Heavy metal pollution will affect the normal growth of crops, and quickly detecting the content of heavy metals in crops has become a problem to be investigated. The traditional detection of heavy metals in plants depends on chemical methods. Although it can realize the high accurate detection of heavy metal content, its operation process is laborious, and it cannot meet the requirements of the high throughput detection, let alone the in-situ micro detection of plant tissues under heavy metal stress. Raman spectroscopy has the advantages of non-destructive detection of molecular vibration information of solid, liquid and gas species, high spectral resolution and insensitive to water. Therefore, it is feasible to monitor the content of heavy metals in crops by Raman spectroscopy. Apple rootstock is the basis of apple seedling grafting, which can ensure the health of the apple tree and apple quality and yield in the later stage. The root of apple rootstock is polluted by heavy metals directly, which hinders its healthy growth and affects the stress resistance of apple seedlings. Therefore, studying the interaction mechanism between heavy metals and apple rootstock root is necessary. In this study, five groups of apple rootstocks under the stress of CuSO₄·5H₂O solution with different concentrations were investigated. Firstly, the Raman scattering spectra of apple rootstocks under different copper ion (Cu²⁺) stress gradients were collected, and the adaptive iterative reweighting partial least squares (air-PLS) and S-G smoothing method were applied to preprocess the obtained raw Raman spectrum data for removing the fluorescence effect and correcting the baseline. Secondly, partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) discriminant models were established to estimate the different heavy metal stress concentrations. Results showed that the accuracy of the SVM model and PLS-DA model could reach 100% and 96%, respectively, which is promising for predicting apple rootstocks’ heavy metal Cu stress situation; finally, the chemical imaging was mapped based on the characteristic Raman spectrum peaks at 1 096, 1 329, 1 605 and 2 937 cm^-1. It was illustrated that the Raman signal intensity increased first and then decreased with the increase of stress concentration in the exact wavenumber. These findings demonstrated the potential of micro-Raman scattering for measuring apple rootstock heavy stress, which provides anovel method for detecting heavy metal stress of crops.