Rapid and noninvasive sensory analyses of food products by hyperspectral imaging: Recent application developments

Abstract Background Sensory analysis is the evaluation of the signals received through senses of sight, sound, taste, smell, and touch. The traditional methods for the analysis of sensory properties including trained sensory panel, colourimeters and texture analysers are invasive, laborious, and small-scale procedures. Hyperspectral imaging systems (HSI) has emerged as a less time-consuming and non-destructive method to determine the sensory properties of a diverse range of food products. Scope and approach This review provides a comprehensive overview of recent application developments since 2010 for identifying sensory properties including colour, defects, texture, flavour, freshness, and maturity in various food products by HSI. Key findings and conclusions The Vis-NIR (400–1000 nm) hyperspectral imaging is most used for the assessment of sensory properties. Moreover, the commonly applied multivariate analysis in the sensory evaluation by HSI is linear regression algorithms (PLSR and MLR), but non-linear analysis achieves better performance for the prediction of sensory features. Regression models are applied for determining the texture, colour, flavour, and freshness of food products while classification models are used to detect the defects and maturity by which the most accurate results are obtained. Furthermore, the combination of algorithms and the integration of spectral and spatial information yield more accurate results in the assessment of sensory features. Although there are still some limitations to overcome for the improvement of the HSI system, studies on the application of HSI in the real-time monitoring of sensory properties have proved a great potential for industrial applications.