Study of parallel image processing with the implementation of vHGW algorithm using CUDA on NVIDIA'S GPU framework

Abstract

This paper provides an effective study of the implementation of parallel image processing techniques using CUDA on NVIDIA GPU framework. It also discusses about the major requirements of parallelism in medical image processing techniques. Additional important aspect of this paper is to develop vHGW(van Herk/Gill-Werman morphology) algorithm intended for erosion and dilation proposed for diverse types of structuring elements of random length and along with random angle parallely on NVIDIA's GPU GeForce GTX 860M. The main motive behind implementing image morphological operations is its importance for extracting components of an image. That can be beneficial in the demonstration and explanation of shape of the region. These experiments have been implemented on CUDA 5.0 architecture with NVIDIA's GPU GeForce GTX 860M and got significant results in terms of time.