红外运动目标的灰度分层检测算法

为了减少复杂背景下红外序列图像运动目标检测中的漏检与误检，采用了基于灰度分层的瞬时帧差检测算法。该算法根据红外图像的特点，使用自适应门限将当前帧图像分为两层，采用基于区域的像素间差分代替传统的基于空间坐标的像素间差分，通过对高灰度层检测时得到的高置信度的检测结果去引导对低灰度层的检测。实验表明，该算法可对多类多个运动目标进行有效检测，且具有较高的实时性与鲁棒性。

Detection algorithm for infrared moving objects based on gray delaminating

In order to decrease the missing rate and false rate simultaneity of the moving objects detection in the infrared image sequence with a complex background, an algorithm based on gray delaminating for temporal differencing is presented. According to the properties of the infrared image, the current image is segmented to two layers by an adaptive threshold. The proposed algorithm adopts the pixels differencing based on the local areas instead of the traditional pixels differencing based on the space coordiate, uses the detection results with high confidence in the high-gray level to induce the detection in the low-gray level. Experimental results show that the algorithm can be used for kinds of objects detection effectively, with high real-time and robustness performance.