Shrubs are highly sensitive to habitat and climate change, and its tree-ring data plays an important role in understanding regional environmental evolution, global climate change and environmental protection. Due to the influence of heredity and extreme environment during shrub growth, the growth rings often have serious eccentricity and irregular growth, which makes it difficult for the tree-ring width measurement by professional tree-ring analysis software to accurately reflect the overall radial growth information. In order to explore the measuring indexes and methods of tree-ring, which are suitable for study on shrub dendrochronology in cold and arid regions, this study takes Tamarix ramosissima Ledeb., a common desert plant in this region, as the study object. Based on U-net deep learning method, we created the training model of Tamarix ramosissima tree-ring extraction, which would automatically obtain each year tree-ring springwood area of scanning disc. After tree-ring scanning image and semantic segmentation image were processed by GIS assigning measurement coordinates and ENVI image processing, tree-ring growth rate (TRGR), tree-ring width (TRW), and basal area increment (BAI) of tamarix ramosissima were measured by using GIS editing and measurement tools. Based on Timesat Savitzky-Golay (S-G) filtering time series fitting, seven key parameters of seasonal curve of normalized difference vegetation index (NDVI) were obtained at point-scale, including peak value, rate of increase at the beginning of the season, large seasonal integral and value for the end of the season etc. The correlation between TRGR, TRW, BAI and the key parameters of NDVI seasonal curve were analyzed to test the rationality and to analyze the advantages of TRGR index. The results show that there are high positive correlations between TRGR index of Tamarix ramosissima and the key parameters of point-scale NDVI seasonal curve. The highest correlation coefficient is 0.98.TRGR index can not only reflect the overall radial growth of tree-ring, but also effectively remove trend of diameter at breast height (DBH) in tree-ring index and highlight the index's environmental trend, which is helpful to solve the problem of index detrending in dendrochronology. Compared with TRW and BAI, TRGR index has obvious advantages and it can be used as a new index for shrub dendrochronology. The results provide evidence for NDVI reconstruction from tree-ring information of tamarix ramosissima or vice versa. Research methods have more diverse means than the professional tree-ring analysis system measurement. The results can promote the development of shrub dendrochronology in cold and arid regions.