摘要
针对具有回射特性的合作目标在图像上所呈光斑的外接矩形区域总是图像中对比度较强的区域这一特性,本文提出了一种全新的基于空间对比度的光斑自动定位方法。鉴于空间对比度作为一种归一化的显著度度量可用于描述人眼对局部图像对比度的感知程度,新方法采用空间对比度作为待识别光斑外接矩形区域的基本特征对其进行定位,通过设定显著阈值及非显著阈值作为提取光斑外接矩形的条件,在计算得到光斑区域的灰度重心的基础上确定光斑的亚像素形心。数字仿真实验结果证明本文方法在各种噪声情况下均能够得到较好的光斑形心提取效果,可以达到与传统相关法相当的精度水平。实物仿真试验表明,由本文方法得到的光斑图像坐标进行双目像机交会计算得到的光斑三维位置的精度较高,且不受光斑位置突变的影响,从拍摄图像到计算得到光斑三维位置整个过程的数据输出率为15～20Hz,可以稳定得到高精度的结果。与传统的检测方法相比,本文算法概念新颖,易于实现,能够在满足精度要求的基础上同时满足实时性要求,具有较高的工程应用价值。
In view of the fact that light spots generated by the retro-reflected cooperative targets are always in the image local area of comparative higher contrast, a novel algorithm is presented for automatic light spots detection based on the spatial-contrast. As a normalized salient measure, the spatial-contrast can represent that human eye perceives the degree of the contrast of local images, so it is selected as the feature to localize the light spots areas in the new method. And the salient threshold and the non-salient threshold are set up to determine the beginning and the end of one light spot area＇s localization. Then, the center of light spot is obtained by calculating the sub pixel gray center of the whole local area and is used in the three dimensional position reconstructions of the light spots. Experiment results show that proposed method is simple to realize and meets the precision requirement of the real applications.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2007年第11期82-87,共6页
Opto-Electronic Engineering
基金
国家"863"计划资助项目(2005AA742032)
关键词
空间对比度
显著闽值
非显著阈值
定位
spatial-contrast
salient threshold
non-salient threshold
location
