Segmentation of moving objects in image sequence based on perceptual similarity of local texture and photometric features

K L Chan

doi:10.1186/s13640-018-0308-4

Abstract

The segmentation of moving objects in image sequence can be formulated as a background subtraction problem—the separation of objects from the background in each image frame. The background scene is learned and modeled. A pixelwise process is employed to classify each pixel as an object or background based on its similarity with the background model. The segmentation is challenging due to the occurrence of dynamic elements such as illumination change and background motions. We propose a framework for object segmentation with a novel feature for background representation and new mechanisms for updating the background model. A ternary pattern is employed to characterize the local texture. The pattern and photometric features are used for background modeling. The classification of pixel is performed based on the perceptual similarity between the current pixel and the background model. The segmented object is refined by taking into account the spatial consistency of the image feature. For the background model update, we propose two mechanisms that are able to adapt to abrupt background change and also merge new background elements into the model. We compare our framework with various background subtraction algorithms on video datasets.

Highlights

Moving objects such as humans or vehicles are often the focus of image sequence analysis
The background model is represented by perception-based local ternary pattern and photometric features
The local ternary pattern is robust to random noise and invariant to scale transform

Summary

Introduction

Moving objects such as humans or vehicles are often the focus of image sequence analysis. The segmentation of moving objects is the first key problem which can be formulated as a background subtraction. Object pixels are segmented when they are found to be different from the background model. A common background subtraction framework contains background modeling, joint background/foreground classification, and background model updating. Object segmentation can be improved via background model updating or foreground model. Many background subtraction methods like [5] update parameters of matched background model with a fixed learning factor. The features representing the background scene are very important. We observed various challenges in real scenes Dynamic background elements such as tree and water produce many false-positive errors. It is because the background model does not contain sufficient and representative samples. Let N be the number of initialization image frames and n is the frame

Methods

Results

Conclusion