A network of globally coupled chaotic maps for adaptive multi-resolution image segmentation

In this paper, a computational model for image segmentation based on a network of coupled chaotic maps is proposed. Time evolutions of chaotic maps that correspond to a pixel class are synchronized with one another, while this synchronized evolution is desynchronized with respect to time evolution of chaotic maps corresponding to other pixel classes in the same data set. The model presents the following advantages in comparison to conventional pixel classification techniques: 1) the segmentation process is intrinsically parallel; 2) the number of pixel classes can be previous unknown; 3) the model offers a multi-resolution and multi-thresholding segmentation approach; 4) the adaptive pixel moving process makes the model robust to classify ambiguous pixels; and 5) the model obtains good performance and transparent dynamics by utilizing one-dimensional chaotic maps instead of complex neurons as individual elements.