Set Samples kernel
setSamples - goes through the entire image and collects samples from the image (the current pixel and the next x number of pixels). __global__ void setSamples(cv[[File::cuda::PtrStepSz<float> samples, cv::cuda::PtrStepSz<uchar> img, int dimC) { int i = blockIdx.y*blockDim.y + threadIdxSetSamplesKernel.y; int j = blockIdx.x*blockDim.x + threadIdx.x; if (i >= img.rows png|| j >= img.cols) return; int index = i * img.cols + j; for (int d = 0; d<dimC; d++) { samples(index, d) = (float)img(i, j * dimC + d); } }550px]]
calculateDistance - goes through the image and computes the difference between the samples and the centers from the input image.
__global__ void calculateDistance(cv::cuda::PtrStepSz<float> centers, cv::cuda::PtrStepSz<float> samples, int k, int N, int dim, double* minval, float* D) { // Compute distances between already sampled centers and other input samples. // Update nearest distance if it is smaller than previous ones. int col = blockIdx.x * blockDim.x + threadIdx.x; int row = blockIdx.y * blockDim.y + threadIdx.y; int i = col + row * N; //int i = blockIdx.y*blockDim.y + threadIdx.y; if (i >= N) return; double dist = 0.0; for (int d = 0; d<dim; d++) { double diff = centers(k - 1, d) - samples(i, d); dist += diff * diff; } if (dist < minval[i]) { minval[i] = dist; } *D += minval[i]; }Calculate Distance kernel
generateImage - takes the modified image and then writes it to the file using the function "out()" [[File:CalculateDistanceKernel.png|550px]]
__global__ void generateImage(cv::cuda::PtrStepSz<uchar> out, cv::cuda::PtrStepSz<int> indices, cv::cuda::PtrStepSz<float> centers, int dim) { // Generate output image int i = blockIdx.y*blockDim.y + threadIdx.y; int j = blockIdx.x*blockDim.x + threadIdx.x; if (i >= out.rows || j >= out.cols) return; int index = i * out.cols + j; int ci = indices(index, 0); for (int d = 0; d<dim; d++) { out(i, j*dim + d) = (uchar)centers(ci, d); } }Image kernel
[[File:GenerateImageKernel.png|550px]] After programming these kernel. we noticed an improvement in performaceperformance.
==== Conclusion ====
This program can further be improved by off-loading some more operations from the CPU to the GPU. But this will require more time and research.
=== Assignment 3 ===
