Abstract: To achieve efficient and rapid detection of internal defects in mini-light emitting diodes (mini-LED) on wafers, a hyperspectral imaging detection method based on semiconductor photoluminescence was adopted. Theoretical analysis and experimental verification were conducted, obtaining spectral and imaging data from both normal and abnormal mini-LED samples under various test conditions. Results show that in normal samples, the covariance between photoluminescence spectra and electroluminescence spectra data exceeds 0.977, indicating good spectral consistency. Therefore, photoluminescence spectra can be used to characterize electroluminescence spectra in normal samples. As for abnormal samples, their photoluminescence spectra all exhibit significant yellow-band emission peaks, thus enabling their effective identification. The hyperspectral detection method can not only perform whole-wafer imaging but also collect and analyze the spectra from any local area, thereby achieving comprehensive, efficient, and rapid quality inspection of mini-LED chips. This study provides valuable references for future applications of hyperspectral imaging in mass detection and analysis of semiconductor wafers.