Binary images in general will contain more than one object. In order to identify or classify the objects in the image, we must first identify the connected components, that is the distinctly connected blobs that correspond to each object in the image. At this stage, we are assuming that each object is isolated from the others, that is, there is no occlusion. The sequential scan labelling algorithm we describe comes from Horn's book [2].
We then go on to describe the technique of mathematical morphology for low-level image analysis. The notes in this section follow closely those of Gonzalez and Woods [1].