基于风险估计的弱小目标恒虚警检测

复杂背景下的红外弱小目标检测是目标探测领域的一个难题。为了有效抑制复杂背景的干扰,降低复杂背景所带来的虚警,提出了一种基于风险估计的恒虚警检测方法。首先,分析了背景分布的统计方法,对传统的正态分布统计方法予以改进,对滤波后的图像做局部灰度的分布统计,从而更准确地描述背景图像的分布规律。然后,在传统的恒虚警算法中加入风险估计,将背景复杂度作为风险估计判断依据,利用风险估计自适应调整分割阈值,从而达到抑制背景干扰、减少虚警和误判的目的。最后,实验结果表明:该算法可以显著减少复杂背景造成的虚警,并保证能够有效探测出弱小目标。

Constant false alarm rate algorithm for the dim-small target detection based on the estimated risk

Infrared small target detecting technology under complex background is a difficult subject in target detection.In order to suppress background,an improved constant false alarm rate detection algorithm based on the estimated risk was presented for dim-small target detection under background clutter.Firstly,the background distribution result calculated by the traditional statistical method was analyzed,according to which,the improved statistical method was proposed based on the part of the Ronbinson filtered image to describe the distribution of the background image more accurately.Next,the concept of the estimated risk was introduced into the constant false alarm rate method.Due to the estimated risk determined by the background complexity,the filtered image was adaptively segmented according to the estimated risk to suppress the background interference and reduce the false alarm rate.Finally,the experiment results show that the proposed method is more effective than the traditional constant false alarm rate method and can significantly reduce the false alarm caused by the complex background.