Parallel BFS graph traversal on images using structured grid
Bor-Yiing Su, Tasneem G. Brutch, Kurt Keutzer
Abstract
Bor-Yiing Su, Tasneem G. Brutch, Kurt Keutzer
Abstract
Graph algorithms are widely used in image processing techniques. With technology advancements, image sizes are increasing, and the contents inside images are becoming more complex, resulting in increased runtimes for graph algorithms on these images. Breadth First Search (BFS) is a fundamental graph traversal approach. A key to parallelizing graph algorithms used in image processing is to parallelize the BFS graph traversal operation. In this paper, we propose using highly parallelizable structured grid computations to realize the BFS graph traversal operations. This mapping enables efficient implementation of the BFS graph traversal operations on highly parallel manycore platforms. By using such a mapping, we were able to achieve performance gains of 2× to 33× depending on image complexity.
OpenAlex reports 8 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Graph algorithms are widely used in image processing techniques. With technology advancements, image sizes are increasing, and the contents inside images are becoming more complex, resulting in increased runtimes for graph algorithms on these images. Breadth First Search (BFS) is a fundamental graph traversal approach. A key to parallelizing graph algorithms used in image processing is to parallelize the BFS graph traversal operation. In this paper, we propose using highly parallelizable structured grid computations to realize the BFS graph traversal operations. This mapping enables efficient implementation of the BFS graph traversal operations on highly parallel manycore platforms. By using such a mapping, we were able to achieve performance gains of 2× to 33× depending on image complexity.
Key concepts: Graph traversal, Tree traversal, Computer science, Graph, Breadth-first search, Parallelizable manifold, Grid, Graph database