Achieving tunable focus of electromagnetic field energy at multiple target points is a critical challenge in the wireless power transfer (WPT) domain. In order to solve this problem, some techniques such as optimal constrained power focusing (OCPF) and time reversal (TR) have been proposed. The former presents limited practical applicability while the latter is noteworthy for its adaptive spatiotemporal synchronous focusing characteristics. However, the time reversal mirror (TRM) method necessitates intricate pretesting and has highly complex systems. In this study, we introduce a novel channel processing method, named channel extraction, selection, weighting, and reconstruction (CESWR), to attain balanced power distribution for multiple users, featuring low complexity, high computability, and rapid convergence. Unlike the traditional TR approach, our proposed method, based on channel correlation considerations, filters the channel impulse response (CIR) for multiple targets, dividing them into distinct characteristic and similar components for each target. This method ensures focused generation at both receiving ends while facilitating high-precision regulation of the peak voltage of the received signal. Furthermore, this study implements a rigorous examination of the linearity intrinsic to the proposed method, explicating a singular correspondence between the tuning of theoretical weights and the resultant outcomes. In order to verify the efficacy of this method, we construct a single-input multiple-output time-reversal cavity (SIMO-TRC) system. Subsequent experiments conducted for both loosely and tightly correlated models, provide invaluable insights. Evidently, in the loosely correlated model, the CESWR method exhibits proficiency in attaining a peak voltage ratio (PVR) of nearly 1.00 at the two receivers, with a minuscule numerical discrepancy of merely $8 \times {10^{ - 6}}$ mV. In stark contrast, under the tightly correlated model, the CESWR method demonstrates an enhanced ability to differentiate between two targets, thus offering a noticeable improvement over the classic single-target TR method.