Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Vechalapu_new.pdf (1.71 MB)

ETD Abstract Container

Abstract Header

Real-Time 2D Digital Image Correlation to Measure Surface Deformation on Graphics Processing Unit using CUDA C

vechalapu, uday bhaskar
http://rave.ohiolink.edu/etdc/view?acc_num=akron1522108261138081

Abstract Details

2018, Master of Science, University of Akron, Electrical Engineering.
A digital image correlation (DIC) method is implemented on a graphics processing unit (GPU) computational platform, to analyze full-field displacements of points on the surface of a specimen under structural test, and to provide real-time strain measurement with a level of accuracy sufficient for monitoring for safety and limit checking during test. The full-field displacements are measured to the nearest integer pixel using a DIC method based on a zero-mean normalized cross correlation (ZNCC) measure, and the integer pixel displacements are refined to sub-pixel accuracy using a spatial gradient method. The entire surface deformation analysis application is implemented on GPU controlled by a host CPU (CPU+GPU) using CUDA C, and compared to a version implemented in C on just the host CPU. The parallel computing programming model CUDA helps to implement compute-intensive DIC algorithms to exploit a GPU’s huge computational power to achieve high performance for our real-time application. Using iterative GPU code optimization techniques that are guided by analysis of profiler data on the running code, various code optimization strategies are deployed for both the ZNCC measure and the spatial gradient method to improve the overall performance. Experiments conducted on images ranging in size from 256-pixels × 256-pixels to 512-pixels × 512-pixels show speedup of about 800× to 1000× for the optimized integer pixel calculations done on CPU+GPU over the CPU-only implementation; speedup of the spatial gradient method ranged from about 44× to 72×. The result is a real-time algorithm that analyzes full-field displacements and measures strain of a surface on a GPU computational platform with high computational efficiency and measurement accuracy.
Dr. Joan Carletta (Advisor)
Dr. J. Alexis De Abreu Garcia (Committee Member)
Dr. Jin Wei Kocsis (Committee Member)
Dr. Michael McGaw (Committee Member)
120 p.
Computer Engineering; Electrical Engineering
CUDA, GPU, Digital Image Correlation, DIC, ZNCC, Strain measurement

Recommended Citations

Citations

  • vechalapu, U. B. (2018). Real-Time 2D Digital Image Correlation to Measure Surface Deformation on Graphics Processing Unit using CUDA C [Master's thesis, University of Akron]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=akron1522108261138081

    APA Style (7th edition)

  • vechalapu, uday bhaskar. Real-Time 2D Digital Image Correlation to Measure Surface Deformation on Graphics Processing Unit using CUDA C. 2018. University of Akron, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=akron1522108261138081.

    MLA Style (8th edition)

  • vechalapu, uday bhaskar. "Real-Time 2D Digital Image Correlation to Measure Surface Deformation on Graphics Processing Unit using CUDA C." Master's thesis, University of Akron, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=akron1522108261138081

    Chicago Manual of Style (17th edition)

akron1522108261138081
32
© 2018, all rights reserved.
This open access ETD is published by University of Akron and OhioLINK.