The adaptive Butterworth high pass filter(BHPF) is an effective method for detecting infrared small targets under sea-sky complex backgrounds.However,false alarms will inevitably exist when this approach is used to detect infrared video sequences images.Through estimating the local energy difference of small target regions in the filter images,which is directly caused by the different cutoff frequencies of BHPFs in consecutive two frames,an adaptive method for conducting the detection threshold of small targets in the filtered images was proposed here.Experimental results confirm that our algorithm can not only solve the problem of false alarms mentioned above,but also is effectively adapted to the conditions that the target is temporarily lost or the backgrounds are suddenly changed.