Improvement in the Accuracy of the Moving Object Position by Eliminating Erroneous Sensors with K-Means Clustering Approach

Clustering is the process of grouping objects that have similar features. The integration of data from multiple sensors can improve the accuracy of information than using a single sensor. Electro-Optic Sensor can provide the azimuth and elevation of the moving object at any time instance. So it can give the direction of the moving object, so if we have at least two sensors than the actual position of the object can be calculated with the help of the triangulation method. As we increase the number of sensors then the accuracy in the position of the moving objects increases. But meanwhile, if any of the sensors has erroneous measurement then the final position measurement will be erroneous. So we have to eliminate the effect of this erroneous sensor from the final measurement. This paper summarizes how the k-means clustering technique can be applied to identify and eliminate the erroneous sensor from the measurement. K-means clustering algorithm is applied in such a way that the erroneous measurement will be discarded based on not belonging in the largest cluster. And centroid of the largest cluster will give the accurate position of the moving object.