According to the data of the corn leaf spectra collected by a SVC hyper-spectrometer and the Cu contents measured synchronously in the corn leaves, the high frequency components of fifth layer wavelet decomposition (d5) were obtained by the "Db5" wavelet in Daubechies wavelets for the corn leaf spectra within the wavelength range from 350 to 2500nm, the fractal dimension of d5 could be calculated by the box dimension method, and the changing trend of fractal dimension of corn leaf spectrum under different Cu stress gradient was discussed based on a neighborhood change rate (α) of the fractal dimension, so that the spectral singularity parameters of d5 might be quantitatively calculated and analyzed such as the singular range, singular amplitude and the likes to distinguish the differences on weak information between the corn leaf spectra and the copper pollution levels of corn stressed by different copper ion (Cu²⁺) concentrations. The experimental results showed that the d5could precisely detect the weak spectral singularity information of corn under different Cu stress gradients, and realize the separation of hyperspectral signals of corn leaves at different pollution degrees; the d5fractal dimensions reduced firstly, then risen slowly and finally reached the peak value with the increase of pollution degree, among them the fractal dimension of Cu(100) was the minimum; the α values between CK(0) and Cu(100) were negative but positive in the other two stress gradient intervals, and the absolute value of α rates between Cu(100) and Cu(300) was the smallest. However the absolute value of α rates between Cu(300) and Cu(500) was the largest; it was validated that there was a strong correlation between the Cu content in corn leaf and the singular amplitude and fractal dimension through establishing the model on estimating Cu content in the leaf, the difference of Cu content in each leaf with different pollution degree reached a significant level (is 0.05), and its determination coefficient R²=0.9501. So the fractal dimension and singularity characteristics of spectral high frequency components could be used to diagnose effectively and analyze quantitatively the Cu pollution status of corn, and might provide some reference for monitoring heavy metal pollution of crops.