SLICSuperPixels

Compute super pixels based on SLIC based on “SLIC Superpixels Compared to State-of-the-art Superpixel Methods” (TPAMI 2012). jsk_perception use implementation of https://github.com/PSMM/SLIC-Superpixels, https://github.com/berak/SLIC-Superpixels and https://github.com/garaemon/SLIC-SuperPixels.

Output of this node is an image and each value means label index.

Subscribing Topic

  • image (sensor_msgs/Image)

    Input image.

Publishing Topic

  • ~output (sensor_msgs/Image)

    Output image. The encoding of image is CV_32SC1 and each element value means label index.

  • ~debug (sensor_msgs/Image)

    Debug image, each border of cluster drawn by red contour. (optional, publish if ~publish_debug_images is true)

  • ~debug/mean_color (sensor_msgs/Image)

    Debug image, each cluster is drawn by mean color of the cluster. (optional, publish if ~publish_debug_images is true)

  • ~debug/center_grid (sensor_msgs/Image)

    Debug image, Center of each cluster is plotted by red dot. (optional, publish if ~publish_debug_images is true)

Parameters

  • ~number_of_super_pixels (Integer, default: 100)

    The number of super pixels.

  • ~weight (Integer, default: 4)

    Weight of metrics between color and pixel distance.

  • ~publish_debug_images (Bool, default: false)

    Publish debug images (~debug, ~debug/mean_color, ~debug/centr_grid)

Sample

roslaunch jsk_perception sample_slic_super_pixel.launch