Development of a multispectral structured-illumination reflectance imaging (SIRI) system and its application to bruise detection of apples

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

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/22/2017
Publication Date: 10/31/2017
Citation: Lu, Y., Lu, R. 2017. Development of a multispectral structured-illumination reflectance imaging (SIRI) system and its application to bruise detection of apples. Transactions of the ASABE. 60(4):1379-1389.

Interpretive Summary: Structured-illumination reflectance imaging (SIRI) is a new, promising imaging modality for enhancing quality detection of food. With this imaging technique, special illumination patterns, e.g., sinusoidal, are used to shine the sample, from which reflectance images are acquired. The acquired reflectance images are then demodulated to obtain direct component (DC), which corresponds to regular, uniform illumination, and amplitude component (AC), which can provide more detailed information of the sample. In previous reported studies for SIRI, a broadband light source is used for image acquisition, which can limit its ability of detecting food defects because they may be only sensitive to specific wavelengths or wavebands. In this research, a new multi-spectral SIRI system was developed, which allows to acquire spectral images with 10 or 20 nm increments over the spectral region of 650-950 nm. The system was used to detect fresh bruising for four varieties of apple, i.e., ‘Delicious’, ‘Royal Gala’, ‘Granny Smith’ and ‘Golden Delicious’. Image processing algorithms were developed for the ratio images of DC and AC images for detection of fresh bruises. It was found that the wavelengths between 710 nm and 830 nm were more useful for bruise detection. The SIRI system was able to achieve overall detection errors of 11.7-14.2% for the four apple varieties, while the DC images (i.e., under conventional uniform illumination) failed to identify bruises on these apples. This research showed that a multispectral SIRI system can be useful for enhanced detection of defects and other quality attributes of foods.

Technical Abstract: Structured-illumination reflectance imaging (SIRI) is a new, promising imaging modality for enhancing quality detection of food. A liquid-crystal tunable filter (LCTF)-based multispectral SIRI system was developed and used for selecting optimal wavebands to detect bruising in apples. Immediately after impact bruising, apples of ‘Delicious’, ‘Royal Gala’, ‘Granny Smith’ and ‘Golden Delicious’ were imaged by the system over the spectral region of 650-950 nm with 20 nm increments under sinusoidally-modulated illumination at the spatial frequency of 100 cycles/m. Each sample was subjected to two phase-shifted sinusoidal patterns of illumination with phase offsets of 0 and 120 degrees that were generated by a digital light projector. For comparison, spectral images were also captured under conventional uniform illumination. And a two-phase based demodulation approach was used to retrieve from the SIRI images amplitude component (AC) and direct component (DC) images, from which ratio images were obtained by dividing the AC by the DC images. It was found that the uniform illumination images failed to reveal the bruises in apples, whereas bruises were distinctly visible on the ratio images with contrast varying with wavelength. Principal component analysis showed that seven wavelengths from 710 to 830 nm were more relevant to bruise detection. A modified Otsu’s thresholding method based on the between-class variance was proposed for bruise segmentation from the ratio images at each of the seven wavelengths as well as the first principal component images, which achieved overall error rates of 11.7-14.2%. This study showed that the multispectral SIRI system over the wavelengths of 710-810 nm can provide an effective means for detecting bruises on apples. Further research is needed to develop a general algorithm for defect detection of apples, and upgrade the system towards real-time detection.