In thts paper, we propose a new algorithm which shows a burzed landmine clearly tn an underground zmage captured by a ground penetrattng radar (GPR). By ustng GPR, we could extract the burzed object stgnals from a raw underground data of GPR. In additzon, we could create a high resolutzon underground zmage which focuses on burzed objects. The extractzon of burzed object stgnals from raw GPR data is achzeved by ustng two methods which are a feature extractzon processing method of underground zmage and a ground adaptive mantpulatzon method which manipulates the sensor head of GPR along a ground surface. Moreover, we propose novel synthetic aperture radar processtng, which could be applied to an uneven ground surface of a minefield, for detecttng the landmine clearly. The high resolutzon underground Zmage is obtatned by the proposed synthetic aperture radar processtng which is referred to as ground adaptive synthetic aperture radar algorithm. We developed a landmine detection robot, in which the proposed methods were Zmplemented. To evaluate the proposed methods, experiments concerning landmine detectzon are done and experimental results tllustrate the valtdity of the system.