Potential of hyperspectral imaging technology in monitoring moisture content of desiccated coconut during novel sequential hybrid drying

Desiccated coconuts were dried using a novel sequential infrared-infrared and convective dryer (IR-IRCD). It was observed that the falling rate period happened twice during the first stage of infrared drying (IR) and the later stage of combined infrared-convective (IRCD) drying. Combined (IRCD) drying speeds up the drying time by increasing the drying rate compared to IR drying alone. An in-house hyperspectral imaging system with a push-broom technique under a visible-near-infrared (421–960 nm) range was used to capture images of desiccated coconut samples. The quantitative correlation between spectral reflectance and the measured moisture content at drying time intervals was developed using partial least square regression (PLSR). The effect of multiplicative scatter correction (MSC) preprocessing on the original spectral reflectance and model performance was also investigated. Overall, based on full wavelengths, the PLSR model for calibration and validation with a preprocessed MSC exhibited higher accuracy (R2 > 0.97) compared to the original spectral reflectance (R2 < 0.70). The highest R2 close to 1 and the lowest root means square error (RMSE) of the PLSR model were achieved on moisture content prediction of desiccated coconut during drying. The color distribution map provides evidence that a novel sequential (IR-IRCD) drying with a combination of hyperspectral imaging systems could have great potential in enhancing drying efficiency as well as monitoring the moisture content of desiccated coconut through a non-invasive approach.

Text 123974.pdf - Published Version Restricted to Repository staff only Download (1MB) | Request a copy Official URL or Download Paper: https://link.springer.com/article/10.1007/s40003-0...

Actions (login required)

View Item