论文部分内容阅读
检测红外小目标常用的方法是先使若干帧图像叠加达到能量积累的效果,能量积累达到一定程度后,目标因能量较高而凸现出来。这种方法比较适合于对加性噪声不敏感或信噪比较高的情况。实验表明,当图像信噪比较低、对比度很差时不宜采用该方法。针对低信噪比小目标的检测问题,提出了两种方法:在天水线以下图像内相邻行、列分别差分结合Kapur最大熵分割目标的检测方法;在时域内用模板平滑结合Kapur最大熵分割目标的检测方法。实验证明,这两种方法简单易行,效果较好。
The common method of detecting infrared small target is to make the images of several frames overlap to achieve the effect of energy accumulation. After the energy accumulation reaches a certain level, the target emerges due to the higher energy. This method is more suitable for additive noise is not sensitive or high signal to noise ratio. Experiments show that when the image signal to noise ratio is low, the contrast is not good when using this method. In order to solve the problem of low signal-to-noise small target detection, two methods are proposed: the detection method of differential combination of Kapur maximum entropy segmentation targets in adjacent rows and columns within the image below the Tianshui line, the template smoothing combined with Kapur maximum entropy Split target detection method. Experiments show that these two methods are simple and easy to use, the effect is better.