Application of Image Texture Analysis for Evaluation of X-Ray Images of Fungal-Infected Maize Kernels

Application of an expert system of X- ray micro computed tomography imaging for identification of Sitophilus oryzae infestation in stored rice grains

BACKGROUND

The plausibility of image texture analysis to assess X-ray images of S. oryzae-infested rice after variable storage days (fresh, 45, 90, 135, 180 and 225 days) was investigated using an X-ray micro computed tomography instrument. Subsequently, image acquisition, pre-processing, and the extraction of the image textural features was done using volume graphics VGL 2.2 software. Morphological features (radius, diameter, volume, compactness, sphericity, defect volume, and voids) were extracted from the x, y, and z views of the rice grain and used as inputs for principal component analysis (PCA).

RESULTS

Clear grouping was observed between the fresh, 45 and 225-day-old S. oryzae-infested rice grains with a classification accuracy of 88.34%. The voids (884 248.53 μm³ ) and defect volume distribution (137 428.04 μm³ ) were found to be the maximum in 225-day-old samples. The similarity or the distance indices values between fresh and 255-day-old S. oryzae-infested rice samples were found to be 35 038.08, which resulted in clear discrimination between different storage days in S. oryzae-infested rice grains.

CONCLUSION

This work contributes to the potential use of image texture analysis to aid in distinguishing S. oryzae-infested rice grains from fresh rice grains. © 2019 Society of Chemical Industry.