This repository contains the implementation of the segmentation algorithm as described in: Efficient Graph-Based Image Segmentation by Pedro F. Felzenszwalb and Daniel P. Huttenlocher, published in International Journal of Computer Vision, 59(2) September 2004\n\nThe program accepts a color image in PPM format and generates a segmentation with a distinct color assigned to each region.\n\nTo compile "segment", type "make".\n\nExecute the program by running "segment sigma k min input output".\n\nThe parameters explained:\n\nsigma: Used to smoothen the input image before segmentation.\nk: Value for the threshold function.\nmin: Minimum component size imposed by post-processing.\ninput: Input image.\noutput: Output image.\n\nTypical parameters are sigma = 0.5, k = 500, min = 20. Bigger values for k yield larger components in the result.
dreamlens-superpixelsegmentation-solution's People