摘要
基于对不完全朴素贝叶斯分类器的分析,提出一种离线训练在线识别的目标图像识别与跟踪实验方法。实验将模板图像与目标实时图像之间的特征匹配问题转换为特征分类问题,并在成像自寻的导引系统上运行。实验结果表明,该方法在保持很高鲁棒性的同时,大幅减少在线目标识别的计算量,具有较强的实时性。
Based on the research of Semi-navie Bayesian classifier, a target recognition and tracking experimental method using of^1ine training and online identification has been presented. This method converts features matching problem between the template image and the target image into feature classification. It can run on the imaging homing guidance system. The experimental results show that the method can significantly reduce the computation of the online target recognition while maintaining high robustness, and have a high re- M-time performance.
出处
《实验科学与技术》
2014年第1期19-21,共3页
Experiment Science and Technology
关键词
图像处理
贝叶斯分类器
目标识别
实验
image processing
Bayesian classifier
target recognition
experiment
