基于复合算子的红外复原图像弱小目标检测算法  

Dim Target Detection Algorithm from Restored Infrared Images Based on Compound Operator

作　　者：杜瀚宇 朱伟[1,2] 白俊奇[1,2] 席灿江[1] 

机构地区：[1]中国电子科技集团公司第二十八研究所,南京210007 [2]空中交通管理系统与技术国家重点实验室,南京210007

出　　处：《指挥信息系统与技术》2017年第1期70-74,共5页Command Information System and Technology

摘　　要：基于马尔可夫约束的泊松分布最大似然法(MPML)的超分辨力图像复原技术可恢复红外图像的细节特征,提高图像分辨力,同时带来图像噪声增大的问题。针对采用传统弱小目标检测方法检测红外复原图像虚警率高的问题,基于局部二元模式(LBP)算子提出了包含邻域灰度均值比和信噪比门限以抑制滤波的局部复合算子,并提出了基于复合算子的红外复原图像弱小目标检测算法。该算法在基于现场可编程门阵列(FPGA)的硬件平台上实现难度低,实时性好。试验结果表明,该算法用于红外复原图像弱小目标检测时检测效果好,虚警率较低,场景适应性优。A super-resolution image restoration technology with the maximum likelihood algo-rithm based on Poisson distribution in constraint of Markov （MPML） can recovery the details of the infrared image and improve the image resolution, meanwhile bring a problem of increasing the image noise. Aimed at the problem of the high false alarm rate with the traditional dim target detection methods from the restored infrared images, a local compound operator for the suppres-sion filter including the average ratio of neighbour gray scale and the restrict of signal-to-noise ra-tio （SNR） is proposed based on the local binary pattern （LBP） operator. The algorithm can be realized on the hardware platform based on the field programmable gate array（FPGA）with lower difficulty and high real-time ability. Experimental results show that the algorithm has advantages of good detection effect, low false alarm rate and good adaptability to different scenes in the dim target detection from the infrared restored images.

关 键 词：红外复原图像 局部复合算子 弱小目标检测 现场可编程门阵列 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]