[ Objective ] The aim was to deeply understand the diversity and variation characteristics of phenotypic traits of Lycium ruthenicum fruit, and the correlation mechanism between phenotype and active substances, so as to provide reference for the development and utilization of Lycium ruthenicum resources. [ Method ] Taking Lycium ruthenicum Murr.fruit as the research object, 23 phenotypic traits were extracted by automatic test analyzer, and 3 main active substance indexes were determined and analyzed. [ Result ] The results showed that the absolute value of variation coefficient of fruit phenotype and active substances of L.ruthenicum ranged from 2.92 % to 50.74 %, and the average variation coefficient was 19.75 %. There was rich diversity, among which the variation coefficients of roundness, correlation value and entropy value were 6.42 %, 2.92 % and 5.05 %, respectively. The difference between individuals was small, and the stability was high. The variation coefficients of anthocyanin, flavonoid and hue value were large, and the stability was low, and the selection potential was large. There were extremely significant and significant correlations between phenotypic traits and active substance traits. Among them, 143 pairs of correlation coefficients showed extremely significant levels ( p < 0.01 ), and 20 pairs of correlation coefficients showed significant levels ( p < 0.05 ). The level of active substance content can be judged by phenotypic characteristics ; the cumulative contribution rate of the first three principal components in principal component analysis was 74.893 %, which described the variation of fruit color, size and texture traits in turn. The characteristic vector values of phenotypic indexes such as redness, greenness, blueness, brightness, brightness, grayness, perimeter, area, length, width, angular distance and energy value were higher, which could be used as an important reference factor for fruit quality evaluation. [ Conclusion ] This study provides a new method for the early breeding of fruit quality from the methods of computer vision and quantitative analysis modeling, and also finds a new way for the rapid non-destructive detection technology of fruit quality.