A novel image processing approach for in-line monitoring of visual texture during shrimp drying

• Suggestion of a new image processing approach for extracting the image texture features during drying of foods.
• Eliminating the effects of samples structural and positional changes on the image texture features.
• Profound effect of drying medium temperature on the image texture features.
• Correlating the image texture features with the moisture ratio and geometrical attributes of samples.

In this study, a novel image processing approach based on the combination of Radon transform, pseudo Fourier–Mellin transform, and Fourier spectrum-based fractal dimension was proposed to exclude the undesirable effects of samples structural and positional changes on the image texture features. Images were obtained using a dryer equipped with perpendicular dual-view computer vision system (CVS) at hot air drying (HAD) temperatures of 50−90 °C and superheated steam drying (SSD) temperatures of 110−120 °C. Three drying medium velocities in the range of 1−2 m/s were adjusted for each drying temperature. Unlike the drying medium velocity, the drying medium temperature had a significant effect on the visual texture parameters of the shrimp batch. The zero-order and fractional conversion models along with a modified Arrhenius model were found as the best models for explaining the kinetics of the visual texture features and the temperature dependency of their constants, respectively. Eventually, the linear and cubic regression models satisfactorily correlated the image texture features with the moisture ratio and geometrical attributes of the samples, respectively.