Sugarcane plant height is a comprehensive reflection of soil, meteorology, hydrology and other factors, and it is also an important index for sugarcane growth monitoring and yield estimation. Based on the climate and weather conditions in South China, the dual-pol radar vegetation index (DPRVI) was obtained by preprocessing, matrix transformation and Cloude-Pottier decomposition of the time series Sentinel-1A data of 23 landscapes covering the whole growth period of sugarcane. The dynamic changes of this index and sugarcane growth parameters (plant height) with different growth periods of sugarcane were analyzed. Then four classic empirical regression models (linear, quadratic polynomial, exponential and logarithmic) were used to invert the height of sugarcane plants in different growth periods in the form of piecewise function to establish the best inversion model. Through the experimental results, it can be found that the fitting model had the highest correlation before the tillering stage, and the quadratic polynomial model had the best fitting effect. The determination coefficient R2 and the root mean square error were 0.882 and 0.118cm, respectively. Then the quadratic function model with the best inversion effect before tillering stage was verified and analyzed. The results showed that the determination coefficient R2 was up to 0.839, and the average absolute deviation was 7.4%, which indicated that the model could better guarantee the accuracy of predicting plant height by using DPRVI. Finally, DPRVI was compared and analyzed with other three classical remote sensing parameters. The results showed that the performance of DPRVI was better than that of the other three parameters. Through the analysis, it was concluded that DPRVI can better invert the change of plant height in the early growth stage of sugarcane, and the parameters of plant height can be used as a reference for agricultural management departments.