Abstract: In recent years, due to the influence of natural and human factors, air pollution has become one of the serious problems in the world, which affects people’s life, physical and mental health and daily production. So this paper tried to address this problem by the monitoring and governance of haze. Analyzing the spectral characteristics of vegetation leaves before and after dust retention is an effective way to monitor haze. Plant leaves have the ability to adsorb airborne particles. By comparing the spectral curves of plant leaves before and after dust detention, it provides a scientific basis for the control of haze. In this study, the leaves of 10 common trees were collected for 26 consecutive days in October and November. Ten common tree species included Holly, Chinese rose, Platanus acerifolia, Ginkgo, Potentilla fruticosa, Lilac, Euonymus maackii, Lonicera maackii, Lagerstroemia indica and Magnolia. The hyperspectral image of leaf before and after dust retention was obtained by the hyperspectral imaging device developed by our research group. The hyperspectral image is segmented by thresholding method, and the average spectral curve of the whole leaf is extracted. The comparison of 10 kinds of leaves before and after dust retention was performed, excluding the influence of water, internal structure, and so on, and controlling dust to be the single variable. The changes of spectral characteristics of a single leaf, a variety of spectral characteristics of leaves, and the spectral differences before and after dust retention were analyzed. The basic trends of spectral changes of leaves before and after dust retention were obtained: 1) The variation trend of 10 kinds of leaves was basically the same, and the obvious spectrum changes of them before and after the dust retention were in the 500-550, 550-700 and 730-760 nm bands. 2) There was no change in the position of the red edge of the leaves before and after the dust retention, and the dust retention did not affect the internal physiological structure of the leaves. 3) The spectral response at the 765 nm band of Potentilla fruticosa and Platanus acerifolia differed greatly from other tree species; Euonymus maackii, Lonicera maackii, Lagerstroemia indica, and Magnolia had less sensitivity to dust retention in the 550-570 nm band. The spectral curves of 10 species of leaves were compared and analyzed from 3 aspects: Single leaf spectral characteristics, multiple leaf spectral characteristics, and spectral difference before and after dust retention. The spectral curves of 10 species of leaves were consistent with typical vegetation spectral curves; the spectral reflectance curves of 10 species changed obviously before and after dust retention, and the variation ranges were different, which indicated that different species had different sensitivity to dust retention. The study provides reference for monitoring atmospheric dustfall and controlling haze, and also provides theoretical support for further hyperspectral detection of typical components of haze.