A Simple Character Recognition Algorithm on the Image Based on FPGA

As we know, the characters recognition uses neural network algorithms typically. However, calculation of neural network algorithms is complex. This paper proposes a new threading algorithm which is suitable for FPGA implementation for the character recognition on an image. The algorithm can identify the characters on an image quickly and accurately. The algorithm selects six lines of image data located the region on the image where the character is located after the image is binarized, and counts the number of “1”s therein, then compares with the parameters gotten by software to obtain the determination result. The accuracy of this algorithm is close to 100%. The algorithm is written on Quartus II software (Version 15.0) of Intel company and implemented on Terasic DE1_SOC. It has the advantages of high reliability and high speed through board level verification and signaltap verification.