This paper presents the methodology for acquiring features of motion in an image sequence automatically by means of a competitive evolutionary system that is called Artificial Cell (AC). AC simulates the metabolism of a biological cell by treating pixel values or texture of images as an environment. The adaptability of ACs to the motion feature in the image sequence and competition among ACs enables automatic extraction of the motion features. The technique is applied to extraction of the motion features from the sequential images of amoeba.
展开▼