Research on rapid determination methods for main compositions and sensory quality of pumpkins based on hyperspectral imaging technology.

Rapid detection of pumpkin quality is of great significance for pumpkin production and breeding. In this study, hyperspectral imaging technology was utilized to facilitate the rapid detection of moisture content, starch content, and sensory quality in pumpkins, as well as to investigate their distribution within the pumpkin. The hyperspectral imaging data acquired from pumpkin slices was extracted and averaged. The models for moisture content and starch content in pumpkin built under Multiple Scatter Correction (MSC) pretreatment and Ridge regression proved to be the best ones, whose determination coefficients for cross-validation ( R 2 cv) were 0.968 and 0.869, and the root mean square error for cross-validation (RMSE cv) were 1.142 and 0.365, respectively. Based on the moisture and starch values of pumpkin slices predicted by these models, the sensory quality scores of pumpkin slices can be further estimated. The sensory quality evaluation equation of pumpkin has a correlation of 0.934 to the sensory quality score of pumpkin obtained from the cooking experiment. Additionally, distribution maps summarizing the moisture, starch, and sensory quality of the pumpkin slices were generated, which could well reflect the spatial distribution characteristics of pumpkin quality indexes. • Determination methods for pumpkin qualities based on hyperspectral imaging was proposed. • The HSI models for pumpkin compositions were constructed by using MSC and ridge regression. • Sensory quality can be estimated based on predicted values of pumpkin HSI models. • The distribution of compositions and sensory quality of pumpkin can be visualized with HSI. [ABSTRACT FROM AUTHOR]