The performance of automatic target detection is affected heavily by complex and unknown scene of natural ground images. In the natural ground images

the Rényi entropy and its occurrence probability

which is calculated through the pseudo Wigner-Ville distribution (PWVD) using two dimension window

has a statistical property based on exponential function. And the statistical property will be changed by the appearing of man-made targets. On that basis

a novel method of saliency map generation and target detection based on Rényi entropy is proposed. At first

the image of Rényi entropy is smoothed by the average filter. Then

by the subtraction of fore-and-aft filter images

the image of residual Rényi entropy can be obtained. The saliency map can be obtained by Gaussian filter. Finally

target detection is completed when the saliency map is segmented by a simple and convenient threshold method. Experimental results demonstrate our method can detect the military targets from complex ground scene effectively. In the test of 14 targets in 8 images

the detection probability and false alarm probability of our method are 100% and not more than 7.1% respectively.