Detection of damage caused by on soybean using novel imaging approaches based on computed tomography and image color analysis.

Soybean ( L.) is an important leguminous plant, in which pests trigger significant damage every year. Important members of this community are insects with piercing-sucking mouthpart, especially the southern green stinkbug, L.. This insect with its extraoral digestion causes visible alterations (morphological and color changes) in the seeds. We aimed to obtain precise information about the extent and nature of damage in soybeans caused by using nondestructive imaging methods. Two infestation conditions were applied: one with controlled numbers of pests (six insects/15 pods) and another with naturally occurring pests (samples collected from the apical part of the plant and samples from whole plants). An intact control group was also included, resulting in four treatment groups. Seed samples were analyzed by computed tomography (CT) and image color analysis under laboratory conditions. According to our CT findings, the damage caused by changed the radiodensity, volume, and shape (Solidity) of the soybean seeds during the pod-filling and maturing period. Radiodensity was significantly reduced in all three damaged categories compared to the intact sample; the mean radiodensity reduction range was 49-412 HU. The seed volume also decreased significantly (25%-80% decrease), with a threefold reduction for samples exposed to regulated damage compared to natural ones. The samples exposed to natural damage showed significant but minor reduction in solidity, while samples exposed to regulated damage showed a prominent decrease (~12%). Image color analysis showed that the damaged samples were well distinguishable, and the differences were statistically verifiable. The achieved data derived from our external and internal imaging approaches contribute to a better understanding of the internal chemical processes, and CT analysis helps to understand the alteration trends of the hidden structure of seeds caused by a pest. Our results can contribute to the development of a practically applicable system based on image analysis, which can identify lots damaged by insects.