Extraction and Classifier Design for Image Recognition of Insect Pests on Field Crops

Feature extraction and the classifier design were the crucial parts for image recognition of insect pests on agriculture field crops. The hardware of the detection device for insect pests included the trapping, stunning and buffering unit, the even illumination unit, the scattering and transporting unit, and the image vision unit. The seven morphological features from binary images of the insect pests were extracted and normalized, such as area, perimeter, and complexity. The standard vector model library and the membership functions were established based on the feature mean and the feature standard deviation of the nine species of pests. The fuzzy decisions were analyzed based on the fuzzy principle of the minimum and maximum membership degree. The results showed that the fuzzy classifier could identify the nine species of pests that harmed seriously, and the recognition accuracy was over 86%.