基于类内一致性的红外背景弱目标检测方法

Infrared Background Weak Target Detection Method Based on Consistency in Class

红外背景下检测弱目标较为困难。为此,提出一种基于类内一致性的红外背景弱目标检测方法。定义类内一致性函数,通过直方图上下分割值分割红外图像,使用形态学运算处理图像,填充目标内部空隙以及连通断裂目标,进行多目标区域增长,根据分割出的若干目标形状、大小等信息确认最终目标。实验结果表明,该方法 3帧图像的检测时间分别为0.355 ms、0.363 ms、0.335 ms,优于Ostu方法和均值方法。

It is difficult to detect the weak target in infrared background. In order to solve this problem, this paper proposes an integrated background weak target detection method based on consistency in class. It defines the class consistency function, through the histogram and infrared image segmentation value to make segmentation, uses morphological operation to process images, fills target internal void and connects fracture target, makes multipurpose regional growth, according to the division of several target shape size information to confirm final goal. Experimental results show that the 3 frame image detection times of this method are 0.355 ms, 0.363 ms, 0.335 ms, are better than the Ostu method and average method.