Skip to main content
. 2018 May 2;8:6875. doi: 10.1038/s41598-018-25261-7

Figure 3.

Figure 3

Algorithm for computing the texture in a pixel’s neighborhood. (ac) Generating the response of each pixel to a Leung-Malik filter bank. (d) K-means clustering of response vectors, generated from all cores in the training set, in order to find 50 cluster centroids or textons. (e) Histogram of textons, within a pixel’s neighborhood, comprise the texture-related feature vector T for each pixel.