GAOM - GPU-Acceleration for Optical Measurement

Introduction

This is the source code associated with the book GPU-Acceleration for Optical Measurement, SPIE Spotlight, under peer review. The details within each project folder are explained as below.

• ch2:
  • vec_add.h(.cpp) CPU and GPU implementations of vector addition
  • grad_calc.h(.cpp) Pointwise and tiled implementations of image gradient calculation
• ch3:
  • cu_wff.cu Parallel windowed Fourier filtering (WFF) algorithm
• ch4:
  • icgn.cu Parallel inverse Compositional Gauss-Newton (IC-GN) algorithm
  • precomputation.cu Parallelized precomputation for IC-GN
  • compute.cu Helper functions for the above CUDA kernels

Use the code

Project dependencies:

• CUDA 8.0 for writing GPU code
• OpenCV 3.1 for convenient image I/O

Compilation on Windows

• Download and install Visual Studio 2015+
• Download and install CUDA 8.0
• Install OpenCV 3.1:
  1. Download OpenCV 3.1
  2. Right click "Computer" -> Advanced System Settings -> Advanced -> Environment variables, add a new virable to "System Variables" as "OPENCV_DIR" with the value YOUR_OPENCV_INSTALLATION_PATH\build\x64\vc12(vc14)
  3. If you want your program to run directly, add "%OPENCV_DIR%\bin" to the "User Variables" 's "Path".

Compilation on MAC OS or Linux

• Please follow the CUDA programming guide

**Note: Only the "Release" mode was configured in the provided example code.