Member | Value | Description |
---|---|---|
doFilterBSpline | 2 | This smoothing procedure interpolates using a Cubic B-Spline, which has the advantage of guaranteeing continuity in both the first and second differentials of the image, in both directions. No other parameters are used in this smoothing method, which produces good visual results, though some users find them a little too “smoothed”
|
doFilterCubicSpline | 3 | This is similar to the B-Spline described above, but uses an additional parameter (CubicSplineAlpha) to control the degree of smoothing or edge enhancement. The default value is –0.75, and normal values range fro –0.5 to – 1.25, though higher negative values may be used to produce deliberate edge enhancement, though generally, UnsharpLength and UnsharpEnhancement should be used for this purpose, as they are independent of the smoothing method. |
doFilterDefault | 0 | For MagnificationMode, this acts identically to doFilterReplicate and for MinificationMode it acts as doFilterMovingAverage. This behaviour is the default, and is deigned to minimise regulatory issues by doing the minimum necessary to display all pixel data, without adding extra “image processing”. |
doFilterGpu | 8 | This mode uses interpolation in the GPU to apply magnification or minification. It does not give as much flexibility as the other modes, but may be significant faster, whilst still avoiding a "blocky" effect for magnification and avoiding aliasing on grid-lines for minification. |
doFilterMovingAverage | 4 | This mode is most applicable when zoom is <1, and calculates pixels values as a weighted average of the surrounding values. The number of values used for this calculation depends on the value of FilterLength. If this is 0 (the default), then a suitable values is determined internally, based on the zoom (approximately 2/ActualZoom), but if FilterLength is non-zero then 2*FilterLength values are used. |
doFilterReplicate | 1 | This would be more completely termed doFilterReplicateOrDecimate, and reproduces the behaviour used in versions of DicomObjects prior to 4.2. If zooming up is necessary, then pixels are simply replicated, giving a possibly “blocky” but honest representation of the pixel data. For minification, rows and columns are simply omitted as necessary. |