Potent Features For Identification Of Chinese Character Strokes

Chinese characters contain three levels of visual information: stroke,radical,and structure.Stroke is the basic level of character recognition.Some studies have shown that strokes can enhance the semantic and orthographic representation in the acquisition of Chinese character.After character recognition becomes expertise,strokes still play a role in facilitating recognition which may be more dependent on specific strokes.This shows that recognition of Chinese characters may be based on stroke-based feature processing.Recent studies show that the recognition of letters and words is based on termination or the end point and the common features(intersection point and midsegment)similar to object recognition.But in character recognition,it is not clear what the feature is.Does it share similar end point or common features? We hypothesize that there may be general and culture-specific features in different writing systems.In order to verify this hypothesis,the thesis will focus on two questions: first,what are the key features in the stroke of Chinese characters;moreover,whether the key features and general features in the strokes affect the recognition of Chinese characters.In study 1,we used Bubbles to randomly sample different stroke information and search for the visual information that is important in the stroke recognition of Chinese characters.We selected 29 strokes as visual stimuli and created 14 features that all strokes contain,such as horizontal,vertical and hook.The results show that dot,horizontal hook and right hook are important features in stroke recognition.In the spatial frequency information,we analyze six spatial frequency bands of stroke stimuli,which represent the degree from coarse to fine picture of stroke stimuli.The analysis found that,unlike English letter recognition,stroke recognition was more dependent on the background spatial frequency information of stroke stimuli,focusing on the frequency band of 4-8 cycles/image.This indicates that stroke recognition has a bias to the information of low spatial frequency,which may be related to the writing order of strokes in Chinese characters.The purpose of study 2 is to examine the role of key features found in the stroke recognition.It has been shown that there are common features in word and object recognition,including terminations,vertex and midsegment.In this study,we use delayed segment technique to exam the role of key features of strokes and general features across different objects on Chinese character recognition.Experiment 1 investigates whether the turning and hook features found in research 1 affect Chinese character recognition.Experiment 1 selected 45 high frequency Chinese characters and asked the subjects to perform naming tasks.In experiment 1,it is found that only removing the turning and hook in the word will not have significant influence on the word recognition.There are two potential reasons to explain the results of experiment 1.We did not consider the area of pixel removal when removing features,and the damage of Chinese characters on different pixel areas has an impact on the results.Besides,it is possible that the line relation represented by the hook and turning will have a great influence on Chinese character recognition.In order to confirm these two speculations,in experiment we examine the characteristics of line relationships in Chinese characters,such as terminations,vertex and midsegment,and considers the degree of damage to these characteristics.Experiment 2 found significant main effects at damage degree and feature type.The principle of feature extraction in experiment 2is based on line units,which may impair integrity of strokes in Chinese characters.This may cause different effects on features.To verify this,the experiment 3examed the characteristics of the terminations,vertex and midsegment of stroke-based units.The results show a significant main effect at damage degree,and the reaction times of character recognition is slowed down when 35% and 55% features of characters were eliminated.Among the three features that defined based stroke unit(vertex,midsegment,termination),all the features show their importance in Chinese character recognition.Overall,it can be seen that the key features of stroke recognition(vertical turning,horizontal hook,right hook and dot)and key features of Chinese character recognition(vertex and midsegment and the terminations)conform to the neuronal recycling hypothesis.According to the neuronal recycling hypothesis,human's ability to learn from culture depends on the retrieval of preexisting brain circuits,and the key features in word recognition may be derived from our key features in object processing.The hook is considered to be a special terminations in Chinese characters,and it often appears at the end of a character.The current results suggest that hook play a key role in stroke and character recognition.