It is rather important to detect and recognize those small targets caused by missiles or aircrafts in the series infrared images acquired by satellites. This paper investigates the algorithms for small target detection and proposes the reliable means for searching the point targets at high density clutter background. The method is that first use the background algorithms to acquire a more adaptive signal-to-noise ratio (SNR), then apply the image segmentation means to find the object-existing district and finally detect the point targets by using the recognition algorithms.