Study On Character Recognition Algorithm Based On The Image Of Passport And ID Card

Passport and identity card are the key documents to prove the identity of the holder.Certification needs are required at airports,customs offices,train stations,Banks,etc.At present,we judge the authenticity by the consistency of the information of the artificial verification chip and the surface information of the certificates.With the development of optical character recognition technology,we have tried to apply it to the identification of evidence surface information.But the surface of the certificates have a complex texture and security identification.Moreover,the number of Chinese characters on the identity card is numerous,and there are many close characters and rare characters.Therefore,it is very difficult to apply it directly to engineering.First,aiming at the existence of initial image edge distortion,too much black edge,security features,etc.,this article uses a series of pretreatments to get a clear binary image.Secondly,for characters in different regions,this article uses a regional segmentation algorithm based on the connectivity domain to split the id number and the passport machine readable area,and the improved multi-threshold projection method is used to complete the segmentation of the comprehensive information area of the identity card.At this point we get the single character image of the corresponding region.Thirdly,aiming at the complex characters of Chinese characters,this paper proposes an improved multi-level classification algorithm to identify the characters.We first use the feature distance similarity method to implement the coarse classification.Then we propose the two fine classification algorithms based on improved SURF algorithm matching and internal and external contour feature classification based on improved SPM method in order to complete the identification of the identity card information area characters.Finally,we use the multi-feature fusion direct classification method to complete the English letter and digital recognition of identity card number and the passport machine readable area.Experimental results show that the algorithm proposed in this article effectively reduces the interference of security identification in preprocessing,segments the characters in the target area more efficiently,which is compared with the conventional algorithm.And the character recognition accuracy is higher,the calculation speed is faster,and the anti-interference ability is stronger.The all solution works effectively.