Build a picture from a data table then apply jsl filters on it

Franck_R

Hi,
I work a lot with X/Y/metric data tables which are similar to images and on which I would like to apply JMP filters ("despeckle", "reduce noise",... I was very interested in this presentation by Julia Gong: "Using JMP® Image Analysis Capabilities to Extract Skin Lesion Characteristics") then transform back to data table. It seems easier to use that kind of filter after transforming my data to picture than perform some tricky matrix calculations.

Like:

 

to get for example just the bottom right cluster high level pixels with noise reduction jsl filter, assuming isolated pixels are noise and has to be filtered

 

But I don't understand how to create JMP image matrix from my data.

 

For example if I start from a table with X, Y and a metric below and as an attachment

 

Then:

dt =current data table();
// the X,Y, metric matrix
mat = (:X << getvalues) || (:Y << getvalues) || (:metric << getvalues);

 Neither this command:

 

img = New Image( "r",{mat});

nor this one:

img << Set pixels ("r",{mat});

...seems to works.

 

How can I transform this table into an image matrix and apply JMP filters to it?

Thanks!

 

Franck

data_table.jmp

ian_jmp

You will need to do something like this:

NamesDefaultToHere(1);

dt = Data table("data_table.jmp");

// The values of the metric need to be rescaled to [0, 1]
metric = Column(dt, 3) << getValues;
maxMetric = Max(metric);
minMetric = Min(metric);
metric = (metric - minMetric) :/ (maxMetric - minMetric);

// Column vector to hold the pixel values for the image
pixVals = J(NRow(metric), 1, .);
for (p=1, p<=NRow(pixVals), p++,
	pixVals[p] = RGB Color(metric[p], metric[p], metric[p]);
	);

// Reshape 'pixVals' to be a rectangular matrix (assumes 'X' and 'Y' both start at 1)
xMax = Max(Column(dt, "X") << getValues);
yMax = Max(Column(dt, "Y") << getValues);
pixVals = Shape(pixVals, xMax, yMax);

// Make an image, make it a bit bigger, and put it in a window
im = New Image();
im << setPixels( pixVals );
im << setSize({10*xMax, 8*yMax});
NewWindow("My Image", im);

Use 'Help > Scripting Index' to clarify how the code works.