基于单层感知器的彩色图像目标提取方法  

Object Extraction Algorithm for Color Image Based on Single-Layer Preceptron

作　　者：罗年[1] 钟平[1] 王士乐[1] 涂新星[1] 

机构地区：[1]东华大学理学院,上海201620

出　　处：《光学与光电技术》2010年第6期9-12,共4页Optics & Optoelectronic Technology

基　　金：上海市科学技术委员会基础研究重点基金(08JC1400200)资助项目

摘　　要：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 layer preceptron is proposed in this paper. By creating a single layer preceptron network which with color components as the inputs 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 experiment results of some other mainstream methods, this method could separate the objects from the image more exactly and effective, furthermore, this al gorithm is easy to implement and has a wide range of applications.

关 键 词：目标提取 单层感知器 颜色空间 神经网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]