Abstract: 【Objective】To get systematic and complete information on important seed morpha of maize germplasm resources in China.【Method】Digital color images were scanned from 2 side faces each of 50 kernels each of 194 maize cultivars in China. The authors coded computer programs to automatically extract 27 individual morphologic kernel traits belonging to 4 categories of size, shape, texture and color. These quantitative traits are representative for completeness of kernel morpha and means of individual traits were averaged over kernels within cultivars. Measurements were statistically summarized and cluster-analyzed by SAS software to evaluate and classify the cultivars on individual or group morphologic traits.【Result】 A database was developed of 19300 color images of 9650 kernels. Ranges, means, standard deviations and variation coefficients for each trait were analyzed, and there were large, middle and small variation groups of the morphologic traits. Maize cultivars indicated cluster distributions in most situations. 36 unique germplasm materials were screened out.【Conclusion】This study provided rich primary data on and insights into morphologic seed aspects in kernel research and application in maize. Digital image process as a technique to measure crop kernel morpha showed promising application perspectives.

Key words: Maize, Kernel morpha, Germplasm evaluation, Image process, Cluster analysis