A moving object detection algorithm based on graph

Abstract

This paper realizes a moving object detection algorithm based on graph. This algorithm belongs to the timespace combination object segmentation algorithm, time segmentation uses frame differential image and space segmentation uses Canny edge detector. After the morphological operator handled it formed collection set form, the fusion method based on graph combining the time and space information together. In fusion algorithm the basic elements are blobs, through the moving pixel ratio in blobs' edge to define the moving objects' priority, finally a optimization method based on graph is used to calculate the best combination of blobs, thus getting moving objects.