Abstract: Full-diameter core nuclear magnetic resonance (NMR) analysis is one of the key exploration and analysis techniques in unconventional oil and gas exploration. It provides continuous high-resolution information on rock core porosity, permeability, and fluid saturation. However, due to its large measurement sensitivity area, signals from different positions overlap, resulting in a significantly lower vertical resolution compared to instrument sampling. This limitation hinders its effectiveness in detecting thin interbedded reservoir. To improve the vertical resolution of the full-diameter core NMR measurements, the measured data were modeled as the convolution of the instrument's sensitivity area function and the core's real signal. High-resolution reconstruction of the original signal was achieved using global non-negative least squares, without changing the existing instrument structure or measurement mode. The feasibility of this method was validated through numerical simulations, physical experiments, and actual data analysis. Practical applications show that the high-resolution processed NMR porosity from well logging aligns more closely with gas-filled porosity. This method significantly improves the vertical resolution of full-diameter core NMR measurements, enhancing the detection capabilities for thin interbedded reservoirs.