Random sample consensus

Random Sample Consensus, or RANSAC, is a way to find the best possible match between data points and a model. Imagine you have a bunch of dots on a piece of paper, and you want to draw a straight line that goes through a bunch of them. However, some of the dots might not be very good ones to use in determining where the line goes, because they might be outliers. Outliers are dots that are far away from most of the other dots, or don't seem to fit the pattern you're trying to find.

So how do you deal with these outliers? That's where RANSAC comes in. First, you randomly select a few dots from your group of dots. These are your "inliers" that might be good ones to use in finding your straight line. Then, you use those inliers to create a line, and see how many other dots fit that line. If enough dots fit the line, you assume that it's a good fit for your data, and you keep it as your model.

However, if too many outliers are still causing problems, you start over with a different set of randomly selected inliers, and repeat the process until you find a model that works well enough. This way, even if there are outliers in your data, you can still find a good fit for the majority of the points.