Unintentional islanding in grid-connected photovoltaic inverters(GCPVI) poses a significant challenge to power system reliability and safety. This article introduces a novel islanding detection method that leverages the magnetic characteristics of the GCPVI system. The BH curve, which defines the relationship between the magnetic flux density(B) and the magnetic field strength(H), is derived from the voltage across the inverter-side and grid-side inductors, and the current flowing through them. These BH curves are obtained for each cycle of the measured signals and analysed over successive cycles to calculate the alienation coefficient and cumulative index. The computed coefficients and indices form a time series vector, referred to as the islanding index. This index is compared against a threshold to detect unintentional islanding, even in the non-detection zone(NDZ). The proposed algorithm is experimentally validated on a single-phase hardware-based grid-connected inverter driven by bipolar pulse-width modulation. The measured voltage and current samples of the both side inductors are transmitted to a micro controller for real-time analysis. Using these samples, the method effectively distinguishes islanding from non-islanding events, such as load switching and distributed generation(DG) tripping, within a shorter time frame, adhering to international standards.