The present paper is devoted to address the use and evaluation of measures coming from digital images in an industrial context. Starting from a discussion about the importance and role of this class of measurements, the authors show the architecture of an image-based measurement system and introduce the important issue of the related uncertainty then coming to investigate models and methods for evaluating it. After presenting an original method for evaluating the uncertainty of images deriving measurements, some examples of practical interest are introduced and evaluated leading the reader to the understanding of related problems. The paper is also enriched by the presence of an introductive section discussing a bibliography of the state of the art in the field.