Text Detection and Recognition for Semantic Mapping in Indoor Navigation

Abstract

Detection and recognition of nameplates is used to increase the map of the robot built with locations of objects. This is very useful in service robots applications where many tasks will be the type to wear reading. The most common in the mobile robot service scenario where the problem is moving Android mobile autonomous in the domestic environment, and builds a map as it moves along, with headquarters at it, predictable the objects in their way and place them in the map. In the research arena, utilizing the text has been favorite as a landmark in navigations and until now, different types of algorithms are being processed for indoor surroundings for initially detection and then recognition of objects like texts. The main objective of this project is to use a vision system in an autonomous robot to detect and identify signboards in an inside surrounding in such a way that it is able to recognize the desired nameplate in a given environment. Binarization the output image is brought to the recognition of OCR characters. a database of possible sets of signs that are created in the laboratory. From there, the algorithm more like comparing the strings match. We recorded the names of different signs that are in the native setting and target was accomplished of 80% accurateness in text recognition in an image. Robot Operating System is used for the simulations and a novel technique for detection of texts from the indoor signboards is developed