OCR based interaction for coded objects

As a trend, our everyday goods are changing to program-controlled ones from pure mechanical or electrical ones. The concept of `coded object´ was proposed for such a new generation product by Kitchin and Dooge. We propose a novel interaction method using OCR for coded objects, which is aimed at simplifying the operations on the configuration of coded objects and making the recording, sharing, and distribution of the configurations universal. We constructed a prototype for proof-of-concept using a digital SLR (single lens reflex camera). In this prototype, we utilize open-source OCR engine “tesseract” trained by handwritten character images and adding a preprocessing of projective transforms in order to cope with free angle capture by the camera. This prototype can be configured by shooting handwritten texts.