Automatic Recognition of Indoor Digital Instrument Reading for Inspection Robot of Power Substation

Abstract

In this paper, a method of digital meter reading recognition algorithm was proposed. Our algorithm is based on the template matching pattern recognition method combined with deep learning algorithm with considering the characteristics of substation indoor environment. First, block higher order statistics and threshold via local maximal between-cluster variance are used to locate the digital region on the dashboard, then used the deep learning algorithm with template matching method to compare the digital region with chosen teacher image template, and finally determine the value of digital region. This method is simple and efficient that can be effectively applied in indoor inspection tasks for automatically inspection robot of power substation.