Uso de camêra Omnidirecional para odometria visual.

Lista1 lista1_vc.rar

Lista 3 lista3_implementacao.rar

Visual Odometry with Omni-directional images

Abstract—We describe a method for visual odometry, using optical flow, with a single omni-directional (catadioptric) camera. We show how omni-directional images can be used to perform optical flow, discussing the basis of optical flow and some restrictions needed to it and how unwrap these images, in special how to unwrap omni-directional to a specific kind of view called Bird’s eye view, that correspond to scaled orthographic views of the ground plane. Omni-directional images facilitate landmark based odometry, since landmarks remain visible in all images, as opposed to a small field-of-view standard camera. Also, omni-directional images provide the means of having adequate representations to support accurrate odometry. Tests were performed to measure robustness and performance of our approach with analysis of the data acquired.

Our proposed system can be seen on below. First the preprocesing step containing acquisition of the images by the Omni-directional and camera calibration process. Acquired images pass though a unwrapping process and following to extraction of features, these features feed the optical flow method that estimate motion between consecutive pictures and passing for a odometry process where the motion will be transformed to odometry of the robot and return to unwrapping. These last four steps are defined like Main loop.

Apresentacao gabriel_apresentacao.pdf

Artigo Final- visual_odometry_final_version.pdf

Imagens navegacao corredor Icexteste_corredor_verlab_5m_30fps.rar Obs: imagens nao sao proprias para uso(apenas ilustrativo) são imagens .jpg!

Video baixa qualidade teste1(nao retificado,nos teste usou-se retificado como no teste 2)-video_teste1_nao_retificada.rar

Video baixa qualidade teste2(retificado)-video_teste2_retificada.rar