# Random points from an image using a BernoulliDistribution

I am a Mathematica newbie. I need to sample random points from an image which I have stored in a variable img as:

image=Import["ExampleData/ocelot.jpg"]
img = ImageData[image]


Can you guys suggest how I can use a BernoulliDistribution to sample points from the image using Flatten, Map, RandomVariate, Partition and Image functions?

I know a similar problem was discussed here but the solutions there are very convoluted and I am not able to follow them. Thanks a lot for the help and very sorry if this seems like a duplicate post but I have spent the whole evening banging my head over the link to the post above but am not able to reproduce the same with the above functions. Thanks again!

• Can you upload image.jpg somewhere? Oct 2, 2018 at 5:25
• Hi, I have updated the question. Thanks a lot for the help! Oct 2, 2018 at 5:49
• Can you please clarify if you actually have a grayscale image, or a binary image? The new example you gave is grayscale, so sampling over that is not as clear-cut as the binary image case, which is already dealt with in the thread you linked to. Oct 2, 2018 at 6:44
• The image I need to work with is grayscale. Thanks! Oct 2, 2018 at 7:07

I'm assuming you want to vary the Bernoulli parameter p depending on the pixel grayscale value. Darker areas of the image are more likely to generate a black pixel and lighter areas more likely to be white.

img = Import["ExampleData/ocelot.jpg"];
dat = ImageData[img];
result = Image[
Map[RandomVariate[BernoulliDistribution[#]] &, dat, {2}]]


The positions where a 1 appears are given by:

PixelValuePositions[result, 1]


The sum of many images averaged over time converges to the grayscale image in the limit.

img = Import["ExampleData/ocelot.jpg"];
dat = ImageData[img];
ListAnimate[