摘要
针对传统的杂波抑制方法的不足,提出了基于推广结构张量的红外斑点小目标检测方法.首先利用一阶高斯对称偏导滤波器计算图像的方向场;然后把斑点小目标看成是圆对称模式,根据推广结构张量检测理论,选择相应的复值滤波器应用到方向场,并计算确定性图像;接着利用信息熵对图像复杂度进行评估,并根据评估结果确定分割阈值;最后对确定性图像进行分割和虚假目标删除,得到检测结果.实验结果表明:该算法能快速有效地检测复杂天空背景下的红外斑点小目标,检测概率达96%以上,明显优于传统方法.
The shortcomings of traditional methods of clutter suppression were considered. An infrared spot target detection method was presented based upon generalized structure tensor （GST）. The orientation field was computed using first symmetry derivative of 2D Gaussian. The spot small target was considered as circle symmetry pattern. According to GST detection theory the corresponding complex filter was applied to the orientation field and the certainty image was computed. The complex degree of image was evaluated using information entropy and the segmentation threshold was determined by the evaluation result. The detection results were obtained after segmenting the certainty image and false target deletion, the experiments were performed. The results show the proposed method could detect efficiently spot small target in complex sky background and the detection probability is up to 96%, which is superior to conventional methods.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2009年第7期10-13,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家高技术研究发展计划资助项目(2007AA12Z153)
关键词
红外斑点小目标检测
复杂背景
对称模式
推广结构张量
信息熵
infrared spot small target detection
complex background
symmetry pattern
generalized structure tensor
information entropy
