点击次数：

论文类型：期刊论文

发表时间：2013-01-01

发表刊物：Information Technology Journal

收录刊物：EI、Scopus

卷号：12

期号：20

页面范围：5440-5446

ISSN号：18125638

摘要：Image segmentation is an important issue in the field of computer vision, it serves as a bridge linking the basic image processing methods to the high-level semantic recognition methods. With the increasing applications of the image segmentation methods in the modern industries, such as the defect detection in the production lines, the real-time requirements are greatly raised. Recently, with the advent of the General-purpose Graphics Processing Unit (GPGPU) platform, the parallelized implementations on this new platform open a new way to accelerate the image segmentation methods to meet the real-time requirements. In this work, various methods are analyzed and parallelized on the GPGPU platform, the horizontal comparisons are made to evaluate the potentials of parallelization for different segmentation methods. The parallelization strategies are performed on two levels: on the algorithm development level and on the program development level. It is expected that this investigation may provide guidance to the future parallelization tasks for the more advanced image segmentation methods and other computer vision applications. ? 2013 Asian Network for Scientific Information.