Research on Resistor Value Identification Using Image Recognition

—Method for Reducing the Effect of Specular Highlights Superimposed on Color-Code Image—

Abstract

Specular highlight, which is a direct reflection component of illumination light, is known to be an obstacle when reading color-codes of resistors using image recognition. We propose a method for avoiding the influence of specular highlight through image processing. We have set plural strip-shaped areas on color-codes and extract the color by selecting areas of smaller influence of reflection. We have implemented an algorithm of reading color-codes and conducted reading experiments of resistors having various angle of rotation, which have changing areas of specular highlights. As a result, we have confirmed an identification rate of 99%. We also discuss the changes of the influence of specular highlights due to the rotation of the resistors.