The work aims to research a new number identification method for the invoice printing online inspection system in order to ensure that the invoice containing number printing error does not flow into the society. An invoice number identification method based on structural features was proposed. Firstly, a series of pre-processing for the invoice image was realized, such as collection, graying, binarization, denoising, tilt correction, character positioning, single character segmentation, normalization and so on. A number identification classifier based on structural features was established, and then the number was classified according to the structural characteristic value of each number in the invoice. Based on the experimental results and by means of the proposed structural characteristic method, the invoice number identification rate reached above 99%. After a large number of actual invoice number identification tests, the proposed method has strong anti-interference ability, and the recognition algorithm is fast and the accuracy is high.