Counting Brown Planthoppers in Light-Trap Images Using Morphological Operations

Abstract

As brown plant hopper (BPH) is one of the most dangerous kinds of insect for rice plant, in recent years, there has been increased concern in counting them in light trap images to control their spread in order to reduce their damage on rice plant. This paper proposes an approach to counting BPHs in light trap images based on morphological operations. By applying these operations appropriately, combined with some noise removal techniques based on color, BPHs in the light trap images can be identified. In addition, it is common that the BPHs in the light-trap images are overlapped due to the layout of the light trap. Therefore, an approach to counting the overlapped BPHs based on their size is also introduced while the sequential region labeling algorithm is used to count the number of the separate BPHs. The experimental results show that our proposed approach is suited to identifying and counting the BPHs in images which are overlapped.