KR20170086228A

KR20170086228A - Real Time Handwriting Converting System

Info

Publication number: KR20170086228A
Application number: KR1020160005800A
Authority: KR
Inventors: 성하규; 김시빈; 김민식; 윤혁중
Original assignee: 성하규; 김민식; 윤혁중; 김시빈
Priority date: 2016-01-18
Filing date: 2016-01-18
Publication date: 2017-07-26

Abstract

The present invention is a system capable of converting handwritten text into a text file in real time. This is to solve the problem that the error is very big according to how to shoot because the accuracy of the existing technology is low and the long time until the best conclusion is obtained.
The user can use the camera to scan the character on the screen. The user can then select the best interpreted result by changing the position of the camera on the handwriting. Then convert it to a text file.
According to the present invention, the analytical result according to a large number of positions and conditions can be checked in real time, and the user can directly select the best-interpreted result, thereby greatly reducing the time until the best result is obtained. In addition, the result obtained is converted into a text file so that a person directly corrects the wrong part and makes it usable for many purposes.

Description

{Real Time Handwriting Converting System}

The present invention relates to a system capable of converting a handwriting into a text file in real time, more specifically, analyzing a handwriting in real time and displaying it on a screen so that a user can select a most accurate analysis result by viewing the camera from various angles and positions, And converting it into a text file.

There have been a considerable number of methods and devices for recognizing handwritten characters, which is a technique for recognizing handwritten characters by characters. There are many ways to collect and analyze input data of users and to use programming.

However, existing technologies have various limitations. First, since the handwriting is very different from person to person, the error will be very large regardless of how well the above techniques are used. Therefore, when the handwriting is recognized by the camera, the result interpreted by the camera angle, the amount of light received, and so on are very different.

Also, existing technologies simply have to recognize the input handwriting individually. That is, it takes a considerable amount of time for the user to obtain the most accurate interpretation of the handwriting since the handwriting that requires interpretation is discretely recognized only for each photograph.

As described above, the present invention aims at solving a problem with a very large error due to low accuracy of existing techniques and a long time until a best result is obtained.

According to an aspect of the present invention, there is provided a system for handwritten real-time text conversion, comprising: a camera for continuously inputting and recognizing visual information such as a handwritten text desired to be interpreted, a handwritten sentence desired to be interpreted, An autofocus function for focusing when a similarity degree of a consecutive frame is equal to or greater than a set value when stopping the handwriting sentence, paragraph or the like desired to be interpreted at the center of the screen, and a handwritten sentence, A handwritten character recognition method capable of recognizing a handwritten character composed of Korean, English and numerals and converting the handwritten character into a text format, receiving an input of a paragraph, etc., an output device And a display unit If you do, provide real-time handwritten text conversion system comprising: a tool to save text and save it as a file in text format.

The handwritten document is preferably a handwritten character characterized in that characters do not overlap each other and are written in different colors from each other and colors distinguishable from the background color.

The degree of similarity is based on a ratio of a common portion occupied in the entire frame as a result of comparison between two consecutive frames.

As described above, according to the present invention, it is possible to check analytical results according to a large number of positions and conditions in real time, so that the user can directly select the best-interpreted result, . In addition, the result obtained is converted into a text file so that a person directly corrects the wrong part and makes it usable for many purposes.

1 is a flowchart of a system for receiving a handwritten document in real time and converting and storing the text.
2 is an algorithm of a system for receiving a handwritten document in real time and converting and storing the text.

The present invention will now be described in detail with reference to the accompanying drawings.

1 is a configuration diagram of a handwritten real time text conversion system according to the present invention.

As shown in the figure, the overall system includes a camera, a central processor S100, and an output device S110, and performs functions of an autofocus function, a handwritten character recognition function, and a text storage tool S120 in a central processor.

2 is an algorithm of a handwritten real time text conversion system according to the present invention.

As shown in the figure, when the program is executed, the camera is immediately executed and the handwritten document is recognized by the camera (S200). If the similarity of the consecutive frames is equal to or greater than the set value (S210) The results are displayed on the screen. When the user presses an OK button on the result (S220), the character recognition result is converted into a text file and stored.

This program is a program based on the camera (S200), and when it is executed, the camera function is turned on immediately. Therefore, the camera shooting screen appears immediately. We then move such a camera so that sentences, paragraphs, etc. that want to be interpreted are located in the center of the camera screen.

At this point, we will focus on the auto focus function (S210) on the existing camera if the program remains still without moving the screen to scan the sentence or paragraph. Such an auto-focus function is designed to perform focus detection and character recognition if the similarity of consecutive frames is equal to or greater than a set value, or to continue recognition again.

The program then scans the sentences and paragraphs and analyzes them using various character recognition techniques (S100). There are many techniques such as "feature extraction method for quick offline handwritten character recognition", "handwritten character recognition method and device", "handwritten alphabet and number recognition method" and the like. If you use some of these techniques, you will be able to analyze handwritten characters. In addition, using these techniques online, users can improve the accuracy of such handwriting by correcting the mistakes in the handwriting analysis results and bringing them online.

We display the character recognition result analyzed in the output device (S110) using the program. At this time, if the user likes the result of the translation and the error is small, it is determined if it is satisfactory, and if it is not satisfactory, the user will continue to rotate the camera and repeat the same operation continuously to obtain a better recognition result.

When the user determines (S220), the recognition result is converted into a text file by the program, and the result is stored as a file through the text storage tool (S120) so that the user can correct the typographical errors due to the character recognition error, You can use it.

While the preferred embodiments of the present invention have been described in detail with reference to the accompanying drawings, it is to be understood that the invention is not limited to the disclosed embodiments, and various changes and modifications may be made without departing from the scope of the present invention. Accordingly, the scope of the present invention should be defined by the following claims.

S100: It is a central processor that handles many functions such as handwriting recognition and text storage.
S110: An output device which is a screen for displaying a character recognition result.
S120: A text storage tool that saves the text conversion result as a file.
S200: Receives the time information of the camera in real time.
S210: If the similarity of the consecutive frames is 90% or more, the document at the center of the screen is regarded as the target and character recognition proceeds. If there is a large change in the screen, it is considered that the target is still being searched and the time information is inputted continuously.
S220: The user presses the OK button means that the user is satisfied with the result and wants to convert the text file.

Claims

The system converts the handwriting into a text file by analyzing the handwriting in real time and displaying the result on the screen by selecting the most accurate analysis result by illuminating the camera at various angles and positions,
The handwritten text that you want to interpret,
A camera for continuously inputting and recognizing visual information such as a handwritten sentence or paragraph desired to be interpreted,
An autofocus function for focusing when a similarity degree of a consecutive frame is equal to or greater than a set value when stopping the handwritten sentence, paragraph or the like desired to be interpreted to come to the center of the screen,
A handwritten character recognition method capable of recognizing a handwritten character composed of Korean, English and numerals and converting it into a text format by receiving handwritten sentences, paragraphs and the like on a focused screen by the auto focus function;
An output device for allowing the user to view the text converted by the handwritten character recognition method on the screen,
If you want to find and check the text you want, you can use text storage tool
And converting the handwritten text into a handwritten text.

The method of claim 1,
Letters that do not overlap each other,
Wherein the text is written in a color separable from the background color.

2. The method of claim 1,
As a result of comparing two consecutive frames, a handwritten real - time text conversion system based on the ratio of common parts in the entire frame.