This Project implements a CUDA application with camera input support over v4l2 library. A CUDA Kernel computes an RGB image from the captured Bayer pattern images. The RGB image stream can be displayed with OpenCV or OpenGL. At configure time the support of OpenGL or OpenCV can be disabled. Both sinks are enabled by default.
-
Compile: ./autogen.sh [--without-{opencv|opengl}] make
- Compilation with OpenCV and OpenGL support.
-
Usage: cuda_debayer [-h|--help]
-
or: cuda_debayer [-d|--device=/dev/video0] [-e|--exposure=30000] [-g|--gain=100] [-o|--output=opengl] [-s|--scale=1]
- on a platform with OpenCV and OpenGL runtime library
- Options:
- -d | --device
- Video device name [default:/dev/video0] - -e | --exposure
- Set exposure time [1..199000; default: 30000] - -g | --gain
- Set analog gain [0..244; default: 100] - -h | --help
- Print help message - -o | --output
- Outputs stream over OpenCV/OpenGL [opencv, opengl; default: opengl] - -s | --scale
- Decreasing size by factor [1..20; default: 1]
Run tests with: make check
- The test includes a test whether the CUDA kernel calculate the correct values.