Low-porosity and low-permeability reservoirs are developed in the Enping Formation and Wenchang Formation in the Lufeng Sag, the Pearl River Mouth Basin. Due to their complex pore structure and strong heterogeneity, conventional porosity-permeability relationship models exhibit low accuracy in permeability calculation, making it difficult to meet the actual production needs such as reservoir effectiveness identification and productivity prediction. Starting from the generation principle of porosity spectrum in electrical imaging, and combining with core experiments and forward simulation results, the authors systematically compared the differences between nuclear magnetic resonance (NMR) T2 spectra and porosity spectrum generated by conventional methods, deeply analyzed the reasons for the shortcomings of conventional porosity spectrum in characterizing rock pore structure, and proposed an improved porosity spectrum construction method based on pore volume statistics. The improved porosity spectrum shows excellent consistency with the NMR T2 spectrum, which can more accurately reflect the pore structure characteristics of the reservoir. On this basis, the pseudo-NMR Schlumberger-Doll Research (SDR) model is used for reservoir permeability evaluation, which significantly improves the calculation accuracy over conventional methods. It can better reflect the advantage of high resolution compared to nuclear magnetic logging, providing reliable data support for fine reservoir description and productivity prediction.