抄録
This paper describes a method of generating Freeman chain code from edge coordinates using a State transition graph. Segments which indicate a existence of object across one horizontal line are classified in to seven types according to connectivity between each segments on two adjacent horizontal line. Seven types reclassified in to five features of segments using the state transition graph. The five features are Appearance, Disappearance, Branching, Joining and Connecting. The table including the edge coordinates and the features, generates Freeman chain code easily and identify separation of two or more objects. The capacity of memory for storing the edge coordinates data and the feature data is smaller than that for storing binary image data. The processing time is very short because large amount of data remain not processed.
