Background Extraction from Image using Depth Estimation

This article focuses on creating a 3D depth representation and successfully extracting the background based on the acquired results. Furthermore, this work deals with studying the current state of the art and exploring the possibilities of their practical application. The objective was also to study the issues related to background removal or modification in images and provide an overview of available methods for this particular application. The main goal was to design and implement an algorithm for background extraction. The proposed algorithm was then compared with another existing method, that solves the mentioned problem. The conclusion consists of evaluating the results of each method and describing their strengths and weaknesses.

I. Introduction

In recent decades, information technology has seen rapid development. Modern digital images contain a plethora of data that can be utilized for various applications such as object detection, editing, and more [1], [2]. One of these techniques is background extraction, crucial in computer vision for dis-tinguishing between foreground and background objects [3]. Our paper enhances an existing method for improved back-ground extraction [4], [5]. The paper is structured as follows: Section II reviews related works; Section III introduces our proposed method based on enhancements to the Semi-Global Block Matching (SGBM) method; Section IV discusses the experimental setup; Section V analyzes the results; and Section VI concludes the paper.