To identify the intrinsic relationship between the distribution of seed composition and the change in seed vigour using a nondestructive testing method, this study chooses starch, a major component of corn seeds, as the research object. Combining the terahertz (THz) time-domain spectral reflection imaging technology with moving-window correlation coefficient imaging, a visualised map showing the spatial distribution of starch in corn seeds with different vigour degrees is constructed. Using the Zhengdan 958 corn variety as an example, a test is conducted to apply an artificial ageing method to prepare samples aged 0, 18, 36, 54 and 72 h. Then, the THz spectrometer is used to scan the reflection imaging attachments to obtain THz images of the samples. The THz image at 16.35 cm-1 was used as a benchmark, and the endosperm and seed embryo regions of the seeds were accurately extracted using the threshold segmentation method. By comparing the average absorbance of THz in different tissue regions, it can be obtained that the endosperm and seed embryo spectra differ significantly, and there is an obvious common absorption peak near 51.96 cm-1 for endosperm and starch pure substances. The moving window correlation coefficient method (window width 20, moving step 10) was applied to calculate the correlation coefficient between the terahertz time-domain spectrum of the seeds and the pure maize starch spectrum on a pixel-by-pixel basis and to construct a pseudo-colour heat map based on the correlation coefficient values and the coordinate information to visualise the maize seed starch distribution. Statistics on the percentage of pixel points with a correlation coefficient of >0.8 in the starch distribution map and covering five ageing stages and six spectral region windows lead to the conclusion that within the range of 29.83～67.36 cm-1 of the starch in the endosperm area of the seed exhibits an overall downward trend in vigour decline, presenting a positive correlation with vigour. Test results show that the THz time-domain spectral imaging technology can preliminarily realise the nondestructive detection of the spatial distribution characteristics of starch in corn seeds as seed vigour changes when it is used with the moving-window correlation coefficient pseudo-colour imaging analysis method. This technique provides a new perspective and method for studying the relationship between the chemical composition of seeds and their vitality and for non-destructively analysing changes in the vital activity of seeds and their own physiological and ecological patterns.