Abstract: 【Objective】 Cotton chlorophyll content and leaf area index are rapidly inferred by UAV using multispectral remote sensing data, which is crucial for predicting yield and making field management decisions. 【Methods】 The cotton in Aral, Xinjiang was taken as the research object, the influencing factors of cotton LAI and SPAD value were taken into consideration in the research, and different irrigation levels and different nitrogen levels were set to create a differentiated canopy structure. Vegetation indexes (VIs) were obtained by using a UAV equipped with multispectral sensors to obtain the canopy images of cotton during the main growth periods, and the mean values (MEA), variance (VAR), synergy (HOM), contrast (CON), dissimilarity (DIS), information (ENT), second-order moment (SEM), correlation (COR) and so on were obtained based on the second-order probabilistic statistical filtering (CO-occurrence measures) method (altogether 8 texture features TFs). The estimation models of cotton LAI and SPAD value based on spectral features, texture features and the combination of the two were established, and the differences were compared. 【Results】 (1) The results showed that the LAI and SPAD value of cotton increased first and then decreased during the whole growth period, and the maximum values of LAI and SPAD value of cotton were at the flowering stage. (2) Four VIs (NDVI, OSAVI, NDCI, RVI) and three TFs (CON, ENT, SEM) with high absolute correlation coefficients were screened out, and cotton LAI and SPAD value estimation models were constructed based on SVR, BPNN, RF, and the highest accuracy of the estimation model was the RF model. (3) The estimation effect of the three input variables on cotton LAI and SPAD value was VIs+TFs, VIs, and TFs in order of accuracy. The fused variables have the highest accuracy for the estimation model of cotton LAI and SPAD value (R²=0.97, RMSE=0.07, R²=0.91, RMSE=1.63). 【Conclusion】 RF algorithm model constructed by using VIs and TFs extracted from multi-spectral remote sensing images of UAV can estimate cotton LAI and SPAD value with high accuracy.