In the region analysis work done by Brice and Fennema (1969), a one-pass algorithm was used to initially partition the picture into named, homogeneous, connected components and to set up the region structure (AROS). This one-pass algorithm is rather complex in nature, and several alternatives have been suggested, one of which (a two-pass algorithm) we have chosen to use for comparison purposes because of its simplicity and apparent efficiency. Both algorithms use an L-shaped window to compare the elements of the picture array. If two points differ in properties (only horizontal and vertical comparisons are made), a pair of elementary vectors are placed to separate them. The curves made up of these elementary vectors partition the picture into regions and serve as the boundaries of these regions. The two algorithms discussed here differ in the way they label the picture elements and the boundaries. Section II is a description of the one-pass algorithm, Section III explains the two-pass algorithm, and Section IV is a comparison of their performance.