Identification of object characteristics using the fusion sensor method in wireless sensor network environments

This paper contains an explanation of the research results to solve the problem of selecting the category of mangosteen fruit in a grading process using method image processing through geometry measurements of flat areas, color measurements, shape and texture characteristics. The aim is to implement the selection process through a grading machine to be able to sort two categories of mangosteen fruit based on the characteristics that are trained using the Linear Discriminant Analysis method. The proposed method is the use of four image sensors which are perpendicular to each other in the grading process line. Each sensor provides a decision on the group of objects it identifies, “accepted” if it meets all the requirements or “rejected” if one of the requirements is not met. The results of each sensor decision are sent to the sink node in the wireless sensor network environment to be jointly decided using the method of redundant cooperative sensor whether the objects identified together can be “accepted” or “rejected”. The results of grouping give an error rate of ± 9.8%. The biggest identification error comes from the measurement of colors caused by light noise.