An aerial surveillance method is proposed for a predefined single object tracking. The algorithm takes advantage of template matching to be able to track the selected object in video sequences. The selected templates are chosen from the images by utilizing gradient operation on the Gabor Wavelet representations. The algorithm achieves the tracking task even in the cases where i) the observer is moving, ii) the target object is moving, iii) video record is low quality with noise and iv) there are high illumination changes.