Text recognition and correction for automated data collection in participatory sensing applications

Participatory sensing is an approach that enables mobile devices such as cell phones to be used for data collection, analysis and sharing processes by individuals. In this context, automatic data collection becomes important. In this study, a method is proposed for automatic recognition of information in photographs taken by participants´ cell phones. The proposed method is used to obtain information such as product names and prices from store receipt images. For this purpose, information in store receipts is recognized by image processing methods using Optical Character Recognition (OCR). Then, Knowledge Based Correction (KBC) algorithms in conjunction with the participatory sensing approach are used to correct inaccurate information which cannot be corrected by image processing methods. The proposed method achieves 90% word recognition accuracy and 97% character recognition accuracy.