Target detection method based quadtree structure search and Hurst exponent

Abstract

After the features of Hurst exponent of the manmade target and natural background images are analyzed, a new algorithm is proposed that uses a quadtree structure search method and Hurst exponent computation. In order to extract the man-made objects from their background, the regions are firstly confirmed that probably including man-made objects by a quadtree structure search method. Then compute the Hurst exponent in the regions obtained above to confirm the region that includes the man-made object. Our method has been tested on images that consist of man-made objects and natural background and shows satisfactory performance.