基于形态学的低信噪比图像序列目标提取方法

针对光电探测所得低信噪比图像序列中目标的检测和识别,提出了一种充分应用形态学相关理论和手段的处理方法。应用形态开运算进行有效的背景估计,通过灰度形态学Top—Hat变换实现背景对消,多帧图像的有效积累则借助点目标的形态膨胀来达到,通过数据关联实现目标检测。详细介绍了形态操作过程的实现方法,并通过结构元素模板分解大大减少了运算量。试验结果验证了方法的高效性。

Target Extraction Method of Low SNR Image Sequences Based on Morphology

To detect and recognise target in low SNR image sequences acquired from electro-optical detection, a method which applied morphological theory and means was presented. Background was estimated by morphological opening operation and suppressed by gray-scale morphology Top-Hat transform. Morphological dilation of point target was adopted to accumulate multi- rame images effectively. Target detection was realized by data association.The morphological operation process was introduced.To reduce calculation complexity, the structure elements templates were disassembled. The results of the experiment verify it's efficiency.