Title: ANALYSIS OF A BASIC MULTIPLE-IMAGE ALGORITHM FOR REMOVAL OF FOREGROUND OBJECTS

Year of Publication: Jul - 2013
Page Numbers: 248-255
Authors: Oscar John Gerald Somsen
Conference Name: The Third International Conference on Digital Information and Communication Technology and its Applications (DICTAP2013)
- Czech Republic

Abstract:


In this study I consider a multiple-image algorithm for automatic removal of foreground objects that block the view in the image of an object of interest. Such and other algorithms for image analysis can assist the operator of an optical surveillance system to analyze the large volumes of image data that are typically supplied by the detectors. By considering multiple images from slightly different viewpoints a part of the object of interest that is covered by a foreground object in one image may be visible in another. By combining the information a single enhanced image of the object may be presented to the operator where the foreground objects have been removed. Our algorithm involves alignment and comparison of the images. I was able to successfully process experimental images and produce images were larger foreground objects were indeed removed. Our results show that the alignment is critical and may be adapted to improve the overall result. Also, a careful presentation of the enhanced images is required since otherwise a partially removed foreground object may confuse rather than assist the operator.