Page Banner

United States Department of Agriculture

Agricultural Research Service

Related Topics

Research Project: Technologies for Quality Assessment of Fruits and Vegetables

Location: Sugarbeet and Bean Research

2011 Annual Report


1a.Objectives (from AD-416)
1) Develop a spectral imaging system for assessing and grading quality of pickling vegetables;.
2)Develop a prototype optical instrument for measuring the optical properties of horticultural and food products.


1b.Approach (from AD-416)
Hyperspectral imaging technique in reflectance and transmittance modes will be used for assessing both internal and external quality characteristics of pickling cucumbers and pickles. Algorithms will be developed for sorting and grading these products for quality and defect. Effective wavebands will be identified for efficient detection of product internal defect. Further, light source configuration will be investigated for more efficient assessment of internal quality attributes and defect. Moreover, a computer program will be developed for acquiring and processing spectral scattering profiles from fruit and other foods to determine their spectral absorption and scattering properties. Optimization of the instrument design will be conducted to improve the acquisition of spectral scattering images from fruit samples. Integration of the program into the prototype optical property analyzing instrument will be performed so that the prototoype can automatically acquire and process scattering image data and determine the optical properties.


3.Progress Report

Research was conducted to determine the optical absorption and scattering properties of normal and internally-damaged pickling cucumbers, so that a more effective quality detection technique can be developed. Fifty normal or defect-free pickling cucumbers were first scanned by a hyperspectral imaging system to acquire spectral scattering images for the wavelengths of 700-1,000 nm. Thereafter, the samples were subjected to rolling under a 10-kg load for 30 s to induce mechanical injury to the internal tissue. Spectral scattering images were again acquired for the injured cucumbers one hour and 24 hours after the mechanical stress treatment. After completion of the image acquisition, spectral absorption and scattering spectra for 700-1,000 nm were computed for both normal and defective pickling cucumbers. Statistical analyses were performed to determine how mechanical injury had affected the optical properties of the cucumbers. It was found that after the mechanical stress treatment, values of the absorption spectra for the cucumbers decreased, whereas an opposite trend was observed for the scattering spectra. The bruised cucumbers had significantly higher values in scattering, compared with normal cucumbers. The differences in absorption or scattering for the cucumbers that had been kept for one hour and 24 hours after the mechanical treatment were not significant. This research showed that by enhancing the scattering features can lead to more effective detection of internal defect in pickling cucumbers.

An online hyperspectral imaging system operated in simultaneous reflectance (450-750 nm) and transmittance (750-1,000 nm) modes was used to acquire hyperspectral reflectance and transmittance images for normal (pest-free) and fruit fly-infested pickling cucumbers. Experiments were conducted for 156 pest-free and 173 pest-infested pickling cucumbers of three size classes with the diameter ranging between 15 mm and 30 mm. Because of large size variations in the test cucumber samples, a mathematical method was proposed for correcting the fruit size effect on the hyperspectral image data. A discriminant analysis algorithm was developed to segregate normal cucumbers from the pest-infested cucumbers. The three sensing modes (i.e., reflectance, transmittance, and their combination) were compared for infestation detection. Results showed that the transmittance mode was able to achieve an overall detection accuracy of 88% versus 82% for the reflectance mode. The combination of reflectance and transmittance had an overall detection accuracy of 89%. These results compared favorably to manual inspection, which had an overall detection accuracy of 75%. Manual inspection accuracy decreased significantly as the size of cucumbers decreased. This research demonstrated that hyperspectral imaging technology can be used for detecting pest infestation in pickling cucumbers. Project progress was monitored via meetings, emails and joint sessions for planning, execution and analysis of research with the collaborating researchers.


Last Modified: 7/25/2014
Footer Content Back to Top of Page