Out of Memory Error with Particle Analysis

Share icon

Mon, 05/06/2013 - 09:49 am

Hi,

I have a large image (8960 x 8960) that is made up FP values in the range of -1 to +0.75. This is height data off of a Zygo metrology tool.

I am attempting to do some particle analysis on this data to highlight high and low stops.

When I use panel driven analysis it seems to work. What i want to change is to create a threshold mask based on other criteria and feed it into the particle analysis function:

The panel driven appears to do this:

ImageThreshold/I/T=(0)/Q root:Master
ImageAnalyzeParticles /E/W/Q/M=3/A=48/EBPC stats, root:Master_WMUF:Particles:M_ImageThresh

What I have done is to create a new wave called "fred" for test purposes to create an new threshold wave
fred = master >0.5? 255 : 0
and Fred is an unsigned I8 wave

If I call

ImageAnalyzeParticles /E/W/Q/M=3/A=48/EBPC stats, fred

I get an out of memory error.

What am I doing wrong?
Hello Andy,
Please remember that ImageAnalyzeParticles uses the value 0 to denote a particle -- so make sure that you are not switching between the foreground and the background.

Since your command uses /A=48 it means you could still have more than 1M particles so try to run the command first without the /E/W flags. If you still get an out of memory error please compress the input wave and send it to me to investigate further.

A.G.
WaveMetrics, Inc.

May 6, 2013 at 10:14 am - Permalink

Hi A.G.

Question about which is particle and which is background.
You mentioned that 255 = background and 0 = particle

I ran the imagethreshold with an interval with two cases with and without /I

I have set the interval at 0.1 and 0.8.

In Image without I Flag and this gives me an out of memory error.

With /I Flag

The the areas in the interval are now white (i.e. 255)

This works and does not give me the out of memory error.

It appears that regions of value 255 in the threshold mask are considered particles.
Without I flag.png with I Flag.png Result with I flag.png

May 6, 2013 at 10:43 am - Permalink

hegedus wrote:

The the areas in the interval are now white (i.e. 255)


I'd be very careful with this and verify (by placing a cursor on the image) that the pixels in question are indeed 255.

You need to use ImageThreshold with /i flag to create an input for ImageAnalyzeParticles.

AG

May 6, 2013 at 12:32 pm - Permalink