摘要
为了提高复杂背景下的弱小目标的检测精度,提出了基于差异结构描述符与自适应侧抑制的红外弱小目标检测技术。首先,引入非局部均值滤波,对红外图像进行预处理,降低噪声的干扰;随后,计算输入滤波红外图像的梯度值,引入奇异值分解方法,在梯度域内获取特征信息,从而确定主方向;为了能够根据红外图像的信息变化来自适应增强弱小目标与抑制背景,利用奇异值分解获取的主方向来计算侧抑制系数;并利用特征信息,构建差异结构描述符,联合改进的侧抑制系数,形成了一个自适应侧抑制滤波器,降低其对噪声的敏感性,以更好地对红外图像中的每个像素进行处理;最后,定义像素灰度补偿函数,完成弱小目标检测。实验结果显示:与当前红外弱小目标检测技术相比,在噪声与复杂背景情况下,所提算法能够准确完整地检测出弱小目标,具有更高的信杂比增益与背景抑制因子,呈现出更理想的ROC曲线。
In order to improve the detection precision of infrared dim target in complex background,an infrared image dim target detection algorithm based on differential structure descriptor and adaptive lateral suppression filter was designed in this paper. Firstly,the gradient value of the input infrared image was calculated,and the feature information in a gradient domain was obtained by introducing singular value decomposition method for determining the main direction. For adaptively enhancing the dim targets and suppressing background according to the information change of infrared image,the lateral suppression filter coefficient was calculated based on main direction obtained by singular value decomposition. Then the differential structure descriptor was built by using the feature information,and the adaptive lateral suppression filter was formed by combining the improved lateral suppression filter coefficient for reducing its sensitivity to noise and dealing with each pixel in the infrared image. Finally,the pixel value compensation function was defined to complete the dim target detection. Experimental results show that this algorithm has higher signal clutter ratio gain and background suppressor factor,as well as more ideal ROC curves under the complex background.
出处
《电子测量与仪器学报》
CSCD
北大核心
2018年第7期68-75,共8页
Journal of Electronic Measurement and Instrumentation
基金
国家自然科学基金(61671338)资助项目
关键词
红外弱小目标检测
差异结构描述符
侧抑制
主方向
灰度补偿函数
信杂比增益
infrared dim and small target detection
differential structure descriptor
lateral suppression filtering
main direction
gray compensation function
signal clutter ratio gain
