CUDALink`
CUDALink`

CUDAImageConvolve

CUDAImageConvolve[img,kern]

gives the convolution of img with kern.

CUDAImageConvolve[list,kern]

gives the convolution of list with kern.

CUDAImageConvolve[mem,kern]

gives the convolution of mem with kern.

Details

  • The CUDALink application must be loaded using Needs["CUDALink`"].
  • CUDAImageConvolve works with binary, grayscale, and other images.
  • CUDAImageConvolve works with CUDAMemory and lists.
  • CUDAImageConvolve[image,kern] is equivalent to ImageConvolve[image,kern].
  • If multiple image channels are present, CUDAImageConvolve operates on each of them separately.
  • The following options can be given:
  • "Device"Automaticdevice used for CUDA computation
    Padding"Fixed"value used on boundary
    "OutputMemory"NoneCUDAMemory where output is stored

Examples

open allclose all

Basic Examples  (3)

First, load the CUDALink application:

This performs a mean filter:

Use CUDAImageConvolve on a grayscale image with a Sobel mask to detect vertical edges:

Using CUDAImageConvolve on a graphics object will automatically rasterize the graphic:

Applications  (2)

CUDAImageConvolve can be used to perform heat diffusion:

CUDAImageConvolve can be used to process geographic elevation data: