Weak and Small Infrared Target Automatic Detection Based on Wavelet Transform

A new method about the infrared small target detection in complicated background is studied. According to the nice localization ability and directional characteristic of wavelet multi-resolution analysis, an effective target detection algorithm based on wavelet transform is proposed. Firstly, the original image is decomposed twice by wavelet transform to get its horizontal and vertical high frequency image. Based on the difference spreading properties of small target and noise wavelet transform modulus maxima, noise is suppressed and signal is enhanced by multiplying the high frequency part of same direction under difference scales. Secondly, according to the certain selection criterion, the horizontal and vertical high frequency coefficients are preserved separately in the direction of column and row. Finally, infrared small target is detected by fusing the horizontal and vertical preserved results. Experiments show that the algorithm works efficiently in infrared small target detection.