For infrared small target，the image usually contains little pixels.The intensity is dim and it′s hard to abstract object features.The article analyzes the two common feature abstracting algorithms,and proposes a new feature abstracting algorithm in frequency region based on particle filtering.The article analyzes the changes in the frequency domain after adding background noise to the target.It shows the feasibility of the feature abstracting in frequency region.The experiment results show that the frequency region feature abstracting can describe the target well and distinct the target from background well.It is more accurate and effective than the traditional particle filter method.