An Image-based Bookstore Inventory Check System

Abstract

To check the inventory in a bookstore, it always takes a lot of manpower and time to count the books one by one on the bookshelves. The high cost results in longer inventory check period, such that the bookstore managers can not know the accurate stock information anytime. This paper proposes an image-based approach for bookstore management. A number of web cameras are uniformly distributed around the bookstore, and then the bookshelves’ images are captured in inventory check. A finite state machine is applied to segment the images into individual book spine images, in which the SIFT features are extracted to search for the identities of the books on the shelves. After that, we can get the information about the books which are still on the book shelves. The database is updated immediately according to inventory check results to reflect the latest stock information in time. The experiment results show that the book spine recognition rate is above 90%, which proves the feasibility of the proposed image-based bookstore inventory check system. Most errors come from the book series because the distinguishable features in them are fewer.