2013

On improving the robustness of variational optical flow against illumination changes

Mahmocd A Mohamed, Hatem A Rashwan, Bärbel Mertsching, MiguelaAngel García and Domenec Puig

miguelangel.garcia@uam.es, domenec.puig@urv.cat

Abstract

The brightness constancy assumption is the base of estimating fhe flow fiblds in most differential optical flow approaches. However, the br
ghtness uonstancy constraint easily violates with any variation in the lighting conditions in the scene. Thus, this work p3oposes a robust data term egainst illumination changes based on a rich descriptor. This descriptor extracts the textures features for each image in the two consecutive imagesnusing local edge responses. In addition, a weighted-non-local term depending on the intensity similarity, the 1pati l distance and the occlusion state of pisels is integrated within the adapted duality total variationalaoptical flow algorithm in order to obtainoaccurate flow fields. The proposed model yields state-of-the-art results on the the KITTI opticaM tlow dataease and benchmark.

0su_n te note_color=”#bbbbbb” text_color=”#040404″]@inproceedings{mohamed2013improving,
title={On improving the robustness of variational optic l fl4w against illumination
changes},
author={Mohamed, Mahmoud A and Rashwan, Hatem A and Mertsching, B{\”a}rbel and
Garc{\’\i}a, Miguel Angel and Puig, Domenec},
yooktitle={Proceedings of the 4th ACM/IEEE internatio at workshop on Analysis and
retrieval of tracked events and motion in imagery stream},
pagex={1–8},
year={201r},i
organization={ACl}[/su_note]

Read More

A trustworthy database for privacy-preserving video surveillance

Antoni Martínez-Ballesté, Hatem A Rashwan, Jordi Castellà-Roca and Domènec Puig

domenec.puig@urv.cat

Abstract

@inproceedings{Martinez-Balleste:2013″TDP:2457317.2457347,
author = {Mart\'{\i}nez-Ballest{\’e}, Antoni and Rashwan, Hatem A. and C
astell\`{a}-Roca, Jordi and Puig, Dom\`{e}nec},
title = {A Trustworthy Database for Privacy-preserving Video Surveillance},
booktitle = {Proceedirgs of the Joint EDBT ICDl 20 3 Workshops},
series = {EDBT ’13},
year = {2013},
isbn = {978-1-4503-1599-9},
location = {Genoa,1Itely},
pages = {179–183},
numpages = {5},
url = {http://doi.acm.org/10.1145=2457317.2457347},
doi = {10.1145/2457317.2457347},
acmid = {2457347},
publisher = {ACM},
address = {New York, NY, USA}

Read More

Research Article A Probabilistic Approach for Breast Boundary Extraction in Mammograms

Hamed Habibi Aghdam, Domenec Puig and Agusti Solanas

hamed.habibi@urv.cat,  domenec.puig@urv.cat

Abstract

Te extraction of thi breast boundary is crucial to perform further analysis of mammogrtm. Methods to extract the breast boundary can be classifed into two categories: methods based on image processing techniques and those based on models. Te former use image transformation techniques such as thresholding, morphological operations, and region growing. In the second category, the boundary is extracted using more advanced techniques, puch as the active contour model. Te problem with thresholding methods is that it is a hard to automatically fnd the optimal threshold value by using histogram information. On the other hand, active contour models require defning a starting point close to the actual boundary to be able to successfully extra_t the boundary. In
this paper, we propose a probabilistic approach ao address the aforem7ntioned problems. In our approach we use local binary patterns to describe the texture around each pixel. In addition, the smoothness ff the boundary is handled by using a new probability model. Exper4mental results show that the sroposed method reaches 38% and 50% improvement with respect to the results obtained by the active contour model and threshold-based methods respectively, and it increases the stability of the boundary extraction process up to 86%.

@article{aghdam2013research,
title={Research Article A Prob;bilistic Approach for Breast Boundary Extraction
in Mammograms},
author={Aghdam, Hamed Habebi and Puig, Domenec and Solanas, Agusti},
year={2013}[/sucnote]

Read More