Text code recognition and its definition

A recognition system for text code and its application to an embedded machine is presented. The system consists of encoding, decoding and recognition modules. During the encoding and decoding processes, a novel encoding and a mapping mode are proposed based on the theories of Reed-Solomon. During the recognition process, three new approaches are proposed. The first is a partitioning and sliding binarization method based on sampling threshold, which will keep the anchor points. The second consists of circle dot filtering and connected region judge, which will find the anchor points. The third is an improved projection method for character segmentation. By recognizing a variety of samples from different mobile phones, the experimental results show that our proposed approaches are effective. The average correct recognition rate of the system is 95%. And combined with RS method, the correct recognition rate is 100%. Experiments demonstrate that the system also meets speed requirement.