I have recently tried offloading my image processing code to the GPU but i am not getting the same image transformation output as the CPU code. The issue seems to be occurring when I try converting to a different color space. The code works great in the CPU code but when i try using the GPU module the image becomes fuzzy and distorted and 1/2 of the image is white Has anyone had issues with this call? I am running visual studio 2010 with opencv 2.4.9 and CUDA 6.5. Below is the code:
Mat m_MatSrc; // From file Mat m_MatDst; // After copy or resize cv::gpu::GpuMat dst, src;
capture.read(m_MatSrc);
src.upload(m_MatSrc);
//resize(m_MatSrc, m_MatDst, cv::Size(dst_width, dst_height), 0, 0, 1); gpu::resize(src, dst, cv::Size(dst_width, dst_height), 0, 0, 1);
//cv::cvtColor(m_MatDst, m_MatDst, CV_RGB2RGBA, 4);
gpu::cvtColor(dst, dst, CV_RGB2RGBA, 4);
dst.download(m_MatDst);
cv::imshow("Image2",m_MatDst); //Diagnostic View port
cvWaitKey(1);