Structured-illumination reflectance imaging for enhanced detection of subsurface tissue bruising in apples

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

Submitted to: Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/1/2016
Publication Date: 7/17/2016
Citation: Li, R., Lu, R., Lu, Y. 2016. Structured-illumination reflectance imaging for enhanced detection of subsurface tissue bruising in apples. In: Proceedings of the ASABE Annual International Meeting, July 17-20, 2016, Orlando, FLorida. DOI: 10.13031/ids.2460153; paper no. 2460153.

Interpretive Summary: To reduce food loss and waste after harvest and meet the increasing consumer demand for food quality, defective food items need to be segregated from good ones. Machine vision is now widely used to automatically inspect food products for surface characteristics such as color, size, shape and/or blemishes. However, the technique is ineffective for detecting subsurface or near-surface defects such as bruises and other physiological disorder symptoms that are not visible from the surface. A new structured-illumination reflectance imaging (SIRI) technique was recently developed in our laboratory for enhancing quality detection of fruit and other food products. Instead of conventional diffuse or uniform illumination, the technique applies special patterns of illumination with a pre-selected spatial frequency for two or three phase shifts to product items, from which reflectance images are then acquired. The acquired images are subsequently demodulated or decomposed into direct and alternating components. The direct component (DC) images represent the ones under uniform illumination, while the alternating component (AC) images provides more detailed depth-resolved information about the structural characteristics of subsurface tissues. Experiments were conducted to evaluate the effectiveness of SIRI for detecting bruises on apples. Three levels of mechanical impact bruising were induced on three groups of 20 fruit each for ‘Golden Delicious’ and ‘Delicious’ varieties. Reflectance images were taken under the illumination of sinusoidal patterns at four spatial frequencies (0, 0.10, 0.15, and 0.25 cycles/mm) at the time intervals of less than 1 h, 1-4 h, and 24 h after bruising. An image processing algorithm was developed for identifying bruises on each apple. The extracted AC images were able to detect 70-100% bruises, whereas the DC images had poor detection results of 0-50%. The rate of bruise detection was, however, affected by bruise age; SIRI was most effective when the bruises were within 4-6 h after initial impact. SIRI was superior to uniform-illumination imaging technique for bruise detection, and it provides a new means for detecting subsurface defects like bruise. While implementation of SIRI technique requires acquiring multiple images, further improvements in image acquisition speed should greatly broaden the application of the technique in food quality and safety evaluation.

Technical Abstract: In this research, a novel method of fresh bruise detection was developed using a structured illumination reflectance imaging (SIRI) system. The SIRI system projects sinusoidal patterns of illumination onto samples, and image demodulation is then used to recover depth-specific information through varying the spatial frequency of the illumination pattern. The capability of SIRI was demonstrated through the detection of bruises on ‘Golden Delicious’ and ‘Delicious’ apples with varying levels of bruising. It was hypothesized that by confining the light penetration depth near the surface of each fruit, subsurface defects such as bruising should be more apparent under SIRI than conventional planar illumination imaging. Three 120 degree phase-shifted reflectance images were acquired from 60 fruit each of ‘Golden Delicious’ and ‘Delicious’ variety at each of the four spatial frequencies (i.e., 0, 0.10, 0.15 and 0.25 cycles/mm). The reflectance images acquired by the system were then demodulated into an alternating component (AC) and direct component (DC), where the AC contained depth-specific information and the DC image represented the average reflectance from the apple sample. A bruise detection algorithm was applied to the demodulated AC images and the performance of the system was determined. Detection rates under structured illumination were consistently higher than those obtained under planar illumination. Bruises were successfully detected by the SIRI system at rates ranging from 70-100%, while detection rates under conventional planar illumination ranged from 0-50%. SIRI is promising as a new imaging modality for detecting depth-specific defects in agricultural products.