Book Spine Recognition Based on OpenCV and Tesseract
Lina Cao, Mengdi Liu, Zhuqing Dong, Hua Yang · 2019
In a traditional library, the inventory management of books is a time-consuming and laborious work. Due to such constraint as cost and technical complexity, some technologies, e.g., RFID, are infeasible to be used in such large-scale application scenarios as book inventory in library. We design a prototype system called Book Spine Recognition Based on OpenCV and Tesseract, or BSRBOT for short, which is based on image processing mechanism, and can be further integrated into existing book inventory management systems. Firstly, according to the characteristics of book arrangement on bookshelf, a book spine image segmentation algorithm is designed. Secondly, the recognition and retrieval of the index number at the lower part of the spine is carried out based on character recognition. Finally, each segmented spine image can be matched with the image database to overcome the occasional unsuccessful identification of the call numbers. The experiment shows BSRBOT achieves accurate recognition of the book spines, thus providing a solution for fast, feasible whilst low-cost book inventory management in modern library.