摘要
利用小波分析的多分辨率特性和时频局部窗特性能够克服传统方法的局限,实现小目标的精确定位检测。基于小波变换的小目标检测,先利用改进的小波变换选择合理的小波基和小波变换窗口对小目标图像滤波,后结合直方图阈值分割法分割阈值。实验结果表明该方法不仅能很好保存小目标的形状特征,还能将背景几乎完全消除。
The shortcomings of the traditional methods are overcome and the precision localization of the small target detection is realized with wavelet multi-resolution analysis and time-frequency local window analysis. The small target detection based on wavelet transform is that reasonable wavelet basis and wavelet transform window is selected with improved wavelet transform to filter small target image, and threshold is segmented combining histogram threshold segmentation. Experimental result shows the small target detection method not only can save shape character of small target but also can eliminate background color completely.
出处
《兵工自动化》
2004年第6期67-69,共3页
Ordnance Industry Automation
基金
国防重点实验室武器装备预研-图象信息处理与特征提取的新变换技术研究
关键词
小目标检测
小波变换
小波遗波
阁值分割
Small target detection
Wavelet transform
Filter
Threshold segmentation
