Aplicativo móvel de análise de textura baseado em fractais

Abstract

This work proposes to develop a mobile application to classify texture images, based on fractal theory. The proposal is to develop an Android mobile application that receives a trained model, captures images of a texture, extracts the features using a recent texture analysis method based on the fractal descriptors called FDMIC and then classify it. The trained classification model is generated through a desktop software, developed in Java, which receives input parameters such as an image base, the scale value and the size of the sub-image to be cut. This proposal aims to unite the performance of FDMIC to the mobility provided by a smartphone. Finally, after experimental tests, the accuracy of FDMIC in this proposed solution was measured and through statistical analyzes, the results were compared, emphasizing strengths and weaknesses of this approach.