For low contrast infrared images

if the brightness of point target is not obvious

and with influence of detector noise

the moving point target is hard to be recognized in infrared image. To solve this problem

image registration based on peak gray and the Fourier phase correlation algorithm is applied to two adjacent images acquired by detector

and this method can work out relative translation between the two images. Moving point object can be separated from the background. By absolute value of difference

the Gaussian low-pass filter algorithm and binarization operation on overlapping areas of the two images. The experimental results show that this method can effectively detect the moving point target in low contrast infrared image.