基于局域灰度概率分布的小目标检测方法

文章将红外图像目标与背景的灰度差转换为局部范围内的灰度概率差,在局域概率分布图像上实现小目标的检测,然后通过多帧累加判决的方式将真实目标从噪声点和伪目标中检测出来,仿真实验表明,该方法能有效检测像素占1~3个的小目标,同时具有良好的实时性能。

Dim Targets Detection Based on Local Gray Probability Analysis

In the paper, the difference between the gray level of targets and background is transferred into the difference of theirs local probability, and the target detection is realized in local probability image. Then, targets are selected from noise and false targets by multi frames cumulating. Number experiments show that this method can detect dim targets of 1~3 pixels effectively, at the same time it has good real time performance.