Monitoring on heavy metal pollution and pollution degree of crops is a hot spot in hyperspectral remote sensing research. By conducting the potted-corn experiment stressed by copper, on the basis of the measured spectra and Cu2+ contents of corn leaves under Cu2+ stress with different concentrations, the research was carried out on the effective differentiating method of similar spectra according to the thinking that the corn leaves’spectra with different Cu2+ stress concentrations still have very high similarity and it is difficult to distinguish the different pollution degrees by using the traditional spectral measurement methods. The ED-T-DSGA spectral analysis model was proposed for spectral similarity measurement with the Euclidean distance (ED) and the tangent of spectral differential gradient angle (DSAG), which was proved to be feasible and effective in distinguishing the very small difference of extreme similarity spectra by comparing the traditional spectral measurement method, harmonic analysis (HA) technique and the results of spectral reconstruction of five HA decomposition times. Meanwhile, the ED-T-DSGA spectral analysis model can be used to measure spectral difference and monitor pollution degree of corn leaves stressed by different Cu2+concentrations. The experiment results showed that the greater the value of ED-T-DSGA spectral analysis model was, the greater the Cu2+ stress concentration was, which meant that corn was more seriously polluted by heavy metal copper. And some effective sub-band intervals such as the “yellow edge”, “red valley”, “red edge” and “near-peak B” were extracted by further study based on the ED-T-DSGA analysis model, these interval positions were the spectral responses of Cu2+ stress, the sensitive positions could be used as some favorable basis to monitor Cu2+ pollution degrees.