<p>Through-Wall Imaging (TWI) based on conventional Compressive Sensing (CS) employs random projection matrices to compress the received measurements. However, the high mutual coherence of measurement matrices limits the effectiveness of random projections and may degrade reconstruction performance. Furthermore, Conventional iterative sparse reconstruction algorithms are computationally expensive, and the use of random projection matrices introduces additional receiver hardware complexity. This paper proposes a Parallel Analytical Compressive Sensing (PACS) framework that employs critical-equation selection to reduce the number of effective measurements without requiring random projections. Subsequently, Singular Value Decomposition (SVD) is applied to the measurement matrix, yielding a transformed measurement matrix whose columns are mutually orthogonal. For multi-target scenarios, a Target Categorization (TC) algorithm associates the local maxima of the received signals with individual targets, enabling independent closed-form reconstruction for each target. The proposed PACS framework is validated using two FDTD-based through-wall imaging scenarios involving targets with different reflectivity characteristics. Its performance is compared with Linear Programming (LP) and Alternating Direction Method of Multipliers (ADMM) at compression ratios (CRs) of 5%, 2%, and 1%, while PACS operates at a CR of 1.5%. The results demonstrate that PACS achieves reliable target detection and localization with competitive target-to-clutter ratios, demonstrating reliable target detection and localization under extremely low measurement rates.</p>