Texture Measuring by Means of Perceptually-Based Fineness Functions

Abstract

Abstract. Although texture is a widely analyzed property in computer vision, models which relate this feature with human perception are needed. In this paper, a model that associates computational fineness measures with the human perception of this type of texture property is proposed. The fineness perception is collected from human subjects, performing an aggregation of their assessments. Considering a wide variety of measures, a coarseness-fineness function relating these features with the human assesments is obtained as the function which provides the best fit of the collected data.