GPU constructed image segmentation using first order edge detection operators in CUDA environment

Sanjay Saxena, Neeraj Sharma a

Abstract

This paper presents the command of parallel computing on the GPU in contrast to the substantial computations required in processing of the images. The GPU has long been utilized to quicken 3D applications. With the appearance of anomalous state programmable interfaces, programming to the GPU is simplified and is being utilized to quicken a more extensive class of utilizations. Image segmentation has emerged as an imperative stage in image based applications so our main focus in this work to provide speedup in image segmentation using first order edges detection operators. As it plays a very vital role in various image based application. With the frequency of high level Application Programming Interfaces (CUDA - Compute Unified Device Architecture), the supremacy of the GPU is being leveraged to hasten more common purpose and high performance applications. More precisely, this paper emphasis on CUDA as per its parallel programming platform for first order edges detection operators. Standards are done amongst its parallel execution and its sequential execution.