针对航空影像中目标提取问题中存在的困难，提出了一种顾及目标的区域特征、边缘特征、形状的连续性与平滑性及人的识别能力的彩色影像中目标提取方法，根据这一算法的性质和行为特征，本文将其称为Octopus方法。本文最后给出了将Octopus用于彩红外影像中目标提取的例子。

A new mothod-called Octopus for color images segmentation which is robust,rapid and free of tuning parameters is presented in this paper. It acquires this properties by the synergy of seeded region growing, Snake(Active Contour Models) and Operator of the computer. The process of objects extraction starts with a SRG. The region boundary are modified by Snake on the basis of considering the boundary smoothness and variation of image gradient along the boungary. We illustrate the properties of Octopus on real image where the color information of the object are used.