Product Recognistion System For Blind Persons Using Camera

Abstract

The method deals by means of product identification for the visually impaired, to help them to read text labels from packaged items in their daily routine. To separate the item from cluttered backgrounds in the camera sight, foremost a well-organized and efficient movement based technique to be implemented to identify a region of interest (ROI), ROI is selected by asking the person to shake the item in camera sight. This technique extracts moving item region by a Gaussians-based background subtraction method. In the extracted ROI, text localization and identification are conducted to obtain text information. Further using the background subtraction method, the foreground is single-handedly extracted and the left over parts are neglected. After extracting the foreground, edge detection algorithm is used to find the borders of items in ROI and those borders are improved using edge grow up algorithm. For extracting the label text from the specified input image Optical Character Recognition (OCR) algorithm is used, the algorithm extracts the text whatever font style of text present on the label and the complex background. Once the given item name is extracted, the output is given in the form of voice using android mobile and is also displayed using LCD by interfacing both through a microcontroller. The execution part is done via MATLAB software