This block computes the quick shift segmentation of all database images.
BK = BLOCK_QUICKSEG() Initializes the block with the default options.
BK = BLOCK_QUICKSEG(BK) Executes the block with options and inputs BK.
Required Inputs:
- db
The database of images to extract quick shift superpixels on.
Options:
- bk.ratio
The ratio between spatial consistency and color consistency. See VL_QUICKSEG(), parameter ratio.
- bk.sigma
The standard deviation of the parzen window density estimator. See VL_QUICKSEG(), parameter kernelsize.
- bk.tau
The maximum distance between nodes in the quick shift tree. See VL_QUICKSEG(), parameter maxdist.
- bk.ref_size
Resize to ref_size before performing the segmentation. Segmentation maps are resized to match the original image upon completion.
Fetchable attributes:
- segs
A structure representing the segmentation. Required parameter: seg_id. For each superpixel s in the requested image, the format is: segs(s).ind An linear index to all pixels which form the
superpixel.
segs(s).count The size of the superpixel in pixels. segs(s).color The mean color of the superpixel. segs(s).adj The neighbors of this superpixel.
- segimage
The color segmentation image produced by VL_QUICKSEG(). Required parameter: seg_id.
- segmap
The label map produced by VL_QUICKSEG. Required parameter seg_id. If two outputs are provided, the second output is the labels. If three outputs are provided, the third output is the size of each superpixel.
- labels
A vector of the labels in label map. Required parameter: seg_id
- time
The amount of time taken for a particular image. Required parameter: seg_id.