I have an 8-bit unsigned char image, which is grayscale (values between 0-255), and I want to use the NPP CUDA library documentation, which provides some functions such as nppiSum_8u64s_C1R, to find the sum and average over all the pixels.
But, I don't know what oSizeROI stands for in the nppiSum_8u64s_C1R function? Should I just input my image dimensions w256 x h1024 here?
image_statistics__functions.html#gad5ca4eb684d5414e6026564ea63a1437">http://graphics.im.ntu.edu.tw/~bossliaw/nvCuda_doxygen/html/group_image_statistics__functions.html#gad5ca4eb684d5414e6026564ea63a1437
In the NPP function nppiSum_8u64s_C1R, the argument oSizeROI is the size (in pixels) of the input image. So yes, you should just input the image dimensions here.