JPS5839356B2 - Formattuto Betsu Ninin Shikisuru - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a real-time character recognition device that identifies characters or figures by focusing on the characteristics of the characters or figures when they are drawn with a pen on the screen of a figure input device.

Traditionally, in real-time character recognition, the characters to be identified are kanji and
If there is a wide variety of characters such as katakana, hiragana, alphabets, numbers, etc., the time required to search for discriminant logic will increase. Another drawback was that if the two were very similar, they could not be distinguished and could be confused.

The purpose of this invention is to check the formant and search only the identification logic of the format for documents with a predetermined format such as numbers or katakana in any line or column, thereby saving processing time and To provide a real-time character recognition device that eliminates confusion and recognizes characters in different formats in a more convenient and easy-to-use manner.

Next, the present invention will be explained with reference to the drawings.

Referring to FIG. 1, there is shown a block diagram of an embodiment of a real-time character recognition system according to the present invention.

[Graphic input device]

Number 10 is a figure input device, and when a figure is drawn with a pen on the screen where the figure is to be drawn, an appropriate sampling interval, for example 250
In synchronization with the synchronization signal every 1LS, the digital values of the X and Y coordinate signals representing the position of the pen on the screen are
The signals are obtained on output lines 11 and 12, respectively, and a pen up/down signal 2 is outputted on output line 13.

In this specification, for convenience, the number assigned to a signal line and the number assigned to a signal flowing through that signal line are assumed to be the same and will not be distinguished.

The pen up/down signal 2 referred to here is when the pen is pressed down on the screen and operates with pressure above the -travel (this is referred to as "1 down 1").
This is a signal representing two states: a state where the pen is released from the screen (this is called ++up++).

As an example of such a graphic input device, for example, the
No. 3-7649 T, 0. Mr. ELLIS and M.R.

It is described in detail in ``Digital Computers and Graphic Input Systems'' by Mr. DAVIS.

[Feature extractor]

Reference numeral 20 inputs the output of the graphic input device 10, extracts graphical features of characters, and outputs a feature signal 200.

Such a feature extractor is described in ``Apparatus for automatically reading cursive handwriting'' by Debbie Harmon, Leon, published in Japanese Patent Publication No. 1924/1983.

[Area detector]

Reference numeral 30 denotes an area detector which receives the output of the graphic input device 10 as an input and outputs an area number 300 from its x and y coordinates.

Referring to FIG. 2, the detailed configuration of this area detector 30 is shown.

Numbers 310 and 315 refer to the graphic input device 1, respectively.
When two signals rise at the X and Y coordinates of 0, that is, "
This is a register that stores when the signal goes from "LOW" to "HIGH".

Numbers 330, 331, 332, and 333 are the upper limit of the X coordinate, the lower limit of the X coordinate, the upper limit of the y coordinate, and the lower limit of the y coordinate of the area corresponding to the character field on the screen of the graphic input device 10. It is a read-only memory (hereinafter abbreviated as ROM) that is stored in locations corresponding to area numbers.

The X coordinate stored in register 310 is input to comparators 320 and 321, and the y coordinate stored in register 315 is input to comparators 325 and 326.

Further, the other inputs of the comparators 320 and 321 contain the upper limit of the X coordinate of the area corresponding to the area number in synchronization with the address (area number itself) generated by the oscillator 360 and the counter 370 that receives the oscillator 360 as an input. ROM33
When the values of the ROM 331 containing 0 and the lower limit are respectively input, and the X coordinate stored in the register 310 is a value located between the upper limit and the lower limit, the comparators 320 and 32
1 output to AND circuit 340.

Similarly, regarding the y-coordinate, the value stored in the register 315 is stored in the ROM 332, which contains the upper limit of the y-coordinate of the area corresponding to the area number, and the ROM 33, which contains the lower limit.
2, the comparator 325
and 326 are both AND circuit 340's four inputs are 1"
When the signal reaches HIGH"12, the address output by the counter 370, that is, the area number, is stored in the register 350, and the area number is output to the signal line 300.

Here, for example, the oscillator 360 may be an i o OKH2 oscillator, and if the X and Y coordinates from the graphic input device 10 are 10-bit outputs, the registers 310 and 3
15 is a 4-bit register μPB21 manufactured by NEC.
75 (product name) can be configured with 3 pieces each, with 32 comparators
0, 321, 325 and 326 are NEC 4
It can be configured with three bit comparators μPB2085 (product name).

Further, if the area number corresponding to the character field of the document outputted by this area detector 30 is 0 to 255, then the counter 37
0 is NEC's 4-bit binary counter μPB216
1 (product name), and the ROMs 330, 331, 332, and 333 can each be configured by three 4×256-bit bipolar ROM6300 (product name) manufactured by NMT.

In addition, the register 350 that stores the area number has the area number O.
255, it can be configured with two 4-pit resistors P]32175 (trade name) manufactured by NEC.

In this embodiment, the X and Y coordinates are loaded into the registers 310 and 315 at the rising edge of the Z signal, but a configuration in which the loading is carried out a certain time after the falling or rising edge is also possible.

[Format checker]

The number 40 is the area number 300 output by the area detector 30.
This is a ROM that takes an address as an input and outputs the format stored at that address to a signal line 400.

As mentioned above, if the format signal 400 is 4 bits, the area number 300 output from the area detector 30 is 8 bits.
Product name) Can be configured with one piece.

[Identification memory]

Referring again to FIG. 1, numeral 50 is an identification memory that inputs the feature signal 200 from the feature extractor 20 and searches for identification logic only for a specified format using the format signal 400 from the format checker 40. be.

Referring to FIG. 3, details of this identification memory 50 are shown.

Numbers 52, 53, 54, and 55 are ROMs that store identification logic in numbers, alphabets, katakana, kanji, and formats, respectively, and each inputs the feature signal 200 output from the feature extractor 20 and extracts categories in the form of a truth table. Check the presence or absence of separately memorized features. Signal line 5
It is output to 00.

Also, these ROMs 52, 53, 54, and 55 are THRE
E5TATE ROM, the output of the decoder 51 that decodes the format signal 400 output from the format checker 40 is taken into each data select terminal, and a ROM is selected for each format and the ROM is
The category stored in is output to the signal line 500.

Here, the decoder 51 is a decoder μPB2154 manufactured by NEC.
(Product name) If the THREE STATE ROMs 52, 53, 54, and 55 are both 8-bit configurations for the feature signal 200 and category signal 500, then the MMI 4X256-bit ROM630
1 (Product name) Can be configured with two pieces at a time.

In this example, the format corresponding to a certain area was explained as a type of character, but the format is not limited to characters, but also symbols on maps, benzene nuclei and element symbols that appear in chemical structural formulas, integrals and differentials in mathematics, vectors, etc. It is also possible to configure the information by functional fields, such as signals and basic mask patterns that appear in LSI design.

Further, the real-time character recognition device according to the present invention, excluding the graphic input device, can also be realized by firmware.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a real-time character recognition device according to the present invention, in which 10 is a graphic input device, 20 is a feature extractor, 30 is a region detector, 40 is a format checker, and 50 is an identification memory. FIG. 2 is a diagram showing details of the area detector 30, 310
and 315 and 350 are registers, 320, 321 and 325
and 326 are comparators, 330, 331, 332 and 3
33 is a ROM, 360 is an oscillator, 370 is a counter, 3
40 is an AND circuit. FIG. 3 is a block diagram showing details of the identification memory 50, in which 51 is a decoder, and 52, 53, 54, and 55 are ROMs.

Claims (1)

[Claims] 1. A real-time character recognition device that recognizes characters according to format, characterized by having each of the following components (a) and (b) ← When a figure is drawn with a pen-shaped writing instrument, signals of the X and Y coordinates representing the position of the pen-shaped writing instrument on the screen observed at appropriate sampling intervals and the push of the pen-shaped writing instrument on the screen are generated. (b) A graphic input device that outputs a signal 2 indicating whether or not the graphic input device is in a closed state; , an area detector that detects which area on this screen it belongs to, ← U a format checker that takes the output of the area detector as input and confirms the formant to be drawn in that area, (Y) the above figure input a feature extractor that takes the output of the device as input and extracts the features of characters and graphics drawn on the screen of the graphic input device; ((e) a memory that stores identification logic for each format; An identification memory that receives the output as input and performs identification using identification logic selected according to the format output from the format verifier.