Restrain background clutter in infrared image is the first step before detecting and tracking. As infrared sensor is easily affected by atmosphere hot radiation, long distance and sensor noise, the detected targets in infrared images often present like small targets and drowned in noise. So it's difficulty to detect the small target in complex background, especially doing it with single frame. A new method of infrared small targets detection based on local third order moment is proposed. Firstly, the local third order moment of target and background are analyzed. The result is that the third order moment of target and background has distinctively difference in value. The local third order moment of background tends toward zero or less than zero, while that of target is greater than one. Secondly, the global normalization is carried out after the third order moment operation to single frame image. With this algorithm, the background magnitude is almost near zero, and the potential targets magnitude is near to one. So the background clutter in the infrared images is suppressed effectively. And the target can be detected with fixed threshold. At last, false target judgment criterion is used, which can eliminate remained big bright object boundary effectively. The experiment results proved that this arithmetic has robust performance to restrain different background clutter; even in the situation the background clutter have high grey-level than the small target.