运动目标的分割及其神经网络的实现

SEGMENTATION OF MOVING OBJECT USING NEURAL NETWORK

本文提出了用改进的Kohonen神经网络进行图像自动分割的方法。实验表明该方法不仅能使复杂背景中的目标被较好地提取出来,而且处理速度快,为图像序列分析中运动目标的实时提取工作奠定了良好的基础。

We present an approach which uses improved Kohonen Neural Networks to automatic Image Segmentation. Computer experiments proved that this method can not only ameliorate the object extraction from a complicate background, but also highly speed up the system's processing. It will certainly lay a foundation for the real-time extraction of moving object in Image Sequence Analysis.