Abstract: In order to solve the problem of target strengh weakness of traditional target detection method based on singular value decomposition (SVD), an improved SVD algorithm was proposed for background suppression in dim and small infrared target detection. According to the nature of SVD, nonlinear transformation was adopted to improve the middle order part of image singular values for the largest contribution to the goal. And then, the other singular value was set to zero,finally the target image was obtained by reconstructing image. The experimental results show that the proposed method could preserve and enhance the target signal, improve the signal-to-clutter ratio and contrast ratio,and have good performance in complicated background suppression.