Nondestructive 2D cross-sectional visualization of a Mangosteen

Abstract

A nondestructive inspection of a fruit inside is an essential way for fruit grading. In this paper, we proposed a nondestructive 2D cross-sectional visualization of a Mangosteen using infrared. The infrared was selected because there is no effect against taste and consumer health. By experiments, the infrared in the frequency range of 480-1,025 nm is found to be appropriate for radiating through the Mangosteen. To obtain a 2D cross-sectional image of the Mangosteen, a series of passing through infrared ray was collected. The collection of profiles, a series of the infrared without a rotation along a cross-sectional axis, around the Mangosteen is used to reconstruction the image. There are a series of 56 infrared for each profile and totally 19 profiles for each Mangosteen. We use wavelet transform (Haar) to detect the peel width and determine the diameter of Mangosteen flesh. The diameter of Mangosteen flesh is used as a factor for preprocessing the profile. The 2D cross-sectional image of Mangosteen can be reconstructed by filtered back projection (FBP) algorithm. From experiments using a prototype that we developed, the result was shown that we can inspect the Mangosteen by 2D cross-sectional visualization of the Mangosteen via the nondestructive method. 70% of the classification rate of Mangosteen hardness is shown when using this frequency range.