Graph cut methods are optimization techniques used primarily for image segmentation by partitioning a graph into disjoint subsets that minimize a predefined energy function. These methods leverage graph theory to solve complex computer vision problems with efficiency and scalability.