CFETS Information Technology (Shanghai) Co. ShangHai , China.

---

Abstract—With the rapid development of the financial industry, the realization of automatic recognition of bond information is a hot issue in the current financial science and technology research. This paper realizes the automatic recognition of bond information through the combination of OCR and NLP technology. The text information of bond is extracted from the picture by OCR technology. In view of various kinds of interference in the image, the difference subtraction operation based on HSV model of color image is adopted to get better image information for image-based sequence model. The image-based sequence model is composed of convolution neural network and recursive neural network. The bond information can be well recognized by the model. Finally, the text information is processed by the algorithm based on Levenshtein distance, and the key-value is extracted, which has a certain practical value.

Index Terms—Image processing, OCR; graph sequence model, natural language processing.

[PDF]

Cite: Jizhe Dai, Zhengyan Ma, "Automatic Identification of Bond Information Based on OCR and NLP," Journal of Computers vol. 14, no. 6, pp. 397-403, 2019.