基于类别方差的自动多门限图象分割

从模式识别的分类理论出发，对基于类别方差的自动单门限图象分割方法进行了扩展，给出一种适合于灰度图象的自动多门限图象分割方法，对灰度图象能够自动寻找最优的门限值（一个或多个），与人类的视觉分割过程一致。实验表明，能够自动且有效地进行灰度图象的多级门限值分割。

Approach of Automatic MultiThreshold Image Segmentation Based on Class Variance

： In this paper，we expand the one threshold gray－level image segmentation method and give an approach of automatic multithreshold gray－level image segmentation based on the between class variance using the classification theory in pattern recognition．This approach can automatically select the optimum threshold（one or more）of gray－level image．Experimental results show the effectiveness of this method．