Abstract: Aiming at the problem of infrared aerospace small target's detecting under complex background, an intelligent algorithm was proposed based on the non-subsampled contourlet transform and BP neutral network. By using the morphologic top-hat transform which had adaptive structural element, some infrared background was suppressed. By defining the center vector of subband coefficients, the proposed method constructed the synthetical image at high frequency and suppressed the complex background further. Subsequently, taking pixel's gray level, horizontal gradient, vertical gradient, diagonal gradient, neighbour mean and neighbour variance as input character vector, a BP neutral network which had three layers was constructed by training and infrared small target was detected in the end by this network. The experimental results show that the method can not only realize the suppression for the infrared complex background effectively, but also detect the small target whose SNR is above 2 steadily.

Key words: infrared small target, background suppression, top-hat transform, non-subsampled contourlet, BP neutral network