Parallelizing Canny Edge Detection
Item
Poster Number
35
Poster Title
Parallelizing Canny Edge Detection
First Presenter
Isaiah Fisher
Other Contributors
Mark Trovinger, Omer Yurdabakan
Abstract
Canny Edge Detection is one of the most common methods of edge detection in modern image processing. This algorithm has been implemented in image processing frameworks such as OpenCV. Traditionally, Canny Edge Detection is implemented using a serial approach, however past research has shown it is possible to parallelize this algorithm. In theory parallelizing this algorithm should increase performance with respect to large datasets. In this work, we will compare the performance of a parallel implementation of Canny edge detection to an existing serial implementation. Different types of parallelization will be explored, and the results will be compared on datasets of varying size. An attempt will be made at a solution utilizing graphics processing units (GPUs) using the CUDA platform.
Year
2022
Embargo
no embargo
