Signal-domain registration for change detection in Time-Reversal SAR

Change detection is a well-studied area of radar, involving the coherent comparison of two or more images. This is an important tool used to aid in surveillance, battle damage assessment, and other monitoring applications. For many of these applications, registration of the processed SAR imagery is sufficient, as opposed to registration of the raw signal-domain data. Time reversal SAR, however, is an adaptive waveform transmission technique for SAR that requires coherent processing of the data before image formation algorithms are applied. Change detection for TR-SAR, then, must be performed as a pre-processing step. Ideally, the two change detection data collections would occur over the same trajectory and antenna positions, but this is not practical, and signal registration methods, rather than post-processing image registration, need to be developed and used to compensate for variations in the antenna's positions. In this work, we perform registration in the pre-processing stage, i.e., in the signal domain rather than in the image domain. We consider change detection in a mono-static TR-SAR scenario with the SAR antenna mounted on a linear railing, derive a model for the variations in trajectory positions between the two data collections, present our pre-processing methods, and verify them with experimental data.