Image segmentation is the process of identifying pixels that belong together in an image, and it is used for object recognition and background removal purposes. Geodesic Active Contour is an algorithm that utilises level set formulation to segment an image into objects, even with poorly defined boundaries. The process involves initialisation of contours, converting them to a level set representation, calculating the energy function of contours by combining internal and external energy functions, and finally evolving the contours by iterating through numerical methods. The algorithm updates the level set representation until it converges, capturing the actual boundaries in the image. The precise segmentation of the object leads to connected points that represent the boundary of the segmented object.

source update: The Key to Image Object Separation – Towards AI