TIEVisionImage.fastNlMeansDenoisingColored
Declaration
procedure fastNlMeansDenoisingColored(h: single = 3; hColor: single = 3; templateWindowSize: int32_t = 7; searchWindowSize: int32_t = 21); overload; safecall;
procedure fastNlMeansDenoisingColored(outImage: TIEVisionImage; h: single = 3; hColor: single = 3; templateWindowSize: int32_t = 7; searchWindowSize: int32_t = 21); overload; safecall;
Description
Perform image denoising using the Non-local Means Denoising algorithm with several computational optimizations. Noise is expected to be a gaussian white noise.
Note:
◼Pixel format of input image must be ie24RGB
◼Also available as a
RetouchTool by setting
MouseInteractGeneral to [miRetouchTool] and
RetouchMode to iermIEVisionSmooth
Parameter | Description |
outImage | ie24RGB image for output |
h | Filter strength for image luminance. Larger h values perfectly remove noise but also remove image details, smaller h values preserve details but also noise |
hColor | The same as h but for image color. For most images a value of 10 will be enough to remove colored noise and not distort colors |
templateWindowSize | Size in pixels of the template patch that is used to compute weights. Should be odd. Recommended value 7 |
searchWindowSize | Size in pixels of the window that is used to compute weighted average for a given pixel. Should be odd. This value affects performance linearly, i.e. a larger value for searchWindowsSize will mean a longer denoising time. Recommended value 21 |
| Demos\IEVision\Denoising\Denoising.dpr |
| Demos\ImageEditing\EveryMethod\EveryMethod.dpr |
Also see:
Comparison of smoothing, blurring and noise reduction methods
// Load test image
ImageEnView1.IO.LoadFromFile( 'D:\TestImage.jpg' );
// Perform image denoising using the Non-local Means Denoising algorithm
ImageEnView1.IEBitmap.GetIEVisionImage().fastNlMeansDenoisingColored();
ImageEnView1.Update();
// Perform more aggressive denoising
ImageEnView1.IEBitmap.GetIEVisionImage().fastNlMeansDenoisingColored( 10, 10 );
ImageEnView1.Update();
See Also
◼SymmetricNearestNeighbour