Research On Feature Extraction Of Tangut Script Recognition And Classifier Design

Character recognition is a traditional subject in the field of machine recognition,many researches have been obtained,the recognition of Chinese characters and ancient script is an important research subject in the field of Chinese information processing.The research results of machine recognition have been commercialized,and widely used in face recognition,fingerprint recognition,license plate recognition,office automation and financial business affairs.There are many difficulties in character recognition,because of the Chinese character is very important in the practical application,also has the great significance in theory research,so many studies still study this subject diligently.The recognition of Tangut script is a new field to be developed currently,according to the study,there are many difficulties in the study of the recognition of Tangut script.First,the quantity of Tangut has more than 6000,so it's belonging to the large character set;Second,compared with Chinese,the structure is more complex,the stroke is more cumbersome,and most of the number of strokes are above 14,so Tangut is a high similarity character set.Third,the handwritten Tangut script has different sizes and dot matrix,which increases the complexity and difficulty of the recognition of Tangut script.The most important work of the digitization ancient characters is the identification of ancient script,and the feature extraction is the foundation of character recognition.So mainly introduce the algorithm and process of Tangut script feature extraction in this article.Firstly,introduce the significance of the research on the recognition of Tangut and the present research status at home and abroad;then introduce the preprocessing of Tangut script images,including normalize,binary,smoothing,thinning and tilt correction:and then adopts the Haar-like algorithm and Gabor wavelet algorithm to extract the feature of Tangut script separately,finally uses the AdaBoost algorithm to study the characteristics of the extracted classifier,and compares the classification results between use a single Haar-like algorithm and use Haar-like and Gabor wavelet algorithm cascade way to extract feature,and achieved better classifition results.