Optimized GPU Framework for Speckle Reduction Using Histogram Matching and Region Growing

Author

Shi, Dan ; Li, Xiaoying ; Liu, Dong C.

Author_Institution

Comput. Sci. Coll., Sichuan Univ., Chengdu, China

fYear

2010

fDate

18-20 June 2010

Firstpage

1

Lastpage

4

Abstract

A GPU framework for ultrasound speckle reduction by region growing based on local statistics extracted from the histogram shape is presented. The required image processing is computationally intensive, involving histogram calculation, region growing, box filtering using different sizes of windows, and more. In this paper, we describe the use of a graphics processing unit for implementing image processing algorithms for speckle reduction that achieves a frame rate of 28 fps for the 512×512 image, about 81 times faster than the CPU implementation. Testing results from GPU and CPU are compared in terms of visual image quality and program runtime from different image sizes.

Keywords

biomedical ultrasonics; computer graphics; feature extraction; medical image processing; optimisation; box filtering; feature extraction; graphics processing unit; histogram calculation; histogram matching; histogram shape; image processing; local statistics; optimized GPU framework; program runtime; region growing; ultrasound speckle reduction; visual image quality; Central Processing Unit; Filtering; Graphics; Histograms; Image processing; Shape; Speckle; Statistics; Testing; Ultrasonic imaging;

fLanguage

English

Publisher

ieee

Conference_Titel

Bioinformatics and Biomedical Engineering (iCBBE), 2010 4th International Conference on

Conference_Location

Chengdu

ISSN

2151-7614

Print_ISBN

978-1-4244-4712-1

Electronic_ISBN

2151-7614

Type

conf

DOI

10.1109/ICBBE.2010.5517199

Filename

5517199