Abstract: As the most direct physical attribute of targets,geometric features are usually applied to the target discrimination and classification of SAR images.The processing flow for extracting geometric features from vehicle chips in high-resolution SAR images is presented.First,the vehicle chip is segmented in region segmentation step to form a binary image on which the geometric features are calculated.And then,the two methods,minimum enclosing rectangle method and Radon transform method,are introduced to estimate the length and width of vehicle targets.Finally,the real vehicle target SAR image data in MSTAR database are used to test the methods mentioned above and to validate the discriminating performance of these geometric features.