ABSTRACT

The application of machine vision technology involves dealing with many visual variables. In most cases one must be forgiving of some variables and sensitive to others, the system must detect. Even simple applications have to contend with variables that a person performing the same task can easily dismiss. Label pres­ ence and/or absence detection is often cited as a machine vision application. One variable a system must be able to contend with is hue saturation. For example, a label may be perfectly acceptable as long as its "color" is yellow, from pastel to virtually orange. A person can easily accept this range and still detect a missing or skewed label. A machine vision system may be tricked into thinking the label is missing when actually the acceptable hue saturation change is what was experi­ enced.