DETEKSI JUMLAH KENDARAAN DENGAN ALGORITMA GAUSSIAN MIXTURE MODEL DI AREA JALAN RAYA

Abstract

Road safety monitoring systems are developing at this time. The transportation sector is the object of research that continues to be developed and is always an interesting topic. Not only for security purposes and for statistical purposes for the road widening process that supports road user infrastructure, but the detection system is also useful for sales marketing statistics. In this research, propose a vehicle detection system that is useful for widening roads in a certain area or area so that it can reduce traffic congestion and accident rates. The proposed Gaussian Mixture Model method has several weaknesses, such as errors in background substitution with vehicles and failing to distribute the background with vehicle shadows. However, using morphological operations can overcome these problems. The results show a fairly good level of accuracy from the proposed method. It is only less effective when using video objects with poor lighting or at night because in the blob analysis process the detected vehicle objects do not match the actual object. But if the traffic flow is smooth and unidirectional, the proposed method is still acceptable.