Book Image Distortion Correction Based on Bezier Curve Model
Furong Zou, Shengnan Tang, Xiaowen Li, Changyuan Yu · 2022 IEEE 10th Joint International Information Technology and Artificial Intelligence Conference (ITAIC) · 2022
With the development of information technology, image acquisition is becoming an important means to obtain information. Taking pictures of books, using OCR technology can quickly obtain the information of books, but in practical application, the book page will be distorted, which will affect the accuracy of information collection. In view of this situation, on the basis of comparing the correction methods of Bezier curve and machine learning, this paper focuses on the correction effect of text image using Bezier curve under different distortion degrees, and uses the rigid features of the page to transfer the correction model to the text illustrations on the same page, and finally sends the corrected text into OCR recognition, Comparing the recognition accuracy before and after correction, through the result analysis, using Bezier curve bending correction has better correction effect and efficiency without increasing hardware resources.