红外图像弱小点目标检测技术研究

Research on Dim Point Moving Target Detection in Infrared Image

针对低信噪比、背景和噪声干扰严重的红外图像中运动点目标的检测问题,提出了一种背景预测和目标轨迹搜索相结合的高效检测算法。该算法首先对序列图像进行高通滤波,然后对滤波后的图像进行背景预测,将原图像与背景预测图像相减获得残差图像,依据图像阀值分离出少量的候选目标点,再将含有候选目标的序列残差图像经过最大合并算法形成组合帧图像,最后利用目标运动的连续性和一致性在组合帧图像中搜索目标运动轨迹。

For detection of the dim point moving target in the infrared image which is ＇characterized by low SNR （signal noise ratio） and the serious background and noise disturbance, an effective detection algorithm was proposed, which consists of the background prediction and target track searching steps. The algorithm was applied to the high-pass filtering of sequence images, made the background prediction for the filtered images, and subtracted the original image and the background prediction image to obtain the residual images. A few of candidate targets were separated from the residual images according to some threshold. The sequence residual images containing candidate targets are synthesized to form a combination frame image by the maximum combination algorithm. The target track was detected from the combination frame image by utilizing the continuity and consistency of target movement.