CImCombine:RankClip

The RankClip method merges the pixel values after clipping outliers in the upper and lower rank percentiles of the distribution at each point.

Syntax

CImage = CImCombine:RankClip( CImageSet )

CImage = CImCombine:RankClip( CImageSet, pHigh )

CImage = CImCombine:RankClip( CImageSet, pHigh, pLow )

where

CImageSet contains the collection of images to be combined.

pHigh is the upper rank percentile for rejecting points, e.g., pHigh = 2 rejects points in the upper 2 percentile of the distribution—that is, points at or above the 98th percentile; default value is 2.5.

pLow is the lower rank percentile for rejecting points, e.g., plow = 2 rejects points in the lower 2 percentile of the distribution; default vlaue is 2.5.

CImage is a new image created by this method. On failure, nil is returned.

Each pixel in the output image is the mean value of all images after rejection of images having values inside the upper nHigh percentile and images having pixels in the lower nLow percentile of values at each location.

Example

Suppose a CImageSet named S exists. The following script combines the images by the RankClip method by excluding points above the 95th percentile and below the 2nd percentile. The images are normalized by the Median statistic computed inside the central 10% of the image:

C = CImCombine:new()	-- create a CImCombine object
R = NewRect(10,50,10,50)	-- central 10% of the image
C:SetRelRect( R )	-- specify CImage and CRect to measure
C:SetNormStat("median")	-- use the RankClip as the normalization statistic
C:SetNormMethod("scale")	-- scale to normalize the image set
I = C:RankClip( S, 5, 2 )	-- combine the image set and return a new image
I:Display()	-- show the new image