The Construction And Optimization Of A Personalized Font Library Based On Chinese Character Strokes And Structure Information

With the rapid development of the Internet,people are more and more interested in personalized customized products,and personalized fonts emerge as the times require.They are widely used in computers,newspapers,books,magazines,textbooks,documents,package and communication software.In addition,personalized fonts also have their unique value in the field of education and teaching.Through design and creation,personalized fonts have their own characteristics.Individualized fonts can not only optimize education and teaching,make teaching activities interesting and enhance students'interest in reading,but also cultivate students' innovative and divergent thinking and improve their creative ability.At the same time,the generation of personalized fonts can also support educational application,which has application value for the personality of teachers and students in online community interaction.The font library can break the current situation of using the font "thousands of people on one side",and use the font designed by ourselves to communicate online,chat,send micro-blog,send e-mail,"see the word as the face" and "see the word as the person",which is undoubtedly a process full of personality and creativity.This paper presents a method and optimization of Chinese character library construction based on strokes and structural information.Firstly,the calligrapher writes every word in block letters on the digital input device and assists with a certain post-processing method,and finally generates 6763 words in CB2312 standard,we called it standard template of font library.According to the standard Chinese character template,the writer writes 51 words on the touch screen device,calculates the writing characteristics by collecting the skeleton point set of writing,and records the information of the time sequence point set when the writer writes Chinese characters,including the x,y coordinates and writing time of each point in each stroke track point set.Geometric method is used to calculate the writers'writing characteristics,and a feature library template is established,which includes two levels:the feature of components and the feature of words.Then,the standard template is modified according to the writers' writing characteristics to form a feature template.After that,the structure of input word set is divided into four categories:single character,surround structure,upper and lower structure,left and right structure.In the feature template,the user input handwritten Chinese character set is classified and identified according to its structure.The SVM and KNN classification methods of machine learning are used to validate and analyze the style classification of structural data.The test results show that the discrimination rate is about 80%,It shows that this structural classification method can distinguish the writers corresponding to different handwritten Chinese characters.Then,it establishes and improves the situation of strokes and components,increases the diversity of strokes and makes the effect of strokes more natural,and makes the combination effect closer to your writing style.Finally,the skeleton point set of the writer is mapped to the feature template of the font to generate a personalized style font.Through the analysis of the subjective judgment experiments,it is found that whether the original handwritten characters are beautiful or ugly,the optimized words generated by structural classification are closer to the handwritten style of the writers than the original ones,but also more beautiful and natural,and when the original handwritten characters are more beautiful,the optimized spelling effect is better than that of the original handwritten characters;through the Turing experiment,30 testers can find that the optimal spelling effect is better than that of the original handwritten words.There are 694 handwriters correctly selected,accounting for 46.27%.That is to say,the confusion probability of the generated words is 53.73%.This shows that the effect of optimizing the combination has reached the degree of confusion of handwriting,of course,the details can be further optimized.