当前位置:文档之家› A Method for Handwriting Input and Correction on Smartphones

A Method for Handwriting Input and Correction on Smartphones

A Method for Handwriting Input and Correction on Smartphones
A Method for Handwriting Input and Correction on Smartphones

A METHOD FOR HANDWRITING INPUT AND CORRECTION ON

SMARTPHONES

GARETH LOUDON, OLLE PELLIJEFF, LI ZHONG-WEI

Cyberlab Singapore, Ericsson Research,

#18-00 SLF Building, 510 Thomson Road, Singapore 298135 gareth.loudon@https://www.doczj.com/doc/5c14736928.html,, olle.pellijeff@https://www.doczj.com/doc/5c14736928.html,, zhong-wei.li@https://www.doczj.com/doc/5c14736928.html,

This paper describes a new approach for entering text on a “Smartphone” via natural

handwriting input. The approach focuses on ease of use within the confines of a Smartphone

display size and processing limitations. Therefore there are two integrated components to the

approach. The first is a new handwriting recognition engine that has been designed to have a

very high recognition accuracy (98.3% character accuracy), support sentence-based

handwriting input, and have a small memory footprint (84 kb) and fast processing time. The

second is a method that allows very simple editing and correction of recognized characters.

1 Introduction

Mobile devices that are available today are having more and more functionality built into them. There are new devices called “Smartphones” that can support voice communication, PIM tasks, as well Internet access for e-mail and browsing. These Smartphones support text input via a stylus. One example of such a Smartphone is the Ericsson R380 shown in Figure 1. Although a software keyboard can be used to enter text, a more convenient method is to use handwriting.

Figure 1. The Ericsson R380 Smartphone.

Some requirements on handwriting recognition engines for Smartphones are that they must run on slow processors with a low memory footprint, have very high recognition accuracy, enable fast text entry and allow easy correction. Several handwriting recognition engines have been developed to meet these requirements [1][2][3]. Limitations on some of these handwriting recognizers are that users have to write one character at a time and wait for recognition to take place. This waiting time is sometimes known as a “time-out”. One way around the problem of time-outs is to make the user write every character with one stroke. This is done in the Graffiti recognition engine used in the Palm devices. This enables very fast text entry speed. The system also has high accuracy because of the unique strokes representing each character. The disadvantage is that users have to learn the special way to write the

481

In: L.R.B. Schomaker and L.G. Vuurpijl (Eds.), Proceedings of the Seventh International Workshop on Frontiers in Handwriting Recognition, September 11-13 2000, Amsterdam, ISBN 90-76942-01-3, Nijmegen: International Unipen Foundation, pp 481-485

characters. To overcome the problems that exist with other handwriting recognizers we allowed users to write a phrase at a time. To achieve very high recognition accuracy we placed some restrictions on the handwriting styles. Users are only allowed to write in a hand printed style; they must write in lowercase letters and convert to uppercase during an editing phase; and they must write the characters in a particular way as shown in Figure 2. Accents are supported to allow entry of European specific characters. Other characters supported are common punctuation, symbols and also gestures for editing.

Figure 2. Writing styles of characters supported.

2 Methods

The structure for the handwriting recognition architecture is shown in Figure 3.

Figure 3. Handwriting recognition architecture.

The technique used in the handwriting recognition engine is based on discrete Hidden Markov Models (HMMs) [4]. HMMs are used because they handle the problems of non-linear shifting and multiple representations of handwritten characters well. There are two major modes to the handwriting recognition system. The first mode is the training of the HMMs, and the second mode is the recognition. Each handwritten character sequence is stored as a sequence of x,y coordinates in time. The next sections describe the training and recognition methods.

482

2.1 Training

Only handwritten examples of isolated characters are used for training. The following sections describe the training process.

2.1.1 Character Pre-processing

There are five steps in the pre-processing stage.

Character Normalization: Find the mean and variance of a handwritten character, and scale the character to have zero mean and standard variance. Characters are not distorted in shape during the normalization.

Stroke Joining: For characters written in two strokes, connect strokes together to form a one-stroke character. Fill in points between the first and second strokes. The number of points to fill in depends on the distance between the last point in the first stroke and the first point in the second stroke.

Stroke Re-sampling: Resample the pre-processed characters so that the average speed of writing a character is constant across all characters, while maintaining the unique acceleration and deceleration information within a stroke. This is important because some characters are written slowly and some characters are written quickly, but all examples of the same character tend to slow down at corners of a stroke and speed up for straight lines.

Feature extraction: Once the handwriting data has been re-sampled, four features are used to describe the character for each data point (x i , y i , x i - x i-1 , y i - y i-1). If i<2 x i - x i-1 , y i - y i-1 are set to zero.

Vector Quantization: The final step of the pre-processing is the quantization of the features into a fixed set of quantization codes. This helps to reduce the computation complexity in the HMMs. Vector Quantization (VQ) is a standard procedure that can be found described in several textbooks [4]. In the training stage of the process, two codebooks are formed; one 32 element codebook for the x i, y i features and one 32 codebook for the x i -x i-1, y i - y i-1. All handwriting training examples are then quantized using these two codebooks in preparation for the HMM training.

2.1.2 HMM Training

The discrete HMMs are trained using the standard Baum-Welch algorithm [4][5]. Two ten-state HMMs are used for each character. One HMM is based on the features x i, y i and one HMM is based on the features x i -x i-1 , y i - y i-1.

483

2.2 Sentence Recognition

During the recognition stage, handwritten phrases are input into the handwriting recognizer. An example of handwritten text is shown in Figure 4 below. The recognition stage is split into four main steps. The first is the segmentation of handwritten sentence into isolated characters, the second is the extraction of features for character recognition, the third is the search algorithm to find the most likely character, and the fourth is the re-scoring of the recognition results.

Figure 4. An example of a handwritten sentence for recognition.

Character segmentation: Several parameters are calculated for each stroke during the character segmentation step. The key parameters are shown in figure 5.

Figure 5. Some parameters for the handwritten phrase “on the”.

Parameter “x” describes the width of a stroke, parameter “s” describes the space between adjacent stroke, parameter “y” describes the difference in height between the centres of adjacent strokes and parameter “c” describes the distance between the centre of a stroke and the right-hand edge of the preceding stroke. If (s i > max(x i)) or (-s i > 2* max(x i) and y i > max(x i)) stroke i is a character at the end of a word, else if ( c i > 0) stroke i is a character within a word, else stroke i should be merged with the following stroke to form a character. Each character segmented from the input sentence is passed onto the next steps for character recognition.

Character Pre-processing: The pre-processing step during recognition is the same as the training stage, with the only difference being that the codebooks are re-used from the training stage during the VQ process.

Viterbi Search: The time sequence of VQ codes for a character are matched against all the HMMs using a standard Viterbi search algorithm [4]. The probabilities of match between the input data and all characters are recorded and sorted. The five characters with the highest matching probability are stored.

484

Re-scoring: Once the segmentation and character recognition steps have been completed, a re-scoring step takes place that makes use of simple language knowledge to improve recognition performance. The first part of the re-scoring step uses a simple statistical language model based on character unigram and bigram probability values. The second part of the re-scoring step uses a word dictionary, in this case 16,000 words.

2.3 Preliminary Recognition Results

A test set of handwritten English sentences containing 5559 characters (including punctuation and numerals) were used to evaluate the recognition performance. The test set contained handwriting samples collected from seven different writers. All characters were written in a style that conformed with the requirements of the recognizer. The handwriting training set contained 11984 different characters written by thirteen different writers. The average character recognition accuracy was 98.3% for the 5559 characters when not including the use of a dictionary.

3 Editing and Correction of Recognized Characters

The most common approach for correction in recognition systems in the past has been to present a list of alternative characters or words for a user to choose from. If the correct character or word is not in the list then the user has to write the character or word again in some writing area that is separate from the recognized text. The process is also complicated if the user wants to edit a passage of text, because again the user has to switch between a text-editing mode and a handwriting-input mode.

The approach we have taken is to integrate the text editing and correction task with the handwriting input task. We created a mode of correction that is similar to the correction mode used when modifying text on paper with a pen. The user just has to write on top of an existing character to change it. The user can also delete words by crossing them out or inserting new letters or words using a combination of gestures and character writing. Figure 6 gives an example of the correction mechanism. The handwriting recognizer keeps track of re-written characters so that the same recognized character does not keep appearing after recognition.

Figure 6. The correction mechanism.

485

Accents can be added in the correction/editing stage, as shown above, so that specific European characters can be created. To change the case of a character from lower to uppercase (or vice-versa) simply tap on a button (that acts as a shift key) and then tap on the character you wish to change.

4 Implementation

The handwriting recognition engine was coded in “C” and compiled on the EPOC OS used by Ericsson’s Smartphones. The memory size of the basic handwriting recognition engine is 84kb (excluding the re-scoring module). If the dictionary is added the overall memory is 216kb.

5 Conclusions

The approach we have taken for handwriting input has been tailored to meet the needs of Smartphones and other mobile devices of today. The approach not only includes a new handwriting recognition engine but also introduces a new mode of editing and correction that is a very simple and easy to use. It is planned to undertake a subjective evaluation of our handwriting recognition engine as compared with the Graffiti recognition system by involving a large group of users. This is to study the advantages and disadvantages of both methods.

References

1. CalliGrapher by Paragraph, https://www.doczj.com/doc/5c14736928.html,.

2. smARTwriter by ART, https://www.doczj.com/doc/5c14736928.html,.

3. JOT by CIC, https://www.doczj.com/doc/5c14736928.html,.

4. L Rabiner and B,H Juang, Fundamentals of Speech Recognition, Prentice Hall,

1993.

5. L,E Baum, “Inequality and Associated Maximization Technique in Statistical

Estimation of Probabilistic Functions of Markov processes”, Inequalities, vol.

3, 1972, pp.1-8.

486

英文手写体书法练习册123-numbers-writing-workbook

Now I know my numbers! My Very Own Number Writing Practice Workbook 1, 2, 3… One, two, three…

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 one one one one one one

2 2 2 2 2 2 2 2 2 2 2 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 two two two two two two

3 3 3 3 3 3 3 3 3 3 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 three three three three

4 4 4 4 4 4 4 4 4 4 4 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 four four four four four

5 5 5 5 5 5 5 5 5 5 5 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 five five five five five

Handwriting is important

ron.f.hill@https://www.doczj.com/doc/5c14736928.html, 在电脑发展如此迅速的时代,你的书写是否还很重要呢? Handwriting is important I’m speaking for the motion. As a high school student, handwriting is very important. I want to make 3 points of it. In the first place, handwriting shows how much you put your heart in what you write. Whether you can write it beautifully or not does show how much effort you have made. You may say that some people just can’t write well, but in my opinion,it’s the attitude that counts. Once you take it seriously, the handwriting will turn out to be good. In the second place, handwriting is often a symbol of our character. When taking an exam or looking for

a job, handwriting is just like your appearance. It will leave a first impression on others. It’s universally recognized that first impression can be very critical. Just imagine. If you see two exam papers in front of you, which one will you prefer? The one with a nice handwriting, or the one that looks like a mess? The answer is quite obvious. In the third place, handwriting is a picture of our marvelous culture and history. It indicates the fascinating tradition that has passed down from generation to generation. How can you abandon it? The technology may change rapidly, but the culture will last forever. To sum up, I think it essential for students to have good handwriting in all respects. (231words)

handwriting 美国的“颜骨柳肌”,英语手写体字帖

Trace and Write Letter A D’Nealian Style Handwriting Practice Name ________________________________ Trace and write the letter A. A/A/A a/a/a Color the artichoke green. Copy the sentence below. An/artichoke/is/a/ vegetable.///////// ///////////////////// /////////////////////

Trace and Write Letter B D’Nealian Style Handwriting Practice Name ________________________________ Trace and write the letter B. B/B /B b/b /b Color the bat brown. Copy the sentence below. The/boy/hit/the/ball with/his/new/bat./////////////////////////////////////////////

C/C/C c/c/c Color the camel. Copy the sentence below. The/camel/likes/to// count/carrots./////// ///////////////////// /////////////////////

雅思口语Part1范文:Handwriting书法话题

Do you often write things? Do you write everyday? What do you usually write? Do you like writing to people? How often do you send e-mails? What are your main reasons for using e-mails? Do you like to send e-mails? Do you usually write by hand or write using a computer? Nowadays, how do most people write things? Do you think computers might one day replace handwriting? When do children begin to write in your country? How did you learn to write? Do you think handwriting is very important (nowadays)? How can children today improve (or, practice) their handwriting? What impression does a person’s handwriting have on other people? Yes, I write documentaries as part of my job on a daily basis for my clients. There are many things I have to write, for example, I usually write end user training for my clients to teach them how to use the system, or I write a development guide for my technical staffs to teach them how to code. 是的,作为工作的一部分,我每天都为我的客户写纪录片。我需要写很多东西,例如,我通常为我的客户编写终端用户培训来教他们如何使用系统,或者为我的技术人员编写开发指南来教他们如何编写代码。 Yes, I’m really into communicating with my clients by email. I send emails to them on a daily basis from my office, discussing business, guiding them how to use the system and solving their problems. I like sending emails because I can trace the communication in email software. It is much more useful than the telephone. 是的,我非常喜欢通过电子邮件与客户沟通。我每天从办公室给他们发电子邮件,讨论业务,指导他们如何使用这个系统,解决他们的问题。我喜欢发电子邮件,因为我可以用电子邮件软件跟踪沟通。它比电话有用得多。

Three benefits of handwriting 手写的三个好处

Three benefits of handwriting Nowadays, many problems can be solved by using computer or mobile phone particularly along the young people. Fewer and fewer people will use the handwriting. Obviously, people do not realize that handwriting is really helpful for us. In my opinion, handwriting has three main adventures. First, handwriting can exercise the brain. Scientists believed that the brain can become more active by handwriting. Also, People’s interests will be inspired. Second, handwriting can improve people’s writing ability. Many famous writers have the habit of writing by hands. They think it benefits them very much. Last but not least, handwriting can improve the learning efficiency. Studies have shown that when people write an article by handwriting, they pay more attention to the writing. It can help you improve concentration, because there are too many things which may divert their attention from writing. In conclusion, we really need to attaches great importance to the handwriting. We also need to practice our handwriting.

高中英语练字

高一英语练字(一) A seven-year-old girl named Anaya Ellick, who was born without hands, beat fifty competitors to win a national award for handwriting, which makes her parents proud and inspires many people. This story sets me thinking profoundly. Life can place a stumbling block in our way and definitely, it is always easier to give up than to hold on. However, like Anaya, constant effort yields sure success. If you pursue your goals with the courage of never giving up, you will succeed in whatever endeavor you take on. As high school students, we are bound to encounter the twists and turns life offers us. But when we think of those people who suffer more than us, we should have a more optimistic view on study and draw lessons from the bumps and turns along the way. Bear it in mind that we should keep going forward, even in the toughest time.

英语练字-handwriting

English Handwriting 1)书写的规格 (1)应按照字母的笔顺和字母在三格中应占的位置书写。 (2)每个字母都应稍向右倾斜,约为5°,斜度要一致。 (3)大写字母都应一样高,占上面两格,但不顶第一线。 (4)小写字母a,c,e,m,n,o,r,s,u,v,w,x,z写在中间的一格里,上下抵线,但都不出格。 (5)小写字母b,d,h,k,l的上端顶第一线,占上面两格。 (6)小写字母i和j的点、f和t的上端都在第一格中间,f和t的第二笔紧贴在第二线下。

(7)小写字母f g j p q y的下端抵第四线。 (8)小写字母a,d,h,i,k,l,m,n,t和u,它们的提笔是一个上挑的小圆钩,不能写成锐角。 (9)书写单词时,字母与字母之间的间隔要均匀、适当,不要凑得过紧,也不要离得太远。 (10)书写句子时,单词之间必须有适当的距离,一般以空出一个小写字母a 的宽度为宜。

______ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________ ______________ ______________ ________________ _______________

英语练字

____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________ ____________________________________________________________________________________________________

雅思口语范文Part 1 Topic 29 Handwriting 书法/书写

1. Do you like handwriting? I really like writing with a pen. Actually most of my friends just like texting or typing. But from my perspective, I think it’s very nice and fulfilling when I write with a pen and paper. It’s more personal that way. 2. Is your handwriting easy to read for other people? Yeah, I guess my handwriting is easy for other people to read. Actually that’s a skill that I’ve practiced for many years. I think it’s important to take your time and be careful when you write, or else no one will understand your writing. 3. Do you like to receive a handwritten letter / card? I honestly love getting letters or cards with hand-written words. You know, it’s really sweet and it adds a personal touch. Since handwriting takes longer than texting or even emailing, it shows that the person really cares about you. It definitely makes me happy to get something like that. 4. Can you tell someone’s personality from his or her handwriting? Yes, I think you can tell a little bit about someone’s personality from their handwriting. For instance, if someone has messy handwriting, they might be a careless person. But in contrast, people with neat handwriting are usually patient. Of course, this kind of judgement isn’t always accurate. 5. Do you like writing by hand or typing? I prefer writing by hand over typing. Even though typing is fast and convenient on a computer or a cellphone, a pen and paper help me concentrate more on what I want to write. I feel like I’m always being distracted when I write on a computer because I always spend hours browsing the internet.

英语练字2

A new house _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ Yang Ling and her parents live in a new house now. It is _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ n e a r h e r s c h o o l.T h e y l i k e t h e n e w h o u s e v e r y m u c h. _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ I live in a new house now.Is it big? Yes,it is.There are three bedrooms, _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ two bathrooms,a study, a kitchen,a dining-room and a large sitting-room. _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ Is there a garden?No.there isn't a garden. What's in your bedroom? _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ There's a desk beside my bed.On the bed,there are some dolls. Are _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ there any pictures on the wall? No.there aren't any pictures. There's _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ a map of the world on the wall.Is there a telephone? Yes.It's _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ on the desk. b . _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ _____________________________________________________________________________________ 1

初中英语优秀作文字帖

The Hero in My Heart Zhong Nanshan, an 80-year-old scientist ,is a super hero in my heart. We were afraid so much because COVID-19 spread the whole country.Fortunately,there were millions of heroes making efforts to protect us.One of them is Zhong Nanshan whom I respect most.When virus apeared,he went to Wuhan,the most dangerous place,to guide medical work .It was his responsiblity and courage that moved me a lot. The super hero in my heart is Zhong Nanshan ,who is a great man. My Dream My dream is to be a scientist like Zhong Nanshan. Zhong Nanshan, an 80-year-old scientist ,went to

How Can We Become Good learners Everyone is born with the abiliy to learn.How can we become good learners? On the one hand,we should create an interest in what s easier to we learn.Once we ' re interested in something,it ' pay attention to it.On the other hand, “practice makes perfect. W”e should practice more and learn from failures. Learning is lifelong.We should learn wisely and learn well so that we can become successful learners. The Mistake Made Me Grow up The mistake made me grow up. I still remember how my English improved.When I first learned English,I always kept silent to avoid making mistakes. Fortunately,when Miss lee knew about that ,she encouraged me to answer the questions.Luckily,whenever I made mistakes and lost heart ,Miss Lee told me to correct them. “ Practice makes perfect. ” I became a good learner. Don ' t be afraid of making mistakes because the mistakes make us grow up.

相关主题
相关文档 最新文档