基于二维正态云模型算法的红外图像弱小目标检测

Detection of Infrared Dim and Small Target Based on Two-dimensional Normal Cloud Model Algorithm

针对红外图像弱小目标检测的特点,采用二维正态云模型算法。首先利用一维云的特性建立二维云模型,由两个相互独立的一维云模型函数组成,目标像素的分布点为一个云滴,整个像素分布区域形成的云团反映了图像中目标的特性;接着依据目标判别条件函数来通过函数发生器产生正态云模型;最后在红外图像弱小目标检测误差函数下构造各云层的目标函数。实验仿真显示本文算法对红外图像弱小目标检测效果最好,能检测率高,虚警率低,耗时少。

Aiming at the characteristics of infrared small target detection, two-dimensional normal cloud model algorithms are used. Firstly, two-dimensional normal cloud is established with model function of one-dimensional cloud, composed of two independent one-dimensional cloud model function. The distribution of the target pixel is a point cloud droplets and the pixel distribution of the cloud formed in a region reflects the characteristics of the target image. Then normal cloud model is produced by a function generator objective based on determination condition function. Finally the cloud of the objective function is constructed underdetection error function. The simulation results show this algorithm is best for detection of dim and small target in infrared image, with high detection rate, low false rate, and less time.