基于形态学重构的内河红外船舶目标检测

A Freshwater Infrared Ship Target Detection Base on Morphological Reconstruction

针对内河船舶红外图像的成像特点,提出了基于形态学重构的内河船舶红外图像目标检测算法,该算法先对原图像进行中值滤波,然后取滤波后的图像中亮度值最大的像素作为标记图像;原图像经顶帽变换,并进行迭代阈值分割后的图像作为掩模图像来进行形态学重构,从而实现红外船舶目标检测。实验结果表明,该算法能较好地检测出内河红外船舶目标。

A new detection method for freshwater ship infrared targets which based on morphological reconstruction was presented. Firstly, it operates a source picture with a mid-filter and gets the max pixel as a marker image. Secondly, the algorithm operates the source picture with top-hat transform and segment the picture using global threshold to get a mask image. Finally, the algorithm used the maker image and the mask image to reconstruct the target ship. Experiments based on this method shown good target detection results under complicated background.