compute disparity map from stereo pair using SNCC matching function (GPU and CPU code provided)
Copyright (C) Chloe LeGendre
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
Title: SNCC stereo matching for KITTI data Author: Chloe LeGendre, chloelle (at) gmail.com Date: June 16, 2015
Computes disparity maps for KITTI data using SNCC matching function, per the following reference: N. Einecke and J. Eggert. A two-stage correlation method for stereoscopic depth estimation. In International Conference on Digital Image Computing: Techniques and Applications (DICTA), pages 227–234, 2010.
To run the program, you need the following in root directory where source code files exists: cudaSNCC.exe release version
source code files: CudaImageFunctions.cu ImageFunctions.cpp ImageFunctions.h produceDisparityMaps.cpp (main)
input directories containing initial stereo pairs: ./training_image_0 ./training_image_1 left images of stereo pair in folder ./training_image_0/ right images of stereo pair in folder ./training_image_1/
output directories (need these in root directory): ./disparity_GPU_SNCC5x5_13x13 ./disparity_CPU_SNCC5x5_13x13 ./costmap_GPU_SNCC5x5_13x13 ./costmap_CPU_SNCC5x5_13x13
miscellaneous other notes: -For CPU, NCC window size and summation filter window size can be modified in main (default = 5x5 for NCC, 13x13 for summation). -Output directory names should be changed to reflect updated window sizes. -For GPU, NCC window size is fixed at 5 and summation filter window size is fixed at 13. Cannot be changed. -The program takes no arguments. -You can modify whether or not SNCC uses CPU or GPU on line 327 in main function of produceDisparityMaps.cpp -Default / current operation mode = GPU. -You can modify which KITTI sequences and frames are used to compute disparity maps in lines 345 - 348 of produceDisparityMaps.cpp -The exclusive scan functions used in the GPU implementation produce one less disparity column and one less disparity row on the right / bottom of the output. -The GPU version makes guesses for disparity where the CPU does not, so the results will not be 100% identical when CPU produces 0-valued disparities. -KITTI training sequences 000000-000005 are included in this zip file for reference. This is meant as sample data only and is not the entire dataset.