CANDID: Robust Change Dynamics and Deterministic Update Policy for Dynamic Background Subtraction

Background subtraction in video provides the preliminary information essential for many computer vision applications. In this paper, we propose a sequence of approaches named CANDID to solve the change detection problem in challenging video scenarios. The CANDID adaptively initializes the pixel-level distance threshold and update rate. These parameters are updated by computing the change dynamics at a location. Further, the background model is maintained by formulating a deterministic update policy. The performance of the proposed method is evaluated over various challenging scenarios such as dynamic background and extreme weather conditions. The qualitative and quantitative measures of the proposed method outperform the existing state-of-the-art approaches.

I. Introduction

Change detection is one of the preliminary tasks in numerous computer vision applications such as behavior analysis, traffic monitoring, video synopsis, action recognition, visual surveillance, anomaly detection and object tracking. Background subtraction is an effective approach to detect the relevant changes by segmenting the video frames into foreground and background regions. Moving object detection in videos with dynamic background changes, illumination variations and challenging environmental conditions is a challenging task due to the fluctuation and noise in the background appearance. These dynamic changes affect the accuracy of a background subtraction technique.