There is a great interest in developing optical techniques that have the capability of predicting quality attributes, safety parameters and authenticity in real-time assessment. Recently, hyperspectral imaging technique has been widely used for rapid and non-destructive inspection of various food products. Although the technique is currently in an early development stage, its potential is promising. Due to the extensive time needed for the processing of the large volumes of data, hyperspectral imaging technique cannot be directly implemented in an online system. However, selecting some feature wavelengths from hyperspectral images can be useful to develop a multispectral imaging system, which can meet the speed requirement of industrial production. Indeed, the success of multispectral imaging heavily depends on the effectiveness of hyperspectral imaging (HSI) for providing the feature wavelengths. If the high dimensionality of hyperspectral data can be reduced properly in order to design/form a low-cost multispectral imaging sensor based on some selected feature wavelengths for certain applications, the technique would certainly be incomparable for process monitoring and real-time inspection. This review first introduces the fundamental steps for selecting feature wavelengths from hyperspectral data and then describes the feature wavelengths derived from hyperspectral imaging applications to make a more effective and efficient multispectral real-time imaging system. It is anticipated that this review can act as a basis for researchers and industry for further development of online multispectral inspection system for quality, safety and authenticity of muscles food.
ASJC Scopus subject areas
- Food Science