Reference point estimation technique for direct validation of subpixel point detection algorithms for Internet of Things

Abstract

Subpixel point detection algorithms are important in many application spaces, especially those where limitations of the imaging device's resolution need to be overcome. Such algorithms help decrease the overall requirements of the given system. Many factors, such as power consumption and cost, are critical in the context of the Internet of Things. While these algorithms do offer an improvement in the precision of point detection, it is often difficult to directly determine their precision. The main reason for it is the lack of the point of reference that the outputs of subpixel point detection methods can be compared to. In this work, we present a novel method for finding the point of reference for validating the subpixel point detection algorithms directly. Its operation is demonstrated on an experimentally obtained sample dataset.