基于灰度级连通性的FLIR图像的目标检测方法

A New Target Detection Algorithm Based on Grayscale Level Connectivity in FLIR Image Sequences

通过对灰度级连通性及其算法的分析,文中给出目标k级成分区域提取算法,并针对前视红外(FLIR)图像序列提出一种目标检测的新方法。这种方法分为帧内和帧间处理:帧内处理利用目标k级成分区域提取算法实现单帧目标检测,帧间处理利用目标时间连通性实现序列图像间的目标关联和目标检测。仿真结果表明:在噪声和杂波干扰下,所提出的方法对红外图像序列弱小目标检测是有效和稳健的。

Aim. The level-k component zone of grayscale level connectivity was only mentioned in Ref. 4 by Braga-Neto et al. Braga-Neto et al proposed a method of target detection in forward-looking infrared （FLIR） image sequences without using level-k component zone. We now propose a new target detection algorithm based on extracting the level-k component zone of target in FLIR image sequences. In the full paper, we explain the new target detection algorithm in detail. In this abstract, we just add some pertinent remarks to listing the two topics of explanation. The first topic is： the grayscale level connectivity on complete lattice. In the first topic, we discuss three questions： （1） the level-k connectivity, （2） the algorithm based on grayscale level connectivity and （3） the algorithm for extracting level-k component zone of target. The second topic is： the detection of target in FLIR image sequences. In the second topic, we discuss intraframe and interframe processing. In intraframe processing, as explained in the full paper, we give a three-step procedure for detecting target in a single FLIR image. In interframe processing,target association and detection is implemented in terms of a dilation-based connectivity along the time direction in FLIR image sequences. Experimental results are given in a table in the full paper. These results indicate that using our new target detection algorithm raises the value of SNR to, on an average, 5. 9 times that of the SNR previously obtained. These results demonstrate the effectiveness and robustness of the proposed method for dim target detection in the clutter and noise background.