Energy Efficient Filter using an Approximate Multiplier for Fault Identification in Machine Vision

Abstract

Machine Vision systems represent an inclusive area of image processing with wide range of benefits in industrial automation, robotics and so on. As Machine vision systems are trending to be more commercial and complicated, enormous mathematical means are required. Often the system must deal with the early stages of image processing whenever an imaging system acquires an image. In this brief, an energy efficient approximate multiplier based filter is proposed that can be employed for such image processing. The approximate multiplier trades off computational accuracy with energy consumption. Windowing original operands of multiplier with weighted bits and then executing the multiplication exclusively for those windows stands as the principal approach for approximation. The proposed approximate multiplier based filter is employed both as a low pass filter and as a high pass filter in fault identification process for different sample images such as welding chip, ceramic tile and textile using LabVIEW software. The proposed filter proves that it can preserve the quality besides being energy efficient.