摘要
在试验设计方法学的基础上，提出了一种定量评价自动目标识别算法性能的方法；给出了几种刻划场景变化的图像质量指标的定义和度量方法；提出了用响应曲面模型方法建立算法性能模型和描述图像度量与算法性能之间统计关系的方法．实验结果表明，仅需要少量的测试图像就可得到可靠的算法性能模型和可信的评价结果．
n the basis of experimental design methodology, a method for quantitative evaluation ofthe algorithm for automatic target recognition is presented. The definitions and metric methods of several quality indices for the description of the iamges due to scenery variation are given. The responsesurface modelling approach is employed to analyse the performance of the segmentation method oncommon data. Test. results show that only a few frames of tested images are needed to obtain good andreliable performance models and credible results of evaluation.
出处
《华中理工大学学报》
CSCD
北大核心
1996年第2期43-45,共3页
Journal of Huazhong University of Science and Technology
基金
国防科技预研基金
航空航天基础性研究基金
关键词
自动目标识别
算法性能
试验设计
遥感图像
automatic target recognition
performence evaluation of algorithm
experimental designmethodology
image metric
response surface modelling
