Parameters
Input image (default = "") Input binary image from which objects will be identified. This image must be 8-bit and only contain values 0 and 255.
Output objects (default = "") Name of output objects to be stored in workspace.
Black objects/white background (default = "true") When selected, "foreground" pixels are considered to have intensities of 0 and background 255 (i.e. black objects on a white background). When not selected, the inverse is true.
Identify as single object (default = "false") Add all pixels to a single output object. Enabling this skips the connected-components step.
Connectivity (default = "26") When performing connected components labelling, the connectivity determines which neighbouring pixels are considered to be in contact.
- "6" considers immediate neighbours to lie in the cardinal directions (i.e. left, right, in-front, behind, above and below). In 2D this is actually 4-way connectivity.
- - "26" (default) considers neighbours to include the cardinal directions as well as diagonal to the pixel in question. In 2D this is actually 8-way connectivity.
Volume type (default = "Pointlist") The method used to store pixel coordinates. This only affects performance and memory usage, there is no difference in results obtained using difference storage methods.
- "Pointlist" (default) stores object coordinates as a list of XYZ coordinates. This is most efficient for small objects, very thin objects or objects with lots of holes.
- "Octree" stores objects in an octree format. Here, the coordinate space is broken down into cubes of different sizes, each of which is marked as foreground (i.e. an object) or background. Octrees are most efficient when there are lots of large cubic regions of the same label, as the space can be represented by larger (and thus fewer) cubes. This is best used when there are large, completely solid objects. If z-axis sampling is much larger than xy-axis sampling, it's typically best to opt for the quadtree method.
- "Quadtree" stores objects in a quadtree format. Here, each Z-plane of the object is broken down into squares of different sizes, each of which is marked as foreground (i.e. an object) or background. Quadtrees are most efficient when there are lots of large square regions of the same label, as the space can be represented by larger (and thus fewer) squares. This is best used when there are large, completely solid objects.
Enable multithreading (default = "true") Break the image down into strips, each one processed on a separate CPU thread. The overhead required to do this means it's best for large multi-core CPUs, but should be left disabled for small images or on CPUs with few cores.
Minimum strip width (px) (default = "60") Minimum width of each strip to be processed on a separate CPU thread. Measured in pixel units.