diff options
Diffstat (limited to 'pkg/proto/doc/bscale.hlp')
| -rw-r--r-- | pkg/proto/doc/bscale.hlp | 151 |
1 files changed, 151 insertions, 0 deletions
diff --git a/pkg/proto/doc/bscale.hlp b/pkg/proto/doc/bscale.hlp new file mode 100644 index 00000000..af5647f8 --- /dev/null +++ b/pkg/proto/doc/bscale.hlp @@ -0,0 +1,151 @@ +.help bscale Aug91 proto +.ih +NAME +bscale -- linearly transform the intensity scales of a list of images +.ih +USAGE +bscale input output +.ih +PARAMETERS +.ls input +List of images to be transformed. +.le +.ls output +List of output transformed images. If the output list is the same as the input +list the input images are overwritten. +.le +.ls bzero = "0." +The zero point to be subtracted before applying the scale factor. +The options are a numerical value, "mean", "median", or "mode". +.le +.ls bscale = "1." +The scale factor to be applied. The options are a numerical value, +"mean", "median", or "mode". +.le +.ls section = "" +The image section to be used for computing the image statistics. If section +is "", \fIstep\fR is used to define the default image section. \fISection\fR +is used to confine the computation of the mean, median, and mode +to a specific region of the image. +.le +.ls step = 10 +The step size in pixels which defines the default image section to be used +for computing the mean, median, and mode. +In image section notation the default section is equivalent to [*:10,*:10,...]. +\fIStep\fR is used if +the sampling along an axis is not defined by \fIsection\fR. +Subsampling can significantly reduce the memory and +time required for the computation of the mean, median, and mode. +.le +.ls upper = "INDEF" +Upper intensity limit to be used for computing the mean, median, and mode. +.le +.ls lower = "INDEF" +Lower intensity limit to be used for computing the mean, median, and mode. +.le +.ls verbose = yes +Print messages about actions taken by the task? +.le + +.ih +DESCRIPTION + +The specified input images \fIinput\fR are linearly transformed in intensity +and written to the list of output images \fIoutput\fR, using the +zero point specified by \fIbzero\fR and the scale factor specified by +\fIbscale\fR. If the output image list +is the same as the input image list the input images will be overwritten. + +The expression defining the linear transformation is listed below. + + NEW = (OLD - BZERO) / BSCALE + +OLD is the input pixel brightness, NEW is the output +pixel brightness, BZERO is the zero point offset, and BSCALE is the +scale factor. The values of the scaling parameters \fIbzero\fR and +\fIbscale\fR +may be specified explicitly or the mean, median, or mode of the image +may be used for either quantity. If the input image pixel type +is short, integer, or long, overflow or truncation may occur. + +When one of the scaling parameters is the image mean, median, +or mode, then the image mean, median, and mode are calculated. The statistics +computation can be restricted to a section of the input image by setting +the parameter +\fIsection\fR. Otherwise the parameter \fIstep\fR is used to +define a default image section. +Subsampling the image can significantly reduce the memory +and time requirements for computing the statistics of large images. +If numerical values for both the scaling parameters are specified, then +the image statistics are not computed. The statistics computation can +be limited to given intensity range by setting the parameters +\fIlower\fR and \fIupper\fR. + +The mean, median, and mode are computed using the following algorithm. +Note that this algorithm requires that all the data to used for computing +the statistics must be in memory. + +.nf +1. The data in the specified image section is read into a buffer. +2. The data is sorted in increasing order of intensity. +3. The points outside upper and lower are excluded. +4. The median is set to the data value at the midpoint of the remaining + data. +5. The mean and sigma of the remaining data are computed. +6. The histogram bin width (.1*sigma) and separation (.01*sigma) are + computed. +7. The location of the bin containing the most data points is determined. +8. The median of the data values in that bin is used to estimate the mode. +.fi + +.ih +EXAMPLES + +1. Use the mode to subtract a constant background from a list of images. +Overwrite the input images. + +.nf + cl> bscale *.imh *.imh bzero=mode +.fi + +2. Scale a list of images to a unit mean. Overwrite the input images. + +.nf + cl> bscale *.imh *.imh bscale=mean +.fi + +3. Scale a list of images to the intensity range 0 to 511, +where 234. and 1243. are the original data range. Overwrite the input +images. This example uses the CL to calculate bscale. + +.nf + cl> bscale.bzero = 234. + cl> bscale.bscale = (1243. - 234.) / 512. + cl> bscale *.imh *.imh +.fi + +4. Scale an image using a user specified bzero and bscale and create a new +output image: + +.nf + cl> bscale imagein imageout bzero=0.0 bscale=1.10 +.fi + +5. Median subtract a list of input images using the percent replace facility to +create the output image names. + +.nf + cl> bscale images*.imh %i%outi%*.imh bzero=median bscale=1.0 +.fi + +6. Repeat the previous example but use the @ file facility for specifying +the input and output image lists. + +.nf + cl> bscale @infile @outfile bzero=median bscale=1.0 +.fi + +.ih +SEE ALSO +imarith,imcombine +.endhelp |