Accurate and rapid acquisition of soil salinity information under vegetation coverage can provide a basis for soil salinization management. The UAV remote sensing platform was used to obtain multispectral remote sensing images of the Shahao Canal Irrigation Area in the Hetao Irrigation District of Inner Mongolia in July, August and September 2019 and the sampling points were 0~10cm, 10~20cm, 20~40cm and 40~60cm depths of soil salt content (SSC). The spectral index was calculated through multi-spectral remote sensing images, and the normalized vegetation index (NDVI-2) was selected and brought into the pixel binary model (PDM) to calculate the vegetation coverage (FVC). The coverage was divided into four coverage levels: T1 (bare soil), T2 (low vegetation coverage), T3 (medium vegetation coverage), and T4 (high value coverage). The spectral index was screened by a full subset of variables, and partial least squares regression (PLSR) and extreme learning machine (ELM) were used to construct inversion models of soil salinity at various depths under different coverages. The research results showed that the accuracy of the inversion model under bare soil and high vegetation coverage was higher than the accuracy of the inversion model under low vegetation and medium vegetation coverage; comparing the accuracy of the two SSC inversion models, PLSR and ELM, the inversion accuracy of the ELM model was higher than that of the PLSR model; the best inversion depths under the coverage of T1, T2, T3 and T4 were 0~10cm,10~20cm,20~40cm, 20~40cm, respectively.The research result can provide an idea for UAV multi-spectral remote sensing to monitor soil salinization.