摘要
图像灰度的奇异性和多重分形谱特征反映了图像内容的不同属性,并且可以用于在自然背景下对人造目标的检测。基于图像的多重分形谱,多重分形征和多重分形符合误差特征进行了模糊化处理,并且通过对80幅模拟图像的处理和分析获得了两个特征的模糊成员函数,提出了多传感器模糊特征综合函数和判别准则,对80幅样本图像进行目标检测的正确概率可以达到近90%。
The singularity and multifractal spectrum features reflect different attributes of the image content and can be used to detect man-made object in natural background.Based on multifractal spectrum features of image,the fuzzy approaches for multifractal spectrum feature and multifractal fitting error feature are discussed,and two fuzzy membership functions are obtained through analysis and processing of 80 simulation image and the composition of fuzzy features integrated functions is presented.The correct detection probability of man-made objects in 80 sample images is achieved nearly 90% by fuzzy multisensor criterion.
出处
《微计算机信息》
2012年第9期44-45,73,共3页
Control & Automation
基金
基金申请人:张军
基金资助项目名称:地面平台聚束旋转天线极化SAR静止小目标高分辨检测方法研究
基金颁发部门:国家自然科学基金
基金编号:684701
关键词
奇异性
多重分形谱
目标检测
模糊
singularity
multifractal spectrum
target detection
fuzzy
