JPH096912A - Braille recognition system - Google Patents

Abstract

PURPOSE: To provide a braille recognition system which automatically reads braille and automatically generates a code or ink character corresponding to the braille. CONSTITUTION: The braille recognition system is equipped with an image pickup means 1 which irradiates a braille document surface obliquely with light and receives its reflected light to obtain a gray level image, pattern image generating means 41, 43, and 44 which obtain a pattern image of raised points on the braille document, a braille image segmenting means 45 which segments images out of the pattern image of the raised points in braille units, and a braille-code converting means 47 which generates braille codes by the images of the segmented braille units.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a braille recognition system, a braille pattern image generating device, a braille image cutting device and a braille / code converting device.

[0002]

2. Description of the Related Art In the currently used six-point braille, as shown in FIG. 9, three points in the vertical direction and six points in the two horizontal rows are regarded as one unit, and characters are formed by combining the projections of the convex points. Is configured. This unit is called "mass".

[0003] The topmost point in the left column in each cell is a point,
The convex point in the middle row of the left column is a point, the convex point in the bottom row of the left column is a point, the convex point in the top row of the right column is a point, the convex point in the middle row of the right column is a point, and the convex point in the right row is a point. The lower convex points are called points.

[0004] The interval W1 between adjacent left and right convex points in the same cell is usually set to 2.0 to 2.1 mm. In addition, the interval W2 between the adjacent convex points on the left and right between the adjacent cells is usually set to 3.0 to 3.1 mm.

[0005] Braille is written horizontally, and is read from left to right. In the case of six-point braille, 63 kinds of different combinations can be made, but not all characters of kana letters, numbers and alphabets can be represented by 63 kinds. Therefore, when expressing numbers and alphabets, a code called a prefix is attached before the numbers and alphabets to identify the character system.

[0006] Characters in printed matter are called ink characters in contrast to braille characters. Also, as a code corresponding to Braille, generally, a North American Braille code (NABCC: North AmericanBr)
aille Computer Code).

[0007]

Braille characters are expressed by convex dots, unlike general characters. Braille is represented by the presence or absence of six convex dots. Therefore, the present applicant has developed a braille recognition system using a recognition method different from a general character recognition method.

The present invention automatically reads Braille,
An object of the present invention is to provide a Braille recognition system that automatically generates a code or an ink character corresponding to Braille.

It is another object of the present invention to provide a Braille pattern image generating apparatus capable of obtaining an image corresponding to a pattern of convex points of a Braille document.

Another object of the present invention is to provide a braille image cutting device capable of cutting an image in braille units from an image corresponding to a pattern of convex dots of a braille document.

It is another object of the present invention to provide a braille / code conversion device which can automatically generate a code corresponding to an image based on an image corresponding to a pattern of convex points in Braille. And

[0012]

A braille recognition system according to the present invention illuminates the surface of a braille document with light from an oblique direction,
By receiving the reflected light, an image pickup unit that obtains a grayscale image, based on the grayscale image obtained by the image pickup unit,
A pattern image generation means for obtaining a pattern image of convex dots on a braille document, a braille image cutout means for cutting out an image for each braille unit from a pattern image for convex dots, and a braille code for each cut-out image in braille units Braille / code converting means for

Code / ink-character conversion means for converting the braille code obtained by the braille / code conversion means into ink characters may be provided.

As the pattern image generating means, for example, a grayscale image obtained by the image pickup means is subjected to a filter process, so that a portion of the convex point on the Braille document is strongly irradiated with light and a portion having a shadow. Filter processing means for extracting the intermediate part of the, and binarization processing means for performing binarization processing on the filtered image to obtain a binarized image corresponding to the pattern of convex points on the Braille document. What is provided is used.

When the direction of the grayscale image obtained by the image pickup means is defined as the upper portion of the convex point on the Braille document where the light is strongly irradiated, and the lower portion of the shaded portion, the filter processing means For example, a method is used in which the density value of the target pixel is converted into a value obtained by subtracting the density value of the pixel immediately above the target pixel from the density value of the pixel directly below the target pixel.

As the Braille image cutting-out means, for example, projection processing means for performing projection processing on a pattern image of convex points, and the vertical and horizontal lengths of each braille unit based on the projection processing result by the projection processing means. An information extraction unit that extracts information and information about the position of each braille unit, and a unit that cuts out an image for each Braille unit from a pattern image of convex points based on the obtained information are used.

As the projection processing means, means for performing projection processing in the first direction, which is the same direction as the reading direction of braille in the pattern image of convex dots, and orthogonal to the reading direction of braille in the pattern image of convex dots. What is provided with a means for performing a projection process in a direction is used.

As the Braille / code conversion means, for example, the cut-out image in Braille units is divided into a plurality of regions in which convex points can be formed, and for each divided region,
Based on the discrimination means for discriminating the presence / absence of a convex point, the discrimination result for the presence / absence of a convex point for each divided area, and the weighting coefficient set in advance for each divided area, corresponding to the image in the Braille unit cut out What is provided with a code generation means for generating a Braille code is used.

As the determining means, for example, the presence or absence of a convex point is determined for each divided area based on the number of significant pixels in each divided area or the number of significant pixels in a small area set in each divided area. What is done is used.

The braille pattern image generating apparatus according to the present invention illuminates the surface of a braille document from an oblique direction and receives the reflected light to obtain a grayscale image, and a grayscale image obtained by the imaging means. By performing a filtering process on the braille document, a filtering process means for extracting an intermediate part between a part where strong light is projected and a part where a shadow is formed at a convex point on the braille document, and a binary value for the filtered image. A binarization processing means for performing binarization processing to obtain a binarized image corresponding to a pattern of convex points on the Braille document is provided.

When the direction of the grayscale image obtained by the image pickup means is defined as the upper portion of the convex point on the Braille document where the light is strongly irradiated, and the lower portion of the shaded portion, the filter processing means For example, a method is used in which the density value of the target pixel is converted into a value obtained by subtracting the density value of the pixel immediately above the target pixel from the density value of the pixel directly below the target pixel.

The braille image clipping device according to the present invention is
Based on the projection processing means for performing the projection processing on the pattern image of the convex points on the braille document, and the information on the vertical and horizontal lengths of each braille unit and the position of each braille unit based on the projection processing result by the projection processing means. It is characterized in that it is provided with an information extracting means for extracting information, and a means for cutting out an image for each Braille unit from a pattern image of convex points based on the obtained information.

As projection processing means, means for performing projection processing in the first direction, which is the same direction as the Braille reading direction in the convex dot pattern image, and orthogonal to the Braille reading direction in the convex dot pattern image. What is provided with a means for performing a projection process in a direction is used.

The Braille / code converter according to the present invention is
The braille dot convex point pattern image is divided into a plurality of areas in which convex points can be formed, and a discriminating means for discriminating the presence or absence of a convex point for each divided area, and a convex point for each divided area. It is characterized by comprising code generation means for generating a braille code corresponding to the cut-out image in a unit of Braille, based on the presence / absence determination result and a weighting coefficient predetermined for each divided area.

As the determining means, for example, the presence or absence of a convex point is determined for each divided area based on the number of significant pixels in each divided area or the number of significant pixels in a small area set in each divided area. What is done is used.

[0026]

In the Braille recognition system according to the present invention, a grayscale image is obtained by irradiating the surface of the Braille document with light obliquely and causing the reflected light to be received by the image pickup means. Next, a pattern image of convex points on the Braille document is obtained based on the obtained grayscale image. Next, an image is cut out for each Braille unit from the pattern image of convex points. Then, a Braille code is generated for each cut-out image in Braille units.

In the Braille pattern image generating apparatus according to the present invention, a light and shade image is obtained by irradiating the surface of the Braille document with light obliquely and causing the reflected light to be received by the image pickup means. By performing a filtering process on the obtained grayscale image, an intermediate portion between a portion of the convex point on the Braille document which is strongly irradiated with light and a portion where a shadow is formed is extracted. Then, binarization processing is performed on the image after the filtering processing, and the binary processing corresponding to the pattern of the convex points on the Braille document
A binarized image is obtained.

In the braille image clipping device according to the present invention, the projection process is performed on the pattern image of the convex points on the braille document. Information about the vertical and horizontal lengths of each braille unit and information about the position of each braille unit are extracted based on the projection processing result. Then, based on the obtained information, an image is cut out for each Braille unit from the pattern image of convex points.

In the Braille / code conversion device according to the present invention, the convex point pattern image in Braille units is divided into a plurality of areas for each area where convex points can be formed. The presence / absence of a convex point is determined for each divided area. Then, a braille code corresponding to the cut-out image in braille units is generated based on the result of the presence / absence determination of the convex point for each divided region and the predetermined weighting coefficient for each divided region.

[0030]

Embodiments of the present invention will be described below with reference to FIGS.

FIG. 1 shows the overall structure of the Braille recognition system.

The Braille recognition system includes a Braille reading unit 1.
And a Braille recognition unit 2. Braille reading unit 1
Is a manuscript in Braille (hereinafter referred to as Braille document) 10
In addition, the light irradiator 1 for irradiating light obliquely from above.
1, and a CCD camera 14 that receives reflected light from the Braille document 10 via a mirror 12 and a lens 13.

The light irradiation unit 11 is a light source 2 such as a fluorescent lamp.
1, a reflection plate 22 and a filter 23 for correcting uneven brightness. The light emitted from the light source 21 toward the uneven brightness correction filter 23 and the light emitted from the light source 21 and reflected by the reflection plate 22 pass through the uneven brightness correction filter 23 and are obliquely directed to the surface of the Braille document 10. Irradiated.

As described above, the Braille document 10 is irradiated with light obliquely from above in order to determine the presence or absence of the convex points forming the Braille by the shadow formed by the convex points.

FIG. 2 shows a grayscale image of a convex point and its surroundings taken by the CCD camera 14.

Since the convex point 30 forming the Braille is hemispherical, a shadow is formed on the portion 33 on the side opposite to the side on which light is incident. Therefore, the light-incident side portion 31 of the convex point has a color closer to white, which has a lower density than the background, and the portion 33 on the opposite side to the light incident side has a color of higher density, close to black, and the intermediate portion 32 thereof. Then, it becomes gray with almost the same density as the background.

Next, the operation of the Braille recognition unit 2 will be described. The video signal output from the CCD camera 14 is converted into a digital signal (pixel data) by the image capturing unit 41 and then stored in the image memory 42.

In this example, it is assumed that the digital signal is a signal representing the density value of the grayscale image with 8 bits. The light-incident side part 31 of the convex point has a value close to the density value "255" representing white, and the part 33 on the side opposite to the light-incident side has a value close to the density value "0" representing black. The intermediate portion 32 and the background have a value slightly smaller than the density value of the side portion 31 on which the light of the convex point is incident. In this example, the size of one convex point is about 3 × 3 pixels.

Next, the pixel data stored in the image memory 42 is filtered by the filtering section 43.

In this filter processing, the image shown in FIG. 2 is subjected to the filter processing by using a 3 × 3 filter as shown in FIG. That is, assuming that the density data of each pixel in the 3 × 3 matrix centered on the pixel of interest is A to I as shown in FIG. 6, the density data of the pixel of interest is the density expressed by the following formula 1. Converted to data e.

[0041]

[Equation 1]

By carrying out such density conversion for all pixels, an image as shown in FIG. 3 is obtained. That is, in FIG. 2, the light-incident side portion 31, the light-incident side portion 33, and the background portion of the convex point have values close to black (values close to “0”), and the middle of the convex points. Part 3
2 is a value with a lower density than that.

Next, the binarization processing unit 44 performs binarization processing on the filtered pixel data. This binarization process is a process for distinguishing between the light-incident side portion 31, the light-incident side portion 33 and the background portion, and the convex-point intermediate portion 32. As the threshold value, for example, "48" is used.

That is, of the pixel data after the filtering process, the image data whose density value is greater than "48" is the value "
Image data having a density value of "48" or less is converted into black data having a value of "0". As a result, an image as shown in FIG. That is, the black data corresponds to the convex points formed in the Braille document 10.

Next, the cutout processing unit 45 performs a cutout process for cutting out each square from the binarized image. It is assumed that FIG. 7 shows a binarized image (convex pattern image) obtained by the binarization processing unit 44. As shown in FIG. 7, each square has 6 points, 3 points vertically and 2 rows horizontally.
It is made up of dots. In FIG. 7, black circles represent convex points.

In the cut-out processing section 45, horizontal projection calculation and vertical projection calculation are performed. In the horizontal projection calculation, pixel values are added in the horizontal direction (X direction). That is, if each pixel value of the binarized image is f (x, y),
As shown in Equation 2, the sum H (y) of the density values of the pixels included in each line is calculated for each horizontal line.

[0047]

[Equation 2]

In the vertical projection calculation, pixel values are added in the vertical direction (Y direction). That is, assuming that each pixel value of the binarized image is f (x, y), as shown in Formula 3, for each line in the vertical direction, the sum V of the density values of the pixels included in each line is V. (X) is calculated.

[0049]

(Equation 3)

Based on the horizontal projection calculation result, a horizontal projection histogram is obtained as shown in FIG.
Further, a vertical projection histogram is obtained as shown in FIG. 7 based on the vertical projection calculation result.

Then, based on these histograms, as shown in FIGS. 7 and 8, the vertical length L of the frame,
A horizontal length W and a reference position P indicating the position of each frame are obtained. These pieces of information are stored in the position information storage memory 46. Then, each frame is cut out one by one based on the information stored in the position information storage memory 46.

The Braille / code conversion processing unit 47 performs the Braille / code conversion processing on the cut-out frame. FIG. 8 shows an image of one frame that has been cut out. One frame image cut out corresponds to 6 regions in which convex points can be formed in each square, and is 6 in the vertical direction and 3 in the horizontal direction.
It is divided into two areas E _{1 to} E ₆ . Further, as indicated by the broken line in FIG. 8, areas e _{1 to} e ₆ smaller than that are set in the areas E _{1 to} E ₆ . In this example, each area E ₁ ~
The size of E ₆ is about 6 × 6 pixels, and each of the regions e _{1 to} e ₁
The size of ₆ is about 4 × 4 pixels.

Then, for each of the areas e _{1 to} e ₆ , based on the number of pixels (significant pixels) having the value “1” contained therein,
The value of the convex point presence / absence identification data A _i is determined. Where i
Indicates a region e ₁ to e _6, corresponding to e ₁ to e _6. 1 to
It takes a value of 6. For example, the value “1” is set in the areas e _{1 to} e ₆ .
It is determined whether or not the number of pixels of is a predetermined value, for example, 3 or more. If there are three or more pixels with the value "1", it is determined that there is a convex point in the area, and the convex point presence / absence identification data A _i is "
On the contrary, if the number of pixels having the value “1” is less than 3, it is determined that there is no convex point in the area, and the convex point presence / absence identification data A _i is “0”.

Thus, for one frame, e
_When the convex point presence / absence identification data A _{1 to} A ₆ corresponding to _{1 to} e ₆ are determined, the Braille code C is created based on the following formula 4.

[0055]

(Equation 4)

In Equation 4, M _i is each of the areas e _{1 to} e
_The weighting factors M _{1 to} M ₆ are determined respectively corresponding to ₆ , and the weighting factors M _{1 to} M ₆ are set as shown in the following formula 5.

[0057]

(Equation 5)

[0058] By this, for the cut out frame
Thus, codes "0" to "63" are assigned. Soshi
The assigned code is NABCC (North American Braille code
Stored in the code storage memory 48 after being converted to
It is. Such processing is performed for all frames.
You.

Next, the Braille / ink-character conversion processing unit 49 performs Braille / ink-character conversion processing on each code stored in the code storage memory 48. In this Braille / ink-character conversion process, the code is converted to ink-character based on a predetermined rule. The obtained ink character data is stored in the ink character storage memory 50.

Further, the obtained ink characters are displayed on the monitor 52 via the output processing section 51. Also, if necessary, the voice is synthesized by the printer 53 and the voice synthesizer 5
The voice is output by 4.

[0061]

According to the Braille recognition system of the present invention, it is possible to automatically read Braille and automatically generate a code or an ink character corresponding to Braille.

According to the Braille pattern image generating apparatus of the present invention, it is possible to obtain an image corresponding to the pattern of convex dots of a Braille document.

According to the braille image clipping device of the present invention, from the image corresponding to the pattern of the convex dots of the braille document,
Images can be cut out in Braille units.

According to the braille / code conversion device of the present invention, it is possible to automatically generate a code corresponding to an image based on an image corresponding to a pattern of convex points in Braille.

[0065]

[Brief description of drawings]

FIG. 1 is a block diagram illustrating an overall configuration of a Braille recognition system.

FIG. 2 is a schematic diagram showing an image of a convex point and its surroundings captured by a CCD camera.

FIG. 3 is a schematic diagram showing an image that has been filtered by a filter processing unit 43.

FIG. 4 is a schematic diagram showing a binarized image obtained by a binarization processing unit 44.

FIG. 5 is a schematic diagram showing a filter used in a filter processing unit 43.

FIG. 6 is a schematic diagram showing density values of pixels in a 3 × 3 matrix centered on a target pixel.

FIG. 7 is an explanatory diagram for explaining a cutout process by a cutout processing unit 45.

FIG. 8 is an explanatory diagram for explaining the Braille / code conversion processing by the Braille / code conversion processing unit.

FIG. 9 is an explanatory diagram for explaining Braille.

[Explanation of symbols]

 1 Braille reading unit 2 Braille recognition unit 10 Braille document 11 Light irradiation unit 12 Mirror 13 Lens 14 CCD camera 41 Image capturing unit 43 Filter processing unit 44 Binarization processing unit 45 Cutout processing unit 47 Braille / Code conversion processing unit 49 Code / Ink conversion processing unit

Claims (14)

[Claims] 1. An image pickup unit for obtaining a grayscale image by irradiating the surface of a Braille document with light from an oblique direction and receiving the reflected light, and a projection on a Braille document based on the grayscale image obtained by the image pickup unit. A pattern image generating means for obtaining a dot pattern image, a braille image cutting means for cutting an image for each braille unit from a convex dot pattern image, and a braille / code conversion for generating a braille code for each cut braille unit image Braille recognition system comprising means. 2. The braille recognition system according to claim 1, further comprising code / ink conversion means for converting the braille code obtained by the braille / code conversion means into ink characters. 3. The pattern image generation means performs a filtering process on the grayscale image obtained by the image pickup means, so that a portion of the Braille document at which the light is projected strongly at a convex point and a portion where a shadow is formed are formed. Filter processing means for extracting an intermediate portion, and binarization processing means for performing binarization processing on the filtered image to obtain a binarized image corresponding to a pattern of convex points on a Braille document. The braille recognition system according to claim 1, which is provided. 4. When the direction of the grayscale image obtained by the image pickup means is defined as the upper part of the convex point on the Braille document to which light is strongly irradiated and the lower part of the shaded part, the filter processing means 4. The Braille recognition system according to claim 3, wherein the density value of the target pixel is converted into a value obtained by subtracting the density value of the pixel directly above the target pixel from the density value of the pixel directly below the target pixel. 5. The braille image cutout means is a projection processing means for performing projection processing on a pattern image of convex points, and information on the vertical and horizontal lengths of each braille unit based on the projection processing result by the projection processing means. And information extracting means for extracting information on the position of each braille unit, and means for cutting out an image for each braille unit from a pattern image of convex points based on the obtained information. The braille recognition system according to any one of 3 and 4. 6. The projection processing means is a means for performing projection processing in a first direction which is the same direction as the Braille reading direction in a convex point pattern image, and a projection point orthogonal to the Braille reading direction in the convex point pattern image. The Braille recognition system according to claim 5, further comprising means for performing projection processing in a direction. 7. The Braille / code conversion means divides the cut-out image in Braille units into a plurality of regions in which convex points can be formed, and determines the presence or absence of convex points for each divided area. Code generation for generating a Braille code corresponding to the cut-out image in Braille unit based on the discrimination means, the result of discrimination of the presence or absence of a convex point for each divided area, and the weighting coefficient predetermined for each divided area The Braille recognition system according to any one of claims 1, 2, 3, 4, 5 and 6, further comprising: 8. The determining means determines the presence or absence of a convex point for each divided area based on the number of significant pixels in each divided area or the number of significant pixels in a small area set in each divided area. The braille recognition system according to claim 7, which is a thing. 9. An image pickup device for obtaining a grayscale image by irradiating the surface of a braille document with light from an oblique direction and receiving the reflected light, and by performing a filtering process on the grayscale image obtained by the image pickup device. , A filter processing means for extracting an intermediate portion between a portion of the convex point on the braille document where light is strongly irradiated and a portion where a shadow is formed, and binarization processing is performed on the image after the filter processing, and the braille document A braille pattern image generation apparatus comprising: a binarization processing unit that obtains a binarized image corresponding to the pattern of the convex points above. 10. When the direction of the grayscale image obtained by the image pickup means is defined as the upper part of the convex point on the Braille document to which light is strongly irradiated and the lower part of the shaded part, the filter processing means 10. The braille pattern image generation device according to claim 9, wherein the density value of the pixel of interest is converted into a value obtained by subtracting the density value of the pixel immediately above the pixel of interest from the density value of the pixel immediately below the pixel of interest. 11. A projection processing unit for performing a projection process on a pattern image of convex points on a Braille document, and information on the vertical and horizontal lengths of each Braille unit and each information based on the projection processing result by the projection processing unit. A braille image cropping device comprising: an information extracting means for extracting information on a position in braille units; and a means for cropping an image for each braille unit from a pattern image of convex points based on the obtained information. 12. The projection processing means performs projection processing in a first direction, which is the same direction as the Braille reading direction on the convex dot pattern image, and orthogonal to the Braille reading direction on the convex dot pattern image. The braille image cutting device according to claim 11, further comprising means for performing a projection process in a direction. 13. A convex point pattern image in Braille units is divided into a plurality of areas in which convex points can be formed, and a discrimination means for discriminating the presence or absence of convex points in each divided area, and each divided area. The braille code is provided with a code assigning means for generating a braille code corresponding to the cut-out image in the braille unit based on the result of the presence / absence determination of the convex point for each of the divided areas and the predetermined weighting coefficient for each divided area. / Code conversion device. 14. The discriminating means discriminates the presence or absence of a convex point for each divided area based on the number of significant pixels in each divided area or the number of significant pixels in a small area set in each divided area. The braille / code conversion device according to claim 13, which is a device.