JPS60126780A - Picture normalizing circuit - Google Patents

Abstract

PURPOSE:To prevent missing of stroke by providing a circuit detecting a thin stroke from a picture before normalizing and applying additional processing of one picture element to the detected part to apply normalizing processing to the processing result of the processing circuit. CONSTITUTION:A manuscript character input picure data Din is stored once in a picture memory 12 and a stored data Din is transmitted to a normalizing processing section 18 via a processing circuit 16. The memory 12 and the circuit 16 are controlled by an address control section 14. Moreover, the processing section 18 applies the normalizing of the picture twice in total each one time longitudinally and laterally respectively. The processing circuit 16 of this constitution is provided with line buffers 20, 22, a register 24 of matrix constitution and a table circuit 26. Then a picture data Dpin from the memory 12 is inputted to the buffer 20 and register 24, and the output of 3X3 picture elements is outputted to the circuit 26 by using the buffer 22. A part of the output for 9 picture elements' share is fed back to the register 24, one picture element is added to the center in response to the shape of the pattern to prevent missing of the stroke.

Description

DETAILED DESCRIPTION OF THE INVENTION TECHNICAL FIELD The present invention relates to an image normalization circuit, and more particularly to one used in an apparatus or system for reading and recognizing images such as handwritten characters.

BACKGROUND ART For example, in an image recognition device such as an OCR (optical code reader), when input image data with irregular shapes such as handwritten characters is subjected to recognition processing, the size of the input image data is normalized. It will be done. - By the way, the normalization processing described above also includes image size reduction processing. For this reason, in the conventional normalization circuit, if there is a thin stroke part lO in the image (a,) before normalization, as shown in FIGS. 1(a) and (b), after the normalization process, In the image (1)), the thin parts of the strokes often disappeared. Such loss of strokes is a major cause of recognition errors.

- Hajime Tomoshika 7 This invention solves the above-mentioned problems. For example, in a device or system that recognizes and processes irregularly shaped image data such as handwritten characters, it is possible to determine the size of handwritten characters. An object of the present invention is to provide an entire image normalization circuit that can prevent the above-mentioned strokes from disappearing during normalization.

Configuration The image normalization circuit according to the present invention includes a circuit that detects a thin stroke part from an image before normalization and adds at least one pixel to this detected part. The processing result is subjected to normalization processing.

Embodiments of the present invention will be described below based on the drawings.

FIG. 2 shows an embodiment of an image normalization circuit according to the present invention. - The image normalization circuit shown in the figure is composed of an image memory 12, an address control section 14, a processing circuit 16, a normalization processing section 18, and the like.

In FIG. 2, input image data Di such as handwritten characters
n is input to the image memory 12 and temporarily stored. The image data stored in the image memory 12 is sent to the normalization processing section 18 via the processing circuit 16ft, where normalization processing involving image reduction is performed. and,
The output DOut of the normalization processing section 18 is sent to a recognition section (not shown) and subjected to recognition processing. Normalization processing unit 18
The normalization process is performed twice in each direction, once in the vertical direction and once in the horizontal direction of the image. The first processing result is returned to the image memory 12, from where it is sent again to the normalization processing section 18 and subjected to the second normalization processing.

FIG. 3 shows details of the processing circuit 16 incorporated into the normalization circuit shown in FIG.

The processing circuit shown in the figure consists of two line roughs 120.
The register 24 has a 22.3×3 matrix configuration, and a table circuit 26.

In FIG. 3, the input image Dp from the image memory 12 is
in is input to line buffer 20 and register 24, respectively. The register 24 is composed of a shift register, from which outputs for 3×3, that is, 9 pixels can be obtained simultaneously. The output of these 9 pixels is input to the table circuit 26, a part of 0 is fed back to the register 24, and the processing result Dpo of 16 of the processing circuit is fed back to the register 24.
ut is taken out from part of the output of the register 24 or the table circuit 26.

When a pattern portion with a thin stroke appears in the input image, the processing circuit 16 having the above-described configuration adds one pixel to the side of the center pixel depending on the shape of the pattern portion. This ensures that at least every stroke has a width of two pixels or more.

FIG. 4 shows an example of the operation of the processing circuit 16. In FIG. 4, (a) and (C) each show a pixel pattern in a part of the human image before processing. Also, the same figure (1)
) shows the pixel pattern after processing the pixel pattern in (a). In addition, in the same figure (l shows a pixel pattern after processing the pixel pattern in (c). Here, the reference numeral 28 indicates a black pixel.

As illustrated in Figure 4, when a part with a thin stroke and only one pixel width is found (detected), a process is performed to add one pixel to the side of the part numbered seven. As a result, it has a width of at least two pixels or more.

As a result, even if normalization processing that involves reduction processing is performed,
The strokes are reliably preserved, and as a result, recognition errors in the recognition section are reliably prevented. .

FIG. 5 shows an example of the operation of the image normalization circuit described above. In the figure, (a) shows the input image before normalization. Also,
(b) shows an image processed by the processing circuit 16. (C) shows an image that has been normalized by the normalization processing unit 18.

As is clear from the figure, even if the input image before normalization has thin strokes, a normalized image without lost strokes can be obtained.

Note that in the circuit of the embodiment described above, the image memory 12
Address control of each of the processing circuits 16 and 16 is performed by a common address control section 4, thereby simplifying the overall circuit structure.

Effects As explained in the embodiment, the image normalization circuit according to the present invention can eliminate the loss of stock in normalized images by simply adding a relatively simple configuration. , this can prevent recognition errors.

[Brief explanation of drawings]

Fig. 1 is a diagram showing an example of the operation of a conventional image normalization circuit, Fig. 2 is a block diagram showing the configuration of an image normalization circuit according to the present invention, and Fig. 3 is a partial diagram of the circuit shown in Fig. 2. FIG. 4 is a block diagram showing the structure of the circuit; FIG. 4 is a diagram showing an example of the operation of the main part of the circuit according to the present invention; FIG. 5 is a diagram showing an example of the operation of the circuit according to the present invention. lO... Thin portion of stroke, 12... Image memory, 14... Address control unit, 16... Processing circuit,
]8... Normalization processing unit, 20. 22... Line buffer, 24... Register, 26... Table circuit, Din... Input image data, DOut... Normalized output, Dpin... Input image of processing circuit, Dpo+
rc... Processing result of the processing circuit, 28... Black pixel. Applicant's agent Kiyoshi Torii Figure 1 (a) (b) Figure 2 Figure 3 Power? Figure 4 Figure 5 Imino (b) (c)

Claims (2)

[Claims] (1) In a circuit that normalizes an input image in order to recognize images such as handwritten characters, a thin stroke part is detected from the image before normalization, and at least one pixel is added to this detected part. An image normalization circuit characterized in that a circuit for performing processing is provided, and the processing results of this processing circuit are subjected to normalization processing. (2) The above processing circuit includes a plurality of line buffers that store binarized input data line by line, a shift register into which the output of the line buffer is simultaneously input for three lines, and a parallel 3 from output
2. The normalization circuit according to claim 1, further comprising a table circuit that obtains a ×3 matrix output and finds a specific pattern from this output.