Skip to main content
ARS Home » Research » Publications at this Location » Publication #103895


item Lu, Renfu
item Chen, Yud

Submitted to: ASAE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 7/28/1999
Publication Date: N/A
Citation: N/A

Interpretive Summary: Bruises degrade the quality of apples and lower their value in the market. Therefore, development of effective and efficient techniques for detecting bruises in apples would mean increased value for the apple industry and better quality fruit for the consumer. Detecting bruises in apples nondestructively is very difficult because of the presence of the peel that prevents the bruised tissue from being detected. This study explores the potential of hyperspectral imaging for the detection of bruises in Red Delicious apples. Hyperspectral imaging combines the features of imaging and spectroscopy and, therefore, can greatly enhance our capabilities to detect subtle and/or minor features in a fruit. We developed image processing algorithms to create and analyze hyperspectral image data. Results showed that bruised tissue had lower relative reflectance than normal tissue in the near-infrared region between 700 nm and 900 nm. Using an imaging analysis method, called the principal component transformation, we were able to obtain superior images, allowing us to identify all those deliberately created bruises as well as most pre-existing ones. This research shows that hyperspectral imaging is capable of detecting apple bruises nondestructively and has the potential for fruit grading applications. The findings of this research should directly benefit the fruit industry that is still searching for techniques for the detection of bruises and other defects. It is also valuable to any researchers who are interested in developing nondestructive techniques for fruit quality evaluation.

Technical Abstract: This paper reports the results from a study using hyperspectral imaging to detect bruises in apples. A hyperspectral imaging system was developed for quality evaluation of agricultural products. Hyperspectral image data between 430 nm and 900 nm at a spectral resolution of 3.74 nm were acquired from Red Delicious apples immediately before and after bruising and then at t1, 8, 15, and 29 days. Bruised tissue was found to have lower relative reflectance than normal tissue in the near-infrared region between 700 nm and 900 nm. Using the principal component transform technique, superior images were obtained with well-defined bruise features, which could be clearly visualized and effectively classified using the image-processing algorithm described in the paper. This research indicates that hyperspectral imaging can be used effectively to detect bruises in apples with the potential for fruit grading applications.