Image processing methods for a proposed laser pest extraction and monitoring system

Abstract

A laser pest extraction system was proposed in this paper based on image analysis techniques for tracking and identification. High powered laser pointers were proposed and varying laser intensity will be applied based on types of pest detected. Classification is done based on two types of targeted pest. Firstly, the centroid of the pest are obtained from Sobel row edge detection methods and image subtraction methods. The output centroid of the pest edge are further analysed to locate the centroid of the pest. From the centroid, a portion of the pest image, 25 pixels are cropped and fed to the feed forward back propagation neural network for training.