Faint target detection based on wavelet transform and data fusion technology

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.