Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: SORTING AGRICULTURAL MATERIALS FOR DEFECTS USING IMAGING AND PHYSICAL METHODS

Location: Foodborne Toxin Detection and Prevention

Title: Sorting of in-shell Pistachio nuts from kernels using color imaging

Authors
item Haff, Ronald
item Pearson, Tom
item Toyofuku, Natsuko

Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: April 1, 2010
Publication Date: August 1, 2010
Citation: Haff, R.P., Pearson, T.C., Toyofuku, N. 2010. Sorting Of In-Shell Pistachio Nuts From Kernals Using Color Imaging. Applied Engineering in Agriculture.26(4):633-638. Available:http://asae.frymulti.com/toc_journals.asp?volume=26&issue=4&conf=aeaj&orgconf=aeaj2010

Interpretive Summary: Sorting pistachio kernels from in-shell pistachios currently requires automated and hand sorting, an expensive and labor-intensive two-stage process. This paper examines the feasibility and effectiveness of using color imaging as a basis for identifying both regular and small in-shell pistachio nuts from kernels in the pistachio nut process stream. Color imaging would provide the basis for a low cost, high speed sorting device. Two methods were used to distinguish the images of in-shell and small in-shell nuts from kernels. The first method was a discriminant analysis (DA) algorithm, and resulted in a 98.7% overall accuracy for separating regular in-shell pistachio nuts and kernels. The errors were exclusively false negatives (fn), where the algorithm classified an in-shell as a kernel. No errors were made in the other direction. Small in-shell pistachio nuts were harder to discriminate from kernels, resulting in an overall accuracy of 80.7%, with 1.99% fp and 17.31% fn. This was not surprising given the size and coloring similarity between kernels and small in-shell nuts. The second method used a k-nearest neighbors (knn) algorithm. The overall accuracy for knn was 96.2% for kernels and 95.8% for small in-shell nuts. With careful selection of features, color image-based sorting shows promise for providing a cheaper, high speed method for sorting kernels from in-shell pistachio nuts.

Technical Abstract: Sorting pistachio kernels from in-shell pistachios currently requires automated and hand sorting, an expensive and labor-intensive two-stage process. This paper examines the feasibility and effectiveness of using color imaging as a basis for identifying both regular and small in-shell pistachio nuts from kernels in the pistachio nut process stream. Color imaging would provide the basis for a low cost, high speed sorting device. Two methods were used to distinguish the images of in-shell and small in-shell nuts from kernels. The first method was a discriminant analysis (DA) algorithm, and resulted in a 98.7% overall accuracy for separating regular in-shell pistachio nuts and kernels. The errors were exclusively false negatives (fn), no false positives (fp) were made. Small in-shell pistachio nuts were harder to discriminate from kernels, resulting in an overall accuracy of 80.7%, with 1.99% fp and 17.31% fn. The second method used a k-nearest neighbors (knn) algorithm. The overall accuracy for knn was 96.2% for kernels and 95.8% for small in-shell nuts. With careful selection of features this method shows promise for providing a cheaper, high speed method for sorting kernels from in-shell pistachio nuts.

Last Modified: 4/23/2014
Footer Content Back to Top of Page