A Survey on Parallel Image Processing Studies Using CUDA Platform in GPU Programming

被引：0

作者：

Aydin, Semra ^{[1
]}

Samet, Refik ^{[2
]}

Bay, Omer Faruk ^{[3
]}

机构：

[1] Gazi Univ, Bilisim Enstitusu, Ankara, Turkey

[2] Ankara Univ, Bilgisayar Muhendisligi Bolumu, Muhendislik Fak, Ankara, Turkey

[3] Gazi Univ, Elekt Elekt Muhendisligi Bolumu, Teknol Fak, Ankara, Turkey

来源：

JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI | 2020年 / 23卷 / 03期

关键词：

Image processing; parallel computing; GPU; CUDA; RECONSTRUCTION ALGORITHMS; SEGMENTATION ALGORITHM; PERFORMANCE EVALUATION; IMPLEMENTATION; REGISTRATION; ACCELERATION; MRI; EFFICIENCY; SPACE; MODEL;

D O I：

暂无

中图分类号：

T [工业技术];

学科分类号：

08 ;

摘要：

Image processing is used in a variety of fields. Image processing techniques need high processor performance due to increased image resolution day by day. Parallel processing techniques are used to satisfy the requirements related to high performance in real time image processing applications. Recently, GPU programming is one of the most commonly used and preferred methods in parallel processing. CUDA is the most popular platform in GPU programming. In this survey the studies where CUDA platform was used for image processing are presented and evaluated. The major purpose of this survey is to provide a comprehensive reference source for the starters or researchers involved in use of CUDA platform in GPU programming for image processing techniques. Studies using CUDA platform in GPU programming have been classified under 5 areas; image reconstruction, image enhancement, image segmentation, image registration and image classification. Advantages of using CUDA in GPU programming for image processing and issues to pay attention in applications have also been underlined.

引用

页码：737 / 754

页数：18

共 100 条

[61]

NVIDIA Accelerated Computing, 2017, GPU ACC LIB YAYIN TA, V10

[62] High-performance cone beam reconstruction using CUDA compatible GPUs [J].