本文提出了一种基于目标特征的多门限图像分割方法．它通过以特征量匹配的方式将属于景物的先验空间知识引入分割过程，使图像分割由单纯的灰度聚类过程变成灰度一几何空间聚类过程．实验结果表明，本文提出的方法可以完整地分割出复杂背景图像中本身具有多个灰度层次的景物． This paper presents a multi-threshold image segmentation method based on the target features. It introduces the priori spatial knowledge belonging to the scene into the segmentation process by means of the feature quantity matching, which makes the image segmentation change from simple gray clustering process to gray-scale geometric space clustering process. The experimental results show that the method proposed in this paper can completely segment the scene with multiple gray levels in the complex background image.