Connected Component Labeling Algorithms in Image Processing

Connected Component Labeling (CCL) assigns unique identifiers to discrete regions within binary or segmented images by scanning and resolving provisional labels according to defined connectivity (for example, four- or eight-neighbourhood in two dimensions, or six-, eighteen- or twenty-six-neighbourhood in three dimensions). Classical two-pass methods perform a raster scan to assign provisional labels, record label equivalences via data structures such as union-find, and then propagate final labels in a second pass. Block-based strategies group pixels or voxels into processing units to minimise neighbourhood checks and memory accesses. Modern adaptations have leveraged graphics processing units and field-programmable gate arrays to meet real-time constraints in applications spanning medical diagnostics, remote sensing and industrial quality control. Extensions to three-dimensional datasets and ultrahigh-definition video streams further demonstrate the global significance of CCL in extracting meaningful topological and morphological information from increasingly large and complex image volumes.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

A volumetric labelling algorithm for three-dimensional binary images employs two distinct scanning masks that adapt to voxel neighbourhoods, reducing redundant checks and improving efficiency over conventional methods. A parallel CCL algorithm for graphics processing units harnesses CUDA-based tree merging to extract and aggregate component features on the device, dramatically lowering data transfer volume to the host while sustaining real-time performance. A hardware-oriented Union-Retire CCA architecture discards directed tree structures in favour of connectivity-focused processing, cutting memory requirements by over one third and streamlining end-of-row operations on field-programmable gate arrays for embedded vision systems.

Connected Component Labeling Algorithms in Image Processing publication trend

The graph below shows the total number of articles in connected component labeling algorithms in image processing across all publications each year (not limited to Nature Index journals).

Technical terms

Connected Component Labeling: The process of scanning a binary or segmented image to assign a unique label to each set of pixels or voxels that are mutually connected according to a chosen connectivity criterion.

Raster Scan: A systematic, row-by-row traversal of image pixels or voxels for the purpose of initial label assignment in two-pass labelling methods.

Union-Find Algorithm: A data structure that maintains disjoint sets of labels and supports efficient merging and lookup operations to resolve provisional label equivalences.

Voxel: A volumetric pixel representing the smallest distinguishable box-shaped part of a three-dimensional image.

References

1. An Efficient Connected-Component Labeling Algorithm for 3-D Binary Images. IEEE Open Journal of the Computer Society (2022).
2. Block-Based Connected-Component Labeling Algorithm Using Binary Decision Trees. Sensors (2015).
3. Real-Time FPGA Implementation of Parallel Connected Component Labelling for a 4K Video Stream. Journal of Signal Processing Systems (2021).
4. Union-Retire for Connected Components Analysis on FPGA. Journal of Imaging (2022).
5. Parallel Algorithm for Connected-Component Analysis Using CUDA. Algorithms (2023).