A novel algorithm which exploits aspectual invariance and local contrast to discrim- inate landmines from other man-made targets and natural clutter in Synthetic Aperture Radar (SAR) image is proposed in this paper. Firstly, Azimuth Scattering Entropy (ASE), which mea- sures the characteristic of aspectual invariance, is extracted as a detection feature. The detection threshold is then acquired from the histogram of the ASE image with an auto-adaptation method. The ultimate detection result is obtained by fusing results from aspectual invariance detection and Constant False Alarm Rate (CFAR) detection of the full aperture image. The algorithm con- cerns with the use of prior information of the to-be-detected targets and provides new thinking for detection scheme in SAR image. The results of experimental data demonstrate the effectiveness and practicability of the proposed algorithm.