Abstract: Accurate acquisition of pulse energy is very important in the field of atomic energy technology. In the existing energy characterization methods, the multiple voltage thresholds (MVT) sampling method introduces prior information of pulse signals and combines it with trigger times from multiple reference voltages to achieve direct energy characterization, without any pulse shaping techniques. It involves no extra hardware components and has high count rate, rendering it applicable for a range of applications, including oil detection, medical imaging, etc. However, in conventional MVT sampling process, to record trigger times from one reference voltage threshold would consume one time-to-digital converters (TDC) unit. Since a TDC unit is not resource friendly, it would be notable challenging to apply this strategy in a multi-channel scenario, due to the increased power and cost. To address this, a resource-efficient multiple voltage thresholds (REMVT) sampling method was proposed. REMVT harnesses the fact that trigger times from each reference voltage appear sequentially and at a particular order, making it possible to record all of them using only a single TDC unit. To achieve this, the trigger signaling from each reference threshold voltage was routed together and fed to a combinatorial logic circuit. This logic circuit consumes few resources and can alter signal levels if an effective signaling arrive. By this means, the trigger times from each reference voltage were sequentially mapped to different edges. Deciphering interval between these edges would generate sufficient information for pulse reconstruction and hence energy characterization. This “TDC chain reuse” technique significantly reduces resource consumption. For example, in a configuration with 4/8 thresholds, resource consumption is reduced by 75%/87.5%. Additionally, this chain reuse technique reduces measurement variations from using multiple chains, leading to fewer jitters in time measurement and, consequently, better energy characterization performance. With REMVT, the energy characterization errors of 79% pulses are within 20.5%. While with MVT, this value reduces to 72%. Using this REMVT, an energy resolution of 13.2% for 511 keV can be obtained under practical experiment conditions, which is only 1.1% worse than the results obtained from a general high-speed oscilloscope. In addition, a new event detection and pileup event recovery algorithm was proposed in this paper. This algorithm fully exploits the prior information of pulses, namely the sequential order of each trigger time and the respective interval among them, to accomplish reliable event detection and accurate pileup event recovery. Results demonstrate that 93% pileup events can be identified and recovered.