一种基于区域背景预测的红外弱小目标检测方法

Small Target Detection Based on Mean Background Model in IR Images

提出了一种基于区域均值化背景模型的背景预测算法,用于红外弱小目标检测,算法通过“区域均值化背景模型”,来减小背景起伏对背景预测的影响,从而实现对背景更准确的预测,达到提高弱小目标检测性能的目的,算法适用于强对比度云层的空背景、具有人造干扰物的背景和空地背景的红外图像中,具有较强的抗噪音特性,是背景预测算法的一个重要扩展,针对实际红外图像的试验仿真表明,提出的算法是有效的。

For the small targets detection in IR images, a method of background prediction call mean background model (MnBM) is introduced. The MnBM also called “Local Mean Background Model”, which improve the performance of small targets detection by reducing the influence of background gurgitation. This model is applicable to the image, in which the background includes strong contrast cloud, man-made jams such as a building and ground scenes, and it can also restrain strong noise. This model is a significant development to the background prediction algorithm. Also, evolution with some real IR images proved the validity of the algorithm.