鉴于弱小目标检测所固有的难点及常用的单一分辨率下的检测方法还不能准确稳定地检测出目标,提出了一种弱小目标检测新方法.考虑到实际应用中的复杂背景和大量干扰噪声,运用数据融合技术,先对图像进行小波多分辨率分解,然后将不同分辨率下的子图进行最优加权平均融合来检测弱小目标.用实地拍摄的空中弱小目标红外和可见光图像分别进行实验验证,实验图像取256×256像素点阵大小,其中目标占10×10像素左右.结果表明该方法能够准确稳定地检测弱小目标,为后续的跟踪作了很好的铺垫.

Target detection in a sequence of image is difficult when this target is small, faint, and obscured. In this paper, a method based on wavelet multi-resolution analysis and data fusion technology is presented for faint-target detection. The detection method is put forward with the consideration of the practicality of the method, especially while the image involves the complex background and a lot of noise. First, the image is decomposed using wavelet transform. The second part of the algorithm performs a data fusion using the sub-images of different resolution. It is shown that the algorithm can detect the faint target with accuracy and stabilization in the experiment using real images. The size of these experimental images is 256×256 pixels, and the size of target in the image is under 10×10 pixels.