Resampling Estimation based RPC Metadata Verification in Satellite Imagery

Recent advances in machine learning and computer vision have made it simple to manipulate a variety of media, including satellite images. Most the commercially available images go through process orthorectification remove potential distortions due terrain variations. This typically involves use rational polynomial coefficients (RPC) that geometrically remap pixels original image rectified image. paper proposes first method verify authenticity RPC metadata an orthorectified The steps include calculating Residual Discrete Fourier Transform (DFT) pattern from using linear predictor based residual spectral analysis comparing with Expected DFT is obtained associated If correct, then Residual-DFT (which represents data) Expected-Residual-DFT metadata) should be similar. We SSIM (Structural Similarity Index Metric) quantify similarity thereby if data has been tampered or not. Detailed experimental results demonstrate our achieves over 97% accuracy majority binary tampering detection tests.