer or the like. This optical reading device irradiates an original with light from a light source, reads an image with an image pickup device, determines the type of each character (alphanumeric characters or symbols) forming the read image, and performs input. It is a thing.

In such an optical reading device, a character string is extracted from the read image, and then individual characters are extracted for each character string. The extracted character is determined by comparing it with the shape patterns of all the characters stored as a dictionary by the optical reading device.

Problems to be Solved by the Invention Since the conventional optical reader compares the extracted character with the shape patterns of all the characters stored in the dictionary, it is proportional to the number of shape patterns. The time required for discrimination becomes long.

One prior art that solves this problem is disclosed in
55-112687. In this prior art, a character group is determined using the height of characters in a form and a threshold calculated using a predetermined constant.

In such a prior art, since a character font changes in a general input document, a relative upper or lower coordinate difference length from a character height or from a reference line also changes variously. Therefore, when such a prior art is applied to a general manuscript, it is not possible to cope with a change exceeding the threshold value, and it becomes impossible to accurately determine a character group.

An object of the present invention is to provide a character image processing method in which the time required for character discrimination is shortened and the discrimination rate is improved.

Means for Solving the Problems According to the present invention, a character string composed of a plurality of characters arranged along one arrangement direction is read and stored in the image memory 3, the stored contents of the image memory 3 are read, and the characters are read. The upper basic line l2 and the lower basic line 1, which are parallel to each other and indicate the vertical occupying ranges of the main parts of the characters included in the column, are defined respectively, and the upper basic line l2 and the upper extraction coordinates of the plurality of characters are defined. When the absolute difference distance is calculated and the frequency of each absolute difference distance is calculated, and when there are two groups 22 and 23 having a high frequency, those groups 22 and 23 along the upper and lower basic lines L2,1 in the vertical direction. Coordinates between and upper basic line l
The upper threshold line l8 parallel to the upper basic line l2 is defined between the groups 22 and 23 based on the coordinates of 2 and the absolute difference distance between the lower basic line 1 and the lower extracted coordinates of a plurality of characters is calculated. The frequency of each absolute difference distance is calculated, and when there are two groups with high frequency, from the coordinates between the upper and lower basic lines 1 and 12 in the vertical direction and the coordinate of the lower basic line 1, A lower threshold line l9 parallel to the lower basic line 1 is defined between these groups, and a plurality of characters are set according to the positional relationship among the upper basic line l2, the lower basic line 1, the upper threshold line l8, and the lower threshold line l9. It is a character image processing method characterized by being classified into a character group.

Operation According to the present invention, a plurality of characters are arranged along one arrangement direction to form a character string, the character string is read and stored in the image memory 3, and the stored contents of the image memory 3 are read out, An upper basic line 12 and a lower basic line 1 are defined respectively. The technique for defining the upper basic line 12 and the lower basic line 1 can be performed by a known method.

Therefore, the frequency of the absolute difference distance between the upper basic line l2 and the upper extracted coordinates of each character is calculated, and when there are two groups 22 and 23 having a high frequency, those groups 22,23
In an embodiment to be described later, the upper threshold line l8 is determined from the coordinates of the valley and the coordinates of the upper basic line l2, and similarly, the absolute difference distance between the lower basic line 1 and the lower extracted coordinates of each character is calculated. The lower threshold line l9 is defined, and in this way, it is classified into a plurality of character groups according to the positional relationship between the upper and lower basic lines l2,1 and the upper and lower threshold lines l8, l9. In this way, the upper threshold line l8 and the lower threshold line l9 are set dynamically in correspondence with the height of each character, so that they can dynamically respond to changes in the character font of general manuscripts, and a stable character group. That is, the character group can be accurately determined.

Example First, a concept called a basic line is used to classify characters. FIG. 2 is a diagram explaining this basic line. After being read, virtual lines 1 and l2 indicating the vertical occupancy range of the main part of the character are assumed along the character array direction for the character columns divided for each arranged row, and each character is Basically, they are shown aligned on the virtual line 1.

At this time, some of the English characters "g" and "y" project below the virtual line 1, and some of the English characters "h" and "k" project above the virtual line l2. The main part of each character will be written between the virtual lines 1 and 12.
Such virtual lines 1 and 12 are referred to as a lower basic line and an upper basic line.

Such a lower basic line 1 and an upper basic line
FIG. 3 shows a concrete example of characters classified by using l2. In FIG. 3 (1), a part of the character projects above the upper basic line 12 and is referred to as an upper projecting character. In FIG. 3 (2), a part of the character projects below the lower basic line 1 and is referred to as a lower projecting character. In FIG. 3 (3), a character is written between the lower basic line 1 and the upper basic line 12 and is called an intermediate character. In FIG. 3 (4), a part of the character projects below the lower basic line 1 and above the upper basic line 12 and is referred to as the other character. FIG. 3 (5) is a symbol.

FIG. 1 is a block diagram showing a basic configuration of an optical reading device 1 according to the present invention.

The optical reading device 1 includes an image input device 2 using a solid-state image pickup device, an image memory 3 in which an image read by the image input device 2 is stored, and a character string image in which an extracted character string image is stored. Memory 4 and upper basic line coordinate memory 5 in which the coordinates of the upper basic line 12 and the lower basic line 1 extracted for each character string are stored.
And a lower basic line coordinate memory 6, a character extraction coordinate memory 7 in which the extracted coordinates of the extracted character are stored, and a histogram obtained from the absolute difference distance between the character extraction coordinates and the upper basic line coordinates or the lower basic line coordinates. Is calculated from a histogram memory 8 in which is stored, an upper threshold coordinate memory 9 and a lower threshold coordinate memory 10 in which upper threshold coordinates and lower threshold coordinates detected from the histogram are stored, and an upper basic line 12 and a lower basic line 1. A center line coordinate memory 11 that stores the coordinates of the center line and a character group flag memory 12 that stores the character group flag of the determined character.
And a control unit 13 for uniformly controlling the operation of each of the above components
It is comprised including. The coordinates mentioned above mean the second
It is a coordinate that represents the vertical position in FIGS. 3, 3 and 7, and the lines 1 to 19 extending in the horizontal direction on the left and right in FIGS. 2, 3 and 7 are parallel to each other, Therefore, each of the lines 1 to 19 can be specified by the coordinates. The up and down directions in FIGS. 2, 3, and 7 correspond to the left and right horizontal directions in FIG. 6 described later.

FIG. 4 is a flow chart for explaining the basic operation of the optical reader 1.

In step a1, the image read by the image input device 2 is stored in the image memory 3. In step a2,
A character string image is extracted from the images stored in the image memory 3 and stored in the character string image memory 4. In step a3, the upper basic line l2
And the lower basic line 1 are extracted, and virtual coordinate axes in the vertical direction perpendicular to the upper and lower basic lines l2, 1 in FIGS. 2, 3, and 7 are set in the character string image memory 4. , Upper basic line coordinates and lower basic line coordinates are obtained and stored in the upper basic line coordinate memory 5 and the lower basic line coordinate memory 6. In step a4, a character is extracted from the character string image, and the extracted coordinates on the virtual coordinate axis are stored in the character extracted coordinate memory 7.

At step a5, the upper threshold coordinate is detected. FIG. 5 is a flowchart for detecting upper threshold coordinates, and FIG. 6 is
An example of the histogram created when the upper threshold coordinate is detected is shown. A method of detecting the upper threshold coordinate will be described with reference to FIGS. 5 and 6.

In step b1, the absolute difference distance (distance between the upper basic line coordinate and the upper extracted coordinate of each character) is calculated.
At step b2, a histogram as shown in FIG. 6 is created based on the absolute difference distance and stored in the histogram memory 8. In step b3, the two mountain parts formed in the histogram (the mountain part 22 formed by the character upper extraction coordinates existing near the upper basic line 12 and the character upper extraction coordinates existing far from the upper basic line 12) are formed. The valley 21 between the peaks 23) is detected. At step b4, upper threshold coordinates are obtained from the valley coordinates on the histogram and the coordinates of the upper basic line 12 and stored in the upper threshold coordinate memory 9. As an example of a method of obtaining the upper threshold coordinate, for example, the valley coordinate and the coordinate of the upper basic line l2 are subtracted, and the offset pixel number ΔA (see FIG. 6) which is the difference is obtained, and this offset is calculated. The number of pixels ΔA can be added to the coordinates of the upper basic line 12 to obtain the upper threshold coordinates. The upper threshold coordinate is the coordinate of the upper threshold line l8 described later.
This also applies to the lower threshold coordinates described below.

Thus, the upper threshold coordinate is set to the value indicated by the line l8, which is the upper and lower basic lines.
It is defined as an upper threshold line described later that is parallel to l2,1.

Subsequently, in step a6 of FIG. 4, the lower threshold coordinate is detected by the same method as in step a5 and stored in the lower threshold coordinate memory 10.

From step a7 to step a11, the character group is determined. FIG. 7 is a diagram showing a relative positional relationship between a virtual xy coordinate axis, a character, a reference line used for character discrimination, and a character extraction line. The operation of steps a7 to a11 will be described with reference to FIG.

At step a7, the symbol group is determined. First,
The center line l3 is obtained from the upper basic line 12 and the lower basic line 1, and the center line coordinates are stored in the center line coordinate memory 11. As shown in FIG. 7 (1), when the y-coordinate value of the upper extraction line l4 is smaller than the y-coordinate value of the center line l3, that is, the extraction line above the center line l3
If l4 exists below, it is judged as "." or ",". (Similarly, the y coordinate of the virtual coordinate axis is used to determine whether the line is up or down.) If the lower extraction line l5 is above the center line l3, it is determined to be "'" or "".

To determine the symbol "." Or "-", the symbol upper threshold line l6 and the symbol lower threshold line l7 are detected with 1/4 of the width of the upper basic line 12 and the lower basic line 1 as the threshold. The upper extraction line l4 is below the symbol upper threshold line l6, and the lower extraction line l5 is the symbol lower threshold line
If it exists above l7, it is judged to be the symbol "." or the symbol "-". When it is determined that the character group is a symbol as described above, the symbol group flag is set in the corresponding portion of the character group flag memory 12.

At step a8, the upper protruding character group is determined. For the character not determined to be the symbol group in step a7, the upper extraction line l4 exists above the upper threshold line l8 and the lower extraction line l5 exists above the lower threshold line l9 as shown in FIG. 7 (2). In this case, the character is determined to be the upper protruding character, and the upper protruding character group flag is set in the corresponding portion of the character group flag memory 12.

At step a9, the lower protruding character group is determined. For characters that are not determined to be either the symbol group or the upper protruding character group in steps a7 and a8, as shown in FIG. 7 (3), the upper extraction line l4 is below the upper threshold line l8, and the lower part is below the upper threshold line l8. If the extraction line l5 exists below the lower threshold line l9, it is determined to be a lower protruding character, and the lower protruding character group flag is set in the corresponding portion of the character group flag memory 12. At step a10, the determination of the intermediate character group is performed. For characters that are not determined to be any of the symbol group, the upper protruding character group, or the lower protruding character in step a7, step a8, or step a9, as shown in FIG. 7 (4), the upper extraction line l4 is the upper threshold line. l8 below and below l5 l5
Is above the lower threshold line l9, it is determined to be an intermediate character group, and the character group flag memory 12
The intermediate character group flag is set to the corresponding portion of the.

At step a11, determination of other character groups is performed. In step a7, step a8, step a9, and step a10, the character (character as shown in FIG. 7 (5)) which is not judged to be any character group is judged to be another character, and the character group flag memory 12 The other character group flags are set in the corresponding portions of.

Thus, from step a7 to step a11,
A character group extracted from the read image is
It can be limited to one of the five groups of the upper protruding character group, the lower protruding character group, the intermediate character group, and the other character groups. Further, in the case of a symbol, it is possible to determine not only the symbol group but also the type to some extent.

Effect As described above, according to the present invention, the upper and lower basic lines l2,1 indicating the vertical occupying ranges of the main portions of the plurality of characters forming the character string that are read and stored in the image memory 3 are formed. In addition to defining each, the extracted coordinates of the upper and lower parts of each character and the upper and lower basic lines
The frequency is calculated based on the absolute difference distance from l2,1, and the upper and lower threshold lines l8, l9 between the two groups 22 and 23 having the high frequency are determined. In this way, the upper and lower basic lines l2,1
1 and upper and lower threshold lines l8 and l9 are used to classify each character into a plurality of character groups, so that it can dynamically respond to changes in the character font of general manuscripts and accurately classify stable character groups. You will be able to do it.

[Brief description of drawings]

FIG. 1 is a block diagram showing the basic configuration of the optical reading device 1, FIG. 2 is a diagram for explaining basic lines, and FIG. 3 is a diagram showing specific examples of characters classified using the basic lines.
4 is a flow chart showing the basic operation of the optical reader 1, FIG. 5 is a flow chart showing the operation at the time of detecting the threshold coordinate, FIG. 6 is a histogram created at the time of detecting the threshold coordinate, and FIG. 7 is a virtual coordinate axis. It is a figure which shows the relative positional relationship of the character and the reference line used at the time of character discrimination, and the character extraction line. 1 ... Optical reading device, 2 ... Image input device, 3 ... Image memory, 4 ... Character string image memory, 5 ... Upper basic line coordinate memory, 6 ... Lower basic line coordinate memory, 7 ... Character extraction coordinate memory, 8 ... Histogram memory, 9
… Upper threshold coordinate memory, 10… Lower threshold coordinate memory, 11…
Central line coordinate memory, 12 ... Character group flag memory, 13 ... Control unit, 1 ... Lower basic line, l2 ... Upper basic line, l3 ... Center line, l4 ... Upper extraction line, L5 ... Lower extraction line, L6 ... Symbol Upper threshold line, l7 ... Lower threshold line for symbols, l8 ... Upper threshold line, l9 ... Lower threshold line

Claims (1)

[Claims] 1. A character string composed of a plurality of characters arranged in one array direction is read and stored in an image memory 3, the stored contents of the image memory 3 are read, and the main characters of the characters included in the character string are read. The upper basic line l2 and the lower basic line 1 which are parallel to each other indicating the respective occupied ranges in the vertical direction of the part are defined respectively, and the absolute difference distance between the upper basic line l2 and the upper extracted coordinates of each character is calculated, and The frequency of each absolute difference distance is calculated, and when there are two groups 22 and 23 with high frequency, the coordinates between the groups 22 and 23 along the vertical direction of the upper and lower basic lines l2,1 and the upper basic line. l
The upper threshold line l8 parallel to the upper basic line l2 is defined between the groups 22 and 23 based on the coordinates of 2 and the absolute difference distance between the lower basic line 1 and the lower extracted coordinates of the plurality of characters is calculated. The frequency of each absolute difference distance is obtained, and when there are two groups with high frequency, from the coordinates between the upper and lower basic lines 1 and 12 in the vertical direction and the coordinate of the lower basic line 1, A lower threshold line l9 parallel to the lower basic line 1 is defined between the groups, and a plurality of characters are set according to the positional relationship between the upper basic line l2, the lower basic line 1, the upper threshold line l8, and the lower threshold line l9. Character image processing method, characterized in that the character image processing method is characterized in that