DETEKSI DAN PENGGOLONGAN KENDARAAN DENGAN KALMAN FILTER DAN MODEL GAUSSIAN DI JALAN TOL

Abstract

At this presents how to record a moving object and save as new video files (*.avi) resolution 640x480, then stored in storage. Moving object given line ROI (Region of Interest) to easily detected than filtering the moving objects (Vehicles) by using a Gaussian Mixture Model (GMM) with 2 types of distribution, i.e. Bacground and Foreground distribution. The shape of the foreground distribution is filtered by morphological operations and segmented by using Bit Large Object (BLOB) Segmentation to get the vehicle dimensions. Feature extraction results of these vehicles, will be used as data organized for vehicles classification. The results of segmentation BLOB used to kalman filter for counting moving object if segmentation BLOB doesn't found object than continue the next frames. The last Result about these is detection system will calculate for validation using True Positiv, True Negative, False Positive, and False Negative looking for sensitivitas and spesifisitas each conditions dawn, afternoon and dusk.