Research on Information Recognition of VAT Invoice Based on Computer Vision

With the promotion of the bill exchange system throughout the world, the use of VAT invoices has exploded. In order to solve the problems of low efficiency, high error rate and high labor intensity of manual entry of electronic system of VAT invoice, a method of recognizing VAT invoice information based on computer vision was proposed. Firstly, the invoice image was preprocessed, and the tilt correction was implemented by local adaptive threshold and Hough transform. Then the key information area was segmented and the target object was taken out by projection method. Finally, the characters were recognized by OCR technology. The results show that the method has low error rate and high efficiency, and the average character recognition accuracy can reach 96.21%.