Transloadit

What is Region Growing?

Region growing is an image segmentation method that starts from one or more seed pixels. It repeatedly adds connected neighbors whose color, intensity, texture, or other measured properties satisfy a similarity rule.

How Region Growing works

Region growing evaluates a local neighborhood around accepted pixels, so membership decisions propagate outward from each seed rather than being made independently across the frame. Connectivity is usually defined on a pixel grid, while the homogeneity test may use single values or statistics updated as the region expands. Its spatial continuity differs from clustering methods that can group visually similar but disconnected pixels. In an imaging pipeline, it commonly produces a mask for measurement, extraction, or later refinement.

Key facts

  1. Four-neighbor connectivity considers only horizontal and vertical adjacency, whereas eight-neighbor connectivity also admits diagonals and can join regions across corner contacts.
  2. Updating a region’s mean or variance after each accepted pixel makes the criterion adaptive, but the resulting mask can depend on seed order when several regions compete.
  3. Noise and gradual intensity gradients can create narrow leakage paths; smoothing, boundary constraints, or postprocessing such as hole filling are common safeguards.

When Region Growing matters

Choose region growing when the target forms a connected area and representative seeds can be selected reliably. Poor seeds or permissive similarity thresholds can merge distinct regions or allow leakage across weak boundaries.

Turn media knowledge into a working pipeline

Connect uploads, processing, AI, storage, and delivery through one declarative API — with the encoding stack, scaling, and format churn handled for you.

Try Transloadit for free