Location: Sugarbeet and Bean ResearchTitle: Histogram-based automatic thresholding for bruise detection of apples by structured-illumination reflectance imaging Author
|Lu, Yuzhen - Michigan State University|
Submitted to: Biosystems Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/15/2017
Publication Date: 6/12/2017
Citation: Lu, Y., Lu, R. 2017. Histogram-based automatic thresholding for bruise detection of apples by structured-illumination reflectance imaging. Biosystems Engineering. 160:30-41.
Interpretive Summary: Structured-illumination reflectance imaging (SIRI) is a technique that acquires reflectance images from an object under special patterns of illumination, instead of diffuse or uniform illumination that is commonly used for the other existing imaging techniques. SIRI can better control light penetration in biological tissues and enhance image resolutions and contrasts, and it thus offers some special advantages for food quality evaluation, such as detecting defects on fruit like bruises. Image segmentation of SIRI images is an important step for detecting bruise tissues from normal tissues. Conventional techniques are not effective for bruise segmentation. This research was, therefore, aimed at developing an effective methodology for automatic segmentation of bruises from normal tissues from the SIRI images of apple fruit. Two sets of apples with artificial created bruises and naturally occurred bruise were used in the study. Different image segmentation techniques were compared for their performance for detecting bruises. Based on the comparison results, a general image segmentation methodology was proposed. Results showed that the proposed methodology was able to achieve more than 90% accuracies for detecting artificial created fresh bruises (less than 24 hours) and between 77% and 85% for naturally occurred bruises. The methodology provides a simple and effective approach to bruise detection of apples, and it could also be used for detecting other types of defects on apples and other fruits.
Technical Abstract: Thresholding is an important step in the segmentation of image features, and the existing methods are not all effective when the image histogram exhibits a unimodal pattern, which is common in defect detection of fruit. This study was aimed at developing a general automatic thresholding methodology for fast and effective segmentation of bruises from the images acquired by structured-illumination reflectance imaging (SIRI). SIRI images, under sinusoidal patterns of illumination at a spatial frequency of 100 cycles/m, were acquired from 120 apple samples of four varieties with artificially created bruises and from another 40 apples with naturally occurred bruises. Subsequently, three sets of images, i.e., amplitude component (AC), direct component (DC) and ratio (i.e., dividing AC by DC), were derived from the original SIRI images. A unimodal thresholding method, called UNIMODE, was first applied to DC images for background removal, and then nine automatic thresholding techniques, including one unimodal and eight bimodal, were applied to the ratio images for bruise segmentation. It was found that severe over-segmentation occurred when using the bimodal thresholding methods, and this problem was mitigated by confining threshold selection to the lower part of the histogram that contained bruise information. Three bimodal thresholding techniques, i.e., INTERMODE (histogram valley emphasized), RIDLER (iterative thresholding), OTSU (clustering based) achieved the best bruise detection results with the overall accuracies of more than 90%. The overall detection results were further improved by integrating these techniques with the unimodal thresholding, due to reductions in the false positive error. The three bimodal thresholding techniques resulted in overall detection accuracies of 77%-85% for naturally occurred bruises. This study has showed that the proposed automatic thresholding methodology provides a simple and effective tool for bruise detection of apples.