Skip to main content
ARS Home » Midwest Area » East Lansing, Michigan » Sugarbeet and Bean Research » Research » Publications at this Location » Publication #250989

Title: Hyperspectral Waveband Selection for Internal Defect Detection of Pickling Cucumbers and Whole Pickles

Author
item ARIANA, DIWAN - Michigan State University
item Lu, Renfu

Submitted to: Computers and Electronics in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/29/2010
Publication Date: 8/21/2010
Citation: Ariana, D.P., Lu, R. 2010. Hyperspectral waveband selection for internal defect detection of pickling cucumbers and whole pickles. Computers and Electronics in Agriculture. 741(1):137-144.

Interpretive Summary: Pickling cucumbers are susceptible to internal defect, which will cause bloating problems during brining, resulting in low quality pickled products. Currently, machine vision systems are used for inspection of external features (i.e., size, shape, color, and/or surface defects) of pickling cucumbers or pickles but not for internal defects. Our recent research has shown that hyperspectral imaging in transmittance mode, which acquires images in narrow, contiguous wavebands over the spectral range of 740-1,000 nm, is useful for detection of internal defect in fresh cucumbers and whole pickles. However, speed and cost are the main constraining factor for implementation of the technique in commercial processing lines for evaluation of individual cucumbers and pickles. Therefore, this study was aimed at determining a set of optimal wavebands that would be useful for online detection of internal defect in pickling cucumbers and whole pickles. Hyperspectral images were acquired from 300 fresh pickling cucumbers and 280 whole pickles of defective and normal class. The optimal 2, 3, and 4-waveband sets were selected for defect detection of pickling cucumbers and pickles. The best overall classification accuracies of 94.7% for pickling cucumbers and 82.9% for pickles were achieved with the four-waveband sets of 745, 805, 965, and 985 nm and 745, 765, 885, and 965 nm, respectively. The appropriate waveband widths for pickling cucumbers and pickles were 20 nm and 40 nm respectively. The selected wavebands will be useful for further development of a cost effective inspection system for online sorting and grading of pickling cucumbers and pickles.

Technical Abstract: Hyperspectral imaging under transmittance mode has shown potential for detecting internal defect, however, the technique still cannot meet the on-line speed requirement because it needs to acquire and analyze a large amount of image data. This study was carried out to select important wavebands that can be used in further development of an on-line inspection system to detect internal defect in pickling cucumbers and whole pickles. Hyperspectral transmittance/reflectance images were acquired from 300 cucumbers and 280 whole pickles of defective and normal class, using a hyperspectral reflectance (400-740 nm)/transmittance (740-1,000 nm) imaging system. Optimal 2, 3, and 4-waveband subsets were determined by a branch and bound algorithm combined with the k-nearest neighbor classifier. Different waveband binnings were also compared to study the bandwidth requirement for each waveband combination. The highest classification accuracies of 94.7% and 82.9% were achieved using the optimal four-waveband sets of 745, 805, 965, and 985 nm at 20 nm bandwidth for cucumbers and of 745, 765, 885, and 965 nm at 40 nm bandwidth for whole pickles respectively. The selected waveband sets will be useful for online quality detection of pickling cucumbers and pickles.