Research Of Topic Using Cell Phone

With the rapid development of technology and the enrichment of network resources,online education is playing a more and more important role in our daily life.Users take photo of the confused questions and the system will recognize and retrieve the text information in photos.Then it can find the most like questions from question bank and provide right answer and solving approaches to users.What's more,it can also find out a group of questions for you to exercise.It follows that whether it can recognize characters accurately has a big effect on user experience.What's more,the algorithms of text area recognition and subject category recognition are the key point of text recognition from pictures.If we can improve algorithm accuracy of question classification,we can move further towards text recognition from pictures.Towards this end,this paper will do some research and practice into algorithm of text area location and classification,designation of system and related application of this subject.Specific subject as follows,Firstly,in regard to the designation and realization of text area algorithm,the article applies the method of SWT(Stroke Width Transform)to locate text area of questions in pictures taken by phone.By doing this,subsequent operations can focus on text area and the accuracy of algorithm can be approved because of the reduction of division and recognition.Secondly,this article extracts features with multi-features method and analyzes the gray level,shape,line and other visual features roundly.Then,it divides text area into math,Chinese and English type with the help of Multi-class Classification Support Vector Machine based on Normal decision tree.Thirdly,as for the urgent demand of text recognition from pictures,this article designed and realized the system to run text recognition and database retrieval of questions.Two algorithm are realized in my paper.Text detection algorithm performance is as follows:the Word Recall rate is 79.03%,and the Stroke Precision is 79.58%,the Pixel Precision is 90.39%.The precision of classification algorithm for text area is 93.32%.