Identification of Object Characteristics Using The Sensor Fusion Method in Wireless Sensor Network Environments
Rida Hudaya, Feni Isdaryani
Politeknik Negeri Bandung
Abstract
This paper describes the research results to identify the characteristics of objects using image processing with the method of flat area geometric measurement, color measurements, size measurement, shape characteristics, and texture characteristics as data information used to decide on two groups of objects using the method of the Linear Discriminant Analysis for the application of mangosteen grading and sorting process. The proposed solution method is the use of four image sensors placed at four different monitoring angles. Each image sensor provides a decision on the group of objects it identifies, "accepted" if it meets all the objects requirements or "rejected" if one of the objects characteristics requirements is not met. The results of each image sensors decision are sent to the sink node in the wireless sensor network environment to be decided together using the sensor fusion method whether the objects identified in the group are "accepted" by all image sensors or "rejected" by one or more sensors image. The grouping results give a 10% error rate, and the biggest identification error comes from the color measurement.
Keywords: Image Processing, Linear Discriminant Analysis, Sensor Fusion, Wireless Sensor Networks
Topic: Electrical Engineering