|Zhang, Mengyun - University Of Georgia|
|Li, Changying - University Of Georgia|
|Takeda, Fumiomi - Fumi|
|Yang, Fuzeng - Northwest Agricultural & Forestry University|
Submitted to: American Society of Agricultural and Biological Engineers
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/1/2017
Publication Date: 11/1/2017
Citation: Zhang, M., Li, C., Takeda, F., Yang, F. 2017. Detection of internally bruised blueberries using hyperspectral transmittance imaging. American Society of Agricultural and Biological Engineers. 60(5):1489-1502.
Interpretive Summary: Currently, blueberry bruising is evaluated by either visual or tactile inspection or with fruit firmness measuring instruments. These methods are destructive, and time-consuming. The goal of this research was to develop a non-destructive approach for blueberry bruising detection and quantification by using near-infrared reflectance imaging technology in the transmittance mode. The spectra of bruised and healthy tissues were statistically separated. Support vector machine (SVM) classification of the spectra from the region of interest achieved 95% accuracy on the training set. Hyperspectral imaging system in the transmittance mode detected bruise damage expressed as water-soaked and discolored tissues. Therefore, the proposed approach and the bruise ratio index are effective in non-destructively detect and quantify blueberry bruising.
Technical Abstract: The consumption of the blueberries all over the world has been constantly increasing over the past three decades due to its health benefits. The internal bruising damage during harvesting and postharvest handling lower overall quality and cause significant economic losses. The main goal of this study was to nondestructively detect the internal bruised blueberries at early time after mechanical damage using hyperspectral transmittance imaging. A total of 600 blueberries were divided into 20 groups of four storage times (30 minutes, 3 hours, 12 hours and 24 hours), two storage temperatures (73°F and 40°F), and three treatments (stem bruise, equator bruise, and control). A near-infrared hyperspectral imaging system was implemented to acquire transmittance images at a range of 970 nm to 1400 nm. Images were acquired from three orientations (calyx-up, stem-up, and equator-up) for fruit in the control and stem bruise groups, and from four orientations (calyx-up, stem-up, equator-up, and equator-down) in the equator bruise group. Color images of sliced fruit were obtained as references. By comparing against the reference color images, the profiles of spatial and spectral intensities were evaluated to observe the effects of orientations and help extract regions-of-interest (ROIs) of bruised and healthy tissues. A support vector machine (SVM) classifier was trained and tested to classify pixels of bruised and healthy tissues. The classification maps were produced and bruise ratio was calculated to identify the bruised blueberries (bruise ratio > 25%). The average accuracy of blueberry identification was 94.83% using the orientation of stem end facing to the camera. The results indicated that it is feasible to detect bruised blueberry as early as 30 minutes after mechanical damage using hyperspectral transmittance imaging at both 73°F and 40°F storage conditions.