Search results for "GEP"

Depth Enhancement by Fusion for Passive and Active Sensing

2012

This paper presents a general refinement procedure that enhances any given depth map obtained by passive or active sensing. Given a depth map, either estimated by triangulation methods or directly provided by the sensing system, and its corresponding 2-D image, we correct the depth values by separately treating regions with undesired effects such as empty holes, texture copying or edge blurring due to homogeneous regions, occlusions, and shadowing. In this work, we use recent depth enhancement filters intended for Time-of-Flight cameras, and adapt them to alternative depth sensing modalities, both active using an RGB-D camera and passive using a dense stereo camera. To that end, we propose …

Keywords given by authors of scientific articles in database descriptors

2007

Photo Use While Dating: From Forecasted Photos In Tinder To Creating Copresence Using Other Media

2019

While studying Finnish users of the online dating app Tinder searching for long- term partnerships, we paid attention to the importance of photos in their social interactions. Based on our study, we argue that photos play an important role in online dating. Initially, photos are chosen and uploaded to influence future interactions, particularly regarding who will contact them via their profile. We term this particular future tense of photography forecasted photos. Second, photos enabled the creation of copresence between dates, especially via instant messaging services instantly after capture. Third, the classic notions of photos depicting the past became important when wanting to be remind…

To Move or Not to Move?

2019

This chapter deals with the problem of including motion cues in VR applications. From the challenges of this technology to the latest trends in the field, the authors discuss the benefits and problems of including these particular perceptual cues. First, readers will know how motion cues are usually generated in simulators and VR applications in general. Then, the authors list the major problems of this process and the reasons why its development has not followed the pace of the rest of VR elements (mainly the display technology), reviewing the motion vs. no-motion question from several perspectives. The general answer to this discussion is that motion cues are necessary in VR applications—…

Hypergraph imaging: an overview

2002

Hypergraph theory as originally developed by Berge (Hypergraphe, Dunod, Paris, 1987) is a theory of finite combinatorial sets, modeling lot of problems of operational research and combinatorial optimization. This framework turns out to be very interesting for many other applications, in particular for computer vision. In this paper, we are going to survey the relationship between combinatorial sets and image processing. More precisely, we propose an overview of different applications from image hypergraph models to image analysis. It mainly focuses on the combinatorial representation of an image and shows the effectiveness of this approach to low level image processing; in particular to seg…

A novel sequential testing procedure for selecting the number of changepoints in segmented regression models

2023

In this work, we address the problem of selecting the number of changepoints in segmented regression models. We propose a novel stepwise procedure and assess its performance through simulation studies. We demonstrate that our proposal behaves well with the Gaussian and Binomial responses.

Wavelet-based video compression: A glimpse of the future?

2004

Masteroppgave i informasjons- og kommunikasjonsteknologi 2004 - Høgskolen i Agder, Grimstad Even though wavelet-based video compression has been an area of research for more than a decade, motion estimation and compensation has been considered complex and inefficient until recently. We have carried out a thorough investigation of existing research work in this field, and found that fundamental problem with wavelet-based temporal removal while obtaining highly scalability (the wavelet-properties of multiresolution structure in combination with embedded coding), has been solved by performing motion compensated temporal filtering within the wavelet domain of a overcomplete three-dimensional li…

Analysis of binarization techniques and Tsetlin machine architectures targeting image classification

2020

Master's thesis in Information- and communication technology (IKT590) The Tsetlin Machine is a constantly evolving and developing machine learning technique with ever-increasing success. However, for every success, the Tsetlin Machine achieves, a new set of challenges are put ahead. To sufficiently bring the Tsetlin Machine to a broadly used standard, these challenges must be completed. This thesis focuses on the challenge of doing color image classification and will provide an introductory description of how this is possible through the usage of an older technique, namely binarization. A comparison with the various Tsetlin Machine adaptations made public in recent times is also present aft…

An improved image processing chain for mobile terminals

2002

Masteroppgave i informasjons- og kommunikasjonsteknologi 2002 - Høgskolen i Agder, Grimstad Based on the proposal in the report by Tryggve Mikkelsen [Mikkelsen, 2001] this thesis investigates and discusses the implementation of a simplified image processing chain for modern mobile terminals. Previous work on image processing in this field of application has shown that the demosaicing operation, color space conversion and downsampling are resource demanding pre-processing operations. The simplification potential of the signal chain lies within the pictured scenario of using Bayer RGB from the image sensor directly as input to the compression scheme. By omitting these pre-processing operation…

Deep Convolutional Neural Networks for Semantic Segmentation of Multi-Band Satellite Images

2018

Master's thesis Information- and communication technology IKT590 - University of Agder 2018 Semantic segmentation of images is of increasing interest in the eld of computer vision and machine learning. Accurate and e cient segmentation methods is required for many of todays modern applications. This the- sis provides a review of deep learning methods for semantic segmentation of satellite images. Firstly, we compare di erent state-of-the-art methods. Next, we explore the bene ts of using multiple spectral bands of data as compared to the traditional RGB bands. Finally, a look at future possibil- ities with segmentation using capsule networks.