WOLFRAM

Wolfram 语言与系统参考资料中心

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.

更多信息和选项
Details and Options Details and Options
范例  
基本范例  
Applications  
参见
技术笔记
相关指南
按以下格式引用：
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.

更多信息和选项

  • 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

范例

打开所有单元 关闭所有单元

基本范例  (3)

First, load the CUDALink application:

Wolfram Language code: Needs["CUDALink`"]

This performs a mean filter:

Wolfram Language code: CUDAImageConvolve[[image], N[BoxMatrix[1] / 9]]

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

Wolfram Language code: CUDAImageConvolve[[image], {{-1, 0, 1}, {-2, 0, 2}, {-1, 0, 1}}]

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

Wolfram Language code: CUDAImageConvolve[[image], {{1, 1, 1}, {1, -8, 1}, {1, 1, 1}}]

Applications  (2)

CUDAImageConvolve can be used to perform heat diffusion:

Wolfram Language code: initialPoints = ConstantArray[0.0, {1000, 1000}]; Do[ initialPoints[[Sequence@@(RandomInteger[{1, #}]& /@ Dimensions[initialPoints])]] = RandomReal[1.0], {10} ];
Wolfram Language code: frame = CUDAMemoryLoad[initialPoints]; Dynamic[Refresh[ CUDAImageDivide[ CUDAImageConvolve[CUDAMemoryGet[frame], 1.0 * {{0, 1, 0}, {1, -4, 1}, {0, 1, 0}}, "OutputMemory" -> frame], 5.0, "OutputMemory" -> frame]; Image[CUDAMemoryGet@frame], UpdateInterval -> 0]]

CUDAImageConvolve can be used to process geographic elevation data:

Wolfram Language code: gisData = Import[ "http://exampledata.wolfram.com/sdtsdem.tar.gz", "Data" ];
Wolfram Language code: ReliefPlot[CUDAImageConvolve[gisData, {{1, 0, -1}}], ColorFunction -> "GreenBrownTerrain"]
Wolfram Research (2010)，CUDAImageConvolve，Wolfram 语言函数，https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html.

文本

Wolfram Research (2010)，CUDAImageConvolve，Wolfram 语言函数，https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html.

CMS

Wolfram 语言. 2010. "CUDAImageConvolve." Wolfram 语言与系统参考资料中心. Wolfram Research. https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html.

APA

Wolfram 语言. (2010). CUDAImageConvolve. Wolfram 语言与系统参考资料中心. 追溯自 https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html 年

BibTeX

@misc{reference.wolfram_2026_cudaimageconvolve, author="Wolfram Research", title="{CUDAImageConvolve}", year="2010", howpublished="\url{https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html}", note=[Accessed: 09-August-2026]}

BibLaTeX

@online{reference.wolfram_2026_cudaimageconvolve, organization={Wolfram Research}, title={CUDAImageConvolve}, year={2010}, url={https://reference.wolfram.com/language/CUDALink/ref/CUDAImageConvolve.html}, note=[Accessed: 09-August-2026]}