Image saliency based method for detecting automatic landing site of UAV (unmanned aerial vehicle)

The invention discloses an image saliency based method for detecting an automatic landing site of a UAV (unmanned aerial vehicle). The method comprises the following steps: (1), saliency detection of the UAV: salient region detection on acquired UAV flight image data I (I, j) from the angle of frequency domain transformation through discrete cosine transform, that is, an UAV flight sequence diagram I is input and a UAV salient image Inum_SIG is output; and (2), acquisition of the UAV image position: the UAV is positioned in the salient image, that is, the UAV salient image Inum_SIG and threshold delta are input, and UAV image sequence coordinates (xnum_out, ynum_out) are output. The image saliency based method for detecting the automatic landing site of the UAV has the advantages that the UAV can be detected and positioned rapidly, the accuracy and the reliability are high and the like.