논문 상세보기
pp.57-63
해충의 발생시기와 발생량에 대한 정확한 예찰정보는 해충의 효율적인 종합적 방제를 위하여 필수적으로 요구된다. 해충의 효율적인 발생 예찰조사를 위해 디지털 영상처리 알고리즘을 이용하여 벼농경지에서 주요 해충인 멸구류를 자동적으로 인식하고 밀도를 측정하도록 하였다. 야외경작지에서 촬영한 입력영상에 대해 구성인자분해과정, 탑헷(top-hat)변환, 역치적용, 최소/최대 필터링 등의 방법을 적용하여 벼 잎에 붙어 있는 멸구 개체를 인식하고 개체수를 헤아렸다. 평균인식율은 95.8%를 보였다. 또한 인지된 각 멸구류 개체 크기를 측정하여 멸구류의 연령분포 추정을 가능하게 하였다
Accurate forecasting of occurrence time and abundance of insect pests is essential for developing technology of integrated pest management system. Digital image processing algorithms were utilized to automatically recognize rice planthoppers which are major insect pests in the rice cultivation field and were subsequently used to estimate densities in the field for efficient forecasting of insect pests. To the images taken in the rice field, image decomposition, top-hat transformation, threshold, and minimum and maximum filter were implemented for patterning individually the brown planthopper specimens attached at the bottom area of rice stems. In average 95.8cio of images were correctly recognized for estimating densities by the developed system, and the recognition rate was higher than that obtained from direct observations by experienced observers. Furthermore, the size of the recognized specimens was measured and was used for estimating the age structure in the observed brown planthopper populations
