The aim of this study is to compare and screen SPAD estimation models of rice canopy, and to provide a basis for inversing SPAD value of rice by unmanned aerial vehicle (UAV) multispectral remote sensing. This paper used UAV to obtain the canopy multispectral images of rice at jointing stage, heading stage and milky maturity stage. Seven commonly used vegetation indices were selected and three regression methods were used to establish the SPAD value inversion model of rice leaves. The results showed that the vegetation indices with the highest correlation coefficient with SPAD value of rice leaves were different at different growth stages. GNDVI was the highest at jointing stage, CI_Green was the highest at heading stage, and CI_Rededge was the highest at milky maturity stage. The heading stage was the best inversion stage of SPAD value, and the model had good modeling precision and estimation effect. According to the accuracy test, multivariate linear regression model had relatively high modeling accuracy, and partial least squares regression had the best estimation effect. Thus, this research provides a new technology to supervise the growth information of rice and other crops. The experimental results can provide methods and reference for the real-time and nondestructive monitoring of rice growth.