Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 1, 2002
Publication Date: September 1, 2002
Citation: Pearson, T.C., Young, R. 2002. Automated sorting of almonds with embedded shell by laser transmittance imaging. Applied Engineering in Agriculture. 18(5): 637-641. Interpretive Summary: One of the most pressing quality control problems facing the U.S. almond industry is that of shell fragments embedded into the kernel. Since almond are usually used as ingredients in candy and snack foods, consumers may not notice shell fragment embedded into an almond kernel. The incidence of shell fragments lodging into kernels has been increasing over the past several years. Thus, the industry is in need of a method for removing these kernels from the process stream. A prototype device has been constructed to automatically detect kernels with embedded shell fragments. This device images laser light that is transmitted through the kernel. If a shell fragment is present, it will appear as a very dark spot in the image. A computer algorithm was developed to detect these dark spots and activate an air valve to divert nuts with embedded shell from the process stream. Approximately 83% of the kernels with embedded shell were detected while 11% of the kernels with no embedded shell were incorrectly classified as having embedded shell. Kernels classified as having embedded shell can be chopped or sliced to remove the shell fragment. The sorting device has an inspection rate of approximately 40 kernels per second or 100 kg per hour.
Technical Abstract: A quality control problem for the almond industry is that of shell fragments becoming embedded in kernels during hulling and shelling. While embedded shell is rare, with only about 0.1% of shelled kernels exhibiting this problem, the incidence has been increasing over the past several years. The industry therefore needs a method to remove these kernels from the process stream. A prototype device has been constructed which images laser light transmitted through the kernel, automatically detecting and removing kernels with embedded shell fragments. A shell fragment blocks nearly all the transmitted light, forming a dark spot in the image which is detected by a computer algorithm. The computer then activates an air valve to divert the corresponding kernel from the process stream. The sorting device has an inspection rate of approximately 40 kernels per second (100 kg per hour). For a single pass sorting operation, approximately 83% of the kernels with embedded shell were detected and removed. Additionally, 11% of the clean kernels (no embedded shell) were incorrectly classified as having embedded shell and were also removed from the process stream. Running the rejects of the first sorting pass through the system a second time recovered approximately 46% of the previously rejected clean kernels.