Structured-illumination reflectance imaging coupled with spiral phase transform for bruise detection and three-dimensional geometry reconstruction of apples

Authors: LU, YUZHEN - Michigan State University; Lu, Renfu

Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 6/21/2017
Publication Date: 8/1/2017
Citation: Lu, Y., Lu, R. 2017. Structured-illumination reflectance imaging coupled with spiral phase transform for bruise detection and three-dimensional geometry reconstruction of apples. ASABE Annual International Meeting. Paper No. 201700584. doi.org/10.1117/12.22623121.
DOI: https://doi.org/10.1117/12.22623121

Interpretive Summary:

Technical Abstract: Structured-illumination reflectance imaging (SIRI) is a new, promising imaging technique with enhanced, versatile capabilities for quality evaluation of food products. SIRI enables simultaneous acquisition of higher-contrast/resolution and better depth-controlled intensity and phase images for detection of surface or subsurface features of samples as well as reconstruction of the three-dimensional (3-D) geometry. The technique, however, requires multiple image acquisitions with subsequent demodulation, which is a major bottleneck in using it for real-time food quality inspection. This study was, therefore, aimed at providing the mathematical framework and experimental validation of an efficient image demodulation method, called spiral phase transform (SPT), for extracting alternating component (AC), direct component (DC), and phase images from acquired SIRI pattern images, to detect defects in, and reconstruct the 3-D geometry of, apple fruit. Pattern images for four varieties of apples that had been bruised by mechanical impact within a short time period were acquired, using an in-house developed SIRI system with two different configurations, from which AC, DC and phase images were extracted by using the proposed SPT technique. These demodulated images were then used for detecting bruises in apples and for reconstruction of the 3-D surface geometry of the fruit. Results showed that SPT was effective in demodulating the pattern images of apples; the AC and ratio (AC divided by DC) images clearly revealed fresh bruising in apples that could not, otherwise, be detected by conventional imaging technique under uniform illumination, and the ratio images yielded overall error rates of 11.7-14.2% for bruise detection. Using phase analysis for 3-D reconstruction, SIRI yielded relative measurement errors of 7.5% and 6.7% for two reference samples. The 3-D profiles for apple fruit were reconstructed by their phase difference images, which enabled effective identification of surface concavities (i.e., stem/calyx regions) of apples. This study has demonstrated that the SIRI technique, coupled with SPT, can enhance detecting fruit defects by means of analyzing demodulated AC and DC intensity images and phase difference images for 3-D geometry reconstruction.