基于时域背景预测检测红外图像序列中的小目标

Small Target Detection Based on Temporal Predictions of Background in Infrared Image Sequences

研究存在背景噪声和干扰的情况下红外图像序列中弱小运动目标的检测问题。从背景出发,充分利用在图像序列中占绝大部分的背景像素来构造弱小目标的检测,并提出在时域上利用背景预测技术实现检测。通过编程在含有随机目标的实际红外序列图像中的应用,证明了算法的有效性和优越性。

This paper addresses the problem of detecting small, moving, low amplitude objects in infrared image sequences especially for the situation with moving nuisance objects and background noise. Small target is detected by taking full advantage of the information of the background pixels. And the technique of small target detection is put forward based on temporal predictions of the background in infrared image sequences. At last, the effectiveness of the technique is demonstrated by applying the algorithm to real world infrared（IR） image sequences with random targets.