针对复杂背景下的彩色图像的目标提取,提出了一种基于单层感知器的目标提取方法。通过建立以颜色分量为输入、阈值型函数为输出的单层感知器网络并训练得到最适权重,实现彩色图像的目标与背景分离。与现有主流目标提取算法进行对比实验,结果表明,该方法能够更加准确、有效地分离目标和背景,且方法适用范围广,实现简单。

In order to extract objects from color images under complex background, a method based on single layerpreceptron is proposed in this paper. By creating a single layer preceptron network which with color components as theinputs and threshold-type function as the output and training this network, a group of best weights could be obtained,which would be used for extracting the objects from the color image background. Comparing with the experimentresults of some other mainstream methods, this method could separate the objects from the image more exactly andeffective, furthermore, this algorithm is easy to implement and has a wide range of applications.