[4-1015-D] Food Quality (1)

[4-1015-D-05] Quantification of Tofu microstructure by image analysis

*CHIANG WEN-HSIN¹, Yoshito SAITO¹, Kohei OGATA¹, Tetsuhito SUZUKI¹, Naoshi KONDO¹ (1. Graduate School of Agriculture, Kyoto University(Japan))

Keywords:Tofu, microstructure quantification, image analysis, stiffness

Coagulation of soymilk, a complicated process in the production of tofu, is a critical step that complex factors, such as coagulant concentration, cooking temperature, and coagulant type, involved may affect the physical properties of tofu. The microstructure of bio-aggregates is fundamental to their physical properties (Lawrence et al.,2017). Until today, however, there are few researches on the relationship between tofu physical property and microstructure and only qualitative descriptions of tofu curd have been recorded. The aim of this study was to quantitatively and objectively evaluate tofu microstructure using image analysis and verify the relationship between tofu microstructure and stiffness while varying coagulant concentration. The tofu samples were made with varying coagulant concentration. These samples were then photographed using SEM for imaging analysis. In this research, Harilick textural features were calculated to quantify the microstructure of tofu curd and selected by Principal Component Analysis (PCA). 3 geometric parameters (number of holes per area, size of holes(area), and porosity) and selected Haralick textural features were finally correlated with Young’s modulus to verify the relationship between tofu microstructure and stiffness. The proposed methods, Haralick texture analysis and microstructure quantification, have potential discrimination application in tofu SEM image. As coagulant concentration increased, the number of holes and sum average feature also increased, however porosity decreased. From these findings, it was observed that the number of holes, porosity and sum average feature are candidate features for tofu microstructure quantification of SEM images. Moreover, correlations between stiffness with porosity and sum average feature were negative and positive respectively. Although the two tendencies were observed, in the future, more samples with varying concentration are necessary to improve the results.