Identification of the Whitebacked Planthopper (WBPH) on a light trap using digital image processing

Abstract

The whitebacked planthopper (WBPH) Sogatella furcifera (Horvath) (Hemiptera: Delphacidae) is an important rice pest in Asia that has causes severe yield loss. Under favorable conditions, the insect can completely destroy rice plants. WBPH occurs in both upland and wetland rice environments. Outbreaks of the insect pests are closely associated with insecticide misuse, especially during the early crop stages. The farmer has a problem in identify WBPH as there is also another planthopper that feeds on rice plants. Before, the farmer used a manual technique in identifying the WBPH. The technique is slow, less efficient, and have low count accuracy. Thus, the aim of this study is to help farmers identify WBPH using the latest technology. This study will start by identifying and analyzing the characteristic of WBPH. The process will proceed with image capturing using machine vision. Once the image is captured, the image will undergo an image pro-cessing process, which is image acquisition, image segmentation and lastly, feature ex-traction. For image feature extraction, there are two parameters that are chosen which are WBPH size and pattern. From the findings, WBPH can be identified using its unique pattern on its wings. Well for WBPH size, a few parameters were analyzed and a total of 8 parameters was chosen to be used as an indicator to determine the size of WBPH. An algorithm is made using LabVIEW for both feature extractions of WBPH. Next, the al-gorithm that was managed to identified WBPH, again its credibility’s is being tested whether it can identify only WBPH. The algorithm was run for all kinds of insect and if the algorithm managed to reach 90% of accuracy thus the algorithm is a success. To con-clude, this project has successfully identified WBPH based on its feature extraction and the result of accuracy achieved.