USA 2014 Abstracts


Full Papers
Paper Nr: 1
Title:

Toward Moving Objects Detection in 3D-Lidar and Camera Data

Authors:

Clement Deymier and Thierry Chateau

Abstract: In this paper, we propose a major improvement of an algorithm named IPCC (Clement Deymier, 2013a) for Iterative Photo Consistency-Check. His goal is to detect a posteriori moving objects in both camera and rangefinder data. The range data may be provided by different sensors such as: Velodyne, Riegl or Kinect with no distinction. The main idea is to consider that range data acquired on static objects are photo-consistent, they have the same color and texture in all the camera images, but range data acquired on moving object are not photo-consistent. The central matter is to take into account that range sensor and camera are not synchronous, so what is seen in camera is not what range sensors acquire. This work propose to estimate photo-consistency of range data by using his 3D neighborhood as a texture descriptor wich is a major improvement of the original method based on texture patches. A Gaussian mixture method has been developed to deal with occluded background. Moreover, we will see how to remove non photo-consistent range data from the scene by an erosion process and how to repair images by inpainting. Finally, experiments will show the relevance of the proposed method in terms of both accuracy and computation time.

Short Papers
Paper Nr: 2
Title:

Paris-rue-Madame Database - A 3D Mobile Laser Scanner Dataset for Benchmarking Urban Detection, Segmentation and Classification Methods

Authors:

Andrés Serna, Beatriz Marcotegui, François Goulette and Jean-Emmanuel Deschaud

Abstract: This paper describes a publicly available 3D database from the rueMadame, a street in the 6th Parisian district. Data have been acquired by the Mobile Laser Scanning (MLS) system L3D2 and correspond to a 160 m long street section. Annotation has been carried out in a manually assisted way. An initial annotation is obtained using an automatic segmentation algorithm. Then, a manual refinement is done and a label is assigned to each segmented object. Finally, a class is also manually assigned to each object. Available classes include facades, ground, cars, motorcycles, pedestrians, traffic signs, among others. The result is a list of (X, Y, Z, reflectance, label, class) points. Our aim is to offer, to the scientific community, a 3D manually labeled dataset for detection, segmentation and classification benchmarking. With respect to other databases available in the state of the art, this dataset has been exhaustively annotated in order to include all available objects and to allow point-wise comparison.

Paper Nr: 3
Title:

General Road Detection Algorithm - A Computational Improvement

Authors:

Bruno Ricaud, Bogdan Stanciulescu and Amaury Breheret

Abstract: This article proposes a method improving Kong et al. algorithm called Locally Adaptive Soft-Voting (LASV) algorithm described in ”General road detection from a single image”. This algorithm aims to detect and segment road in structured and unstructured environments. Evaluation of our method over different images datasets shows that it is speeded up by up to 32 times and precision is improved by up to 28% compared to the original method. This enables our method to come closer the real time requirements.