Variabilidad estadística en imágenes digitalizadas de rastros de uso: resultados preliminares

Abstract

Work, as physical action, modifies matter. This is the basic principle of use-wear analysis. Changes on matter by work give society those products needed for social reproduction: consumption goods. Lithic tools are part of these. We can identify the tools through use-wear analysis, in order to infer the social action. Moreover, we should think about how to observe change of states in matter. We think we can solve it using images. We define the image as a representation arising from Reality. In this way, any image is a model of reality. We recognise objects through images and, in this process, we define those aspects of the same objects which we consider relevant. This is a good tool for our work, but there are still some problems arising from aspects related to the inherent subjectivity of the observation process. Digital images are a possible solution, because their characteristics allow us to quantify them. We can develop a more complex quantification from the numeric values of pixels, in order to be able to analyse statistically differences between image textures. We have quantified the description of use-wear features in order to develop knowledge that allows us to classify use alterations and determine the original use of the tool. This process is composed of several parts: - Selection of the areas modified by use in all images, that is the selection of pixels, which have grey values related to the use alteration. Those areas are called texels. - We select some features to describe the texels that we will quantify. The selection of features is made from our previous knowledge of use wear analysis. An objective characterisation of texels is generated using geometric variables, which will allow discriminatory measures between different use-related textures. - Finally, a statistical analysis is used to classify the set of images into their corresponding uses. In this paper we explain the development of this methodology using a small sample of images with alterations caused by two working matters –dry wood and dry fur– and two movements –cutting and scraping. Interconnecting both factors gives us four uses. We also explain the preliminary statistical results that we have obtained so far.