The visible-near infrared spectra and leaf copper ion concentration data of maize leaves under different concentration gradient of Cu2+ stress were collected. Then through the short-time Fourier transform (STFT) time-frequency analysis technology，the energy amplitude response of the corn leaf spectrum under different concentrations of Cu2+ stress was studied. Furthermore，the amplitude parameters of the characteristic bands were extracted，and the partial least squares regression (PLSR) method was used to invert the leaf copper ion concentration. It was found that the peak of energy amplitude obtained by STFT transform of the corn leaf spectrum showed a decrease trend first and then increase trend with the increase of Cu2+ stress concentration gradient，and it continued to move to the short wave direction with the increase of Cu2+ concentration gradient. The peak bands of energy amplitude of different concentration gradients were selected as the characteristic bands，and the energy amplitude of the characteristic bands that varied with the frequency domain were used as parameters to establish a partial least square regression model of leaf copper ion concentration inversion. The PLSR model accuracy performed good，and the determination coefficient of R2 was 0.9863. The other two sets of plant data in the same cultivation period were selected as the verification data，and the same STFT transformation for the verification data. The established partial least square regression model was used to invert the leaf copper ion concentration of the two sets of verification data，and correlation analysis with the measured leaf copper ion concentration of the verification group was conducted. The leaf copper ion inversion accuracy R2 was 0.8806 and 0.7331，RMSE was 1.563μg/g and 2.619μg/g，respectively. The research result showed that the spectral time-frequency analysis method can be used for rapid and efficient detection of corn leaves under Cu2+ stress, and provided ideas for the monitoring of heavy metal stress in crops.