A filtering approach for human detection in bionic eye.

TitleA filtering approach for human detection in bionic eye.
Publication TypeJournal Article
Year of Publication2013
AuthorsWang F, Li Y, Barnes N
JournalConference proceedings : ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference
Volume2013
Pagination5279-82
Date Published2013
Abstract

Current bionic eye hardware implants have limited number of separately perceivable electrodes. This challenges the processing algorithms for the visualization unit in bionic eye. Therefore, it is important to ensure that the critical information must remain perceivable in downsampled images. Since the main goal for any bionic eye device is for interaction in the user's everyday life, the location of human subjects is one of the important sources, and detecting humans in natural scenes becomes fundamental in the process of extracting critical visual information. This paper presents an approach to localizing human subjects using a filtering approach. Filtering is one of the most efficient approaches for processing images. In addition, it is very safe in biomedical devices because the behaviour of a filter is well understood. Thus it is easier to pass the regulatory authorities. Our goal is to automatically generate a set of filters, each of which describes a representative pose of human bodies. This is achieved by grouping visual features from images directly. In this process, we use Support Vector Machine followed by an efficient post-processing procedure. Then, these filters naturally define human bodies in a fine granularity for effective image processing. Experiments on a large number of images suggest our approach is practical for bionic eye.

DOI10.1109/EMBC.2013.6611204
Alternate JournalConf Proc IEEE Eng Med Biol Soc