Edge-segment-based Background Modeling: Non-parametric online background update

Abstract

For background-subtraction-based moving object detection, reliable background modeling is the most important component. Pixel-based methods are sensitive to illumination change, and edge-based methods can solve illumination-related problems, but have shape distortion problems. In this paper, we propose an edge-segmentbased statistical background modeling algorithm and an online update mechanism to detect moving objects from consecutive frames, which creates a balance between the pixel-and edge-based methods. Our background modeling method uses a statistical map to model the frequency of the background-edges, as distributions that comprise support regions approximated with a quadratic function and enhanced with color and gradient information, to overcome the edge-distortion problem by matching the edge-segments to the modeled distributions. To adjust the changing background in the scenes, we propose an online-background update step for every incoming frame that updates the statistical map and enhances the information held by the distributions support regions. Furthermore, our experiments show that the proposed method obtains better results and detects moving edges efficiently.

On the Role and the Importance of Features for Background Modeling and Foreground Detection

“…Edges based approaches used in addition with intensity or color features are the most investigated approaches and allow to combine the advantage of the two features. Forthe approahces based on edges alone, only two main works emerged that are the works of based on edge segments [365][412] [265][366] [264][411] [266] and the work based on subpixel edge [226]. These approaches appear to be relevant too and merit to be more investigated.…”

“…Thus, basic comparison of edge-segments produces similar results as edge-pixel-based approaches. To solve this problem, statistical edge-segment-based methods extract movement of edge-segments including edge distortion [365][412] [265][366] [264][411] [266]. Thus, these methods solve the edge-variation problem by accumulating edge existence from a training sequence [267].…”

Unattended object detection based on edge-segment distributions

“…Moreover, existing edge-based methods [6] have many false alarms, because they use a simple edge dif- ferencing method. To solve this problem, edge-segmentbased methods [7,8,10,11] model background using the connected edges instead.…”