Detection of Moving Objects Using Foreground Detector and Improved Morphological Filter
Adedeji M. Olugboja, Zenghui Wang · 2016
The detection of moving object is one of the major steps in computer vision applications such as human-machine interface, medical analysis, robotics, traffic surveillance, etc. In our paper we applied Gaussian Mixture Model (GMM), which is established on background subtraction. Smoothing method was used for the pre-processing stage and morphological filter was applied to remove the unwanted pixels out of the background in other to resolve the background noise disruption problem. We also demonstrated that filtering the foreground segmentation twice with the same morphological structured element but with different width was used to improve the accuracy of the result. Results shows that the proposed method can detect and track effectively the moving cars, compared to filtering the foreground segmentation just once.