Wireless sensor networks for image recognition

Abstract

With the expanding of wireless sensor networks (WSNs) application, WSNs are not only designed to measure slow changing scalar values and fast changing dynamic signals such as temperature and vibration but also used to recognize more complex information like image. The image recognition system based on WSNs is a low cost and convenient alternative method for traditional wire image recognition system. This paper firstly reviews the current state of the art of WSNs for image recognition and its application in fire detection, smart metering device, medical care, and target tracking. The main research challenges and potential solutions in this field are then discussed. Optimized image recognition algorithm for WSNs nodes and low- power FPGA based WSNs nodes will be the effective solutions. Finally, a reading recognition system for the traditional rolling digital water meter is proposed and verified. The experimental results indicate that the recognition rate of the presented system reaches 97.4% for the complete number and 83.2% for the rolling number.