Nicolas Paparoditis

Road Sign Detection in Images: A Case Study

Road sign identification in images is an important issue, in particular for vehicle safety applications. It is usually tackled in three stages: detection, recognition and tracking, and evaluated as a whole. To progress towards better algorithms, we focus in this paper on the first stage of the process, namely road sign detection. More specifically, we compare, on the same ground-truth image database, results obtained by three algorithms that sample different state-of-the-art approaches. The three tested algorithms: Contour Fitting, Radial Symmetry Transform, and pair-wise voting scheme, all use color and edge information and are based on geometrical models of road signs. The test dataset is made of 847 images 960x1080 of complex urban scenes (available at www.itowns.fr/benchmarking.html). They feature 251 road signs of different shapes (circular, rectangular, triangular), sizes and types. The pros and cons of the three algorithms are discussed, allowing to draw new research perspectives.

TerraMobilita/iQmulus urban point cloud analysis benchmark

The objective of the TerraMobilita/iQmulus 3D urban analysis benchmark is to evaluate the current state of the art in urban scene analysis from mobile laser scanning (MLS) at large scale. A very detailed semantic tree for urban scenes is proposed. We call analysis the capacity of a method to separate the points of the scene into these categories (classification), and to separate the different objects of the same type for object classes (detection). A very large ground truth is produced manually in two steps using advanced editing tools developed especially for this benchmark. Based on this ground truth, the benchmark aims at evaluating the classification, detection and segmentation quality of the submitted results. Graphical abstractDisplay Omitted HighlightsVery rich data: high accuracy, high resolution, many attributes.Massive data: 160 million annotated points thanks to a performant web based annotation tool (and many hours of work).Rich semantics organized in a semantic tree with various levels of generalization.Very objective evaluation metrics.

Vehicle localization using mono-camera and geo-referenced traffic signs

Vision based localization is a cost effective method for indoor and outdoor application. However, it has drift problem if none global optimization is used. We proposed a geo-referenced traffic sign based localization method, which integrated the constraints of 3D traffic signs with local bundle adjustment to reduce the drift. Comparing to global bundle adjustment, Local Bundle Adjustment(LBA) has low computational cost but suffers the drift problem for large scale localization because of the random error accumulation. We reduced the drift by means of the constraints from geo-referenced traffic signs for bundle adjustment process. The original LBA model was extended for the constraints and the traffic signs were detected in images and matched with 3D landmark database automatically. From the experiments of simulated and real images, our approach can reduce the drift and have better locating results than none-constraint LBA based localization method.

Estimating meteorological visibility using cameras: a probabilistic model-driven approach

Estimating the atmospheric or meteorological visibility distance is very important for air and ground transport safety, as well as for air quality. However, there is no holistic approach to tackle the problem by camera. Most existing methods are data-driven approaches, which perform a linear regression between the contrast in the scene and the visual range estimated by means of reference additional sensors. In this paper, we propose a probabilistic model-based approach which takes into account the distribution of contrasts in the scene. It is robust to illumination variations in the scene by taking into account the Lambertian surfaces. To evaluate our model, meteorological ground truth data were collected, showing very promising results. This works opens new perspectives in the computer vision community dealing with environmental issues.

Generation of an integrated 3D city model with visual landmarks for autonomous navigation in dense urban areas

In the context of urban autonomous navigation systems for going from point A to point B, a practicable trajectory which takes into account drivable areas and permanent obstacles should be designed first. A robot should then follow this trajectory while avoiding not only dynamic obstacles such as cars and pedestrians but also permanent obstacles such as road sides and central islands. To this end, a robot must be aware of its exact position and must be informed of what its immediate environment is at all times. In dense urban areas, GNSS systems generally suffer from lack of precision due to masks and multipaths. Localization systems have to model these phenomena and even merge with vision based methods in order to obtain the high accuracies required in the process. In this paper we propose an integrated geographic database enabling, on the one hand, the GNSS and vision based localization methods to obtain the required accuracy, and on the other, to provide the robots with information about its surroundings such as drivable surfaces and permanent obstacles. The database is comprised of 3D buildings, 3D roads and a set of 3D visual landmarks. Our system provides most of the information required for autonomous navigation in dense urban areas and has successfully been embedded in real experiments, thanks to a real-time querying system.

Extracting Outlined Planar Clusters of Street Facades from 3D Point Clouds

This paper presents an approach for extracting 3D outlined planar clusters of street facades. Terrestrial laser data are acquired using a Mobile Mapping System (MMS). Mapping of street facades is of great interest in various digital mapping and robotic research topics. After a filtering step of the 3D point cloud, the dominant hypothetical facade planes are detected using an adapted Progressive Probabilistic Hough Transform (PPHT). The corresponding planar clusters are extracted using a priori geometric knowledge of street. The clusters are horizontally and vertically delimited using heuristic approaches. The adapted PPHT allows the automatic extraction of georeferenced planar clusters of facades with a fine detection of dominant facade lines and a low computation time. The adopted approach has been tested on a set of point cloud acquired in the city of Paris under real conditions. Examples and experimental results show the efficiency and the potential of the proposed approach.

A web-based 3D mapping application using WebGL allowing interaction with images, point clouds and models

Nowadays we see a lot of Rich Internet Applications (RIA), but real interactive 3D applications have just started years ago, and it is only in 2011 that a GPU library was integrated directly in the major internet browsers. This opened a new way of creating 3D applications on the web and a new public, as much more users can access them without complicated plugin installations and this is just what 3D web mapping was waiting for. We present our street view application allowing to move through cities from a pedestrian point of view and interact with it in order to create and update precise urban maps enabling accessibility diagnostics, services and applications around mobility. We cared about precision in the data acquisition (images and laser) in order to be able to make accurate 3D measurements for professional usages. The user can show large laser clouds in his web-browser, access information on any points, draw 3D bounding boxes and export selections to databases. He can also make measurements from images, integrate collada models, display OpenData layers, etc. Finally we use projective texture mapping in real time to texture any mesh with our images.

Increasing interactivity in street view web navigation systems

This paper presents some interactive features we have added on our street-view web navigation application. Our system allows to navigate through a huge amount of data (panoramas and laser clouds) and also to interact with it. We will detail 4 aspects of this interactivity. First, the labelling, displaying of features directly into the images in the 3D space, useful for general public but also researchers in image processing and computer vision. Secondly we propose a crowd sourcing mode for blurring people not automatically detected. Thirdly we offer the possibility for the web user to localize and measure in 3D all objects visible in the images by plotting only in one image. Finally we developed a multimedia editor that allows public administrations (like town halls, museums, operas, theaters, etc.) to add interactive content like video or images at the exact 3D position/orientation/size they chose with an easy manipulating editor to augment with realism the static scenes with dynamic or fresher elements.

Use intermediate results of wrapper band selection methods: A first step toward the optimization of spectral configuration for land cover classifications

Intermediate results of two state-of-the-art wrapper feature selection approaches (GA and SFFS) associated to a classifier (linear SVM) applied to hyperspectral data sets were used to derive information about band importance for specific land cover classification problems. The impact of the number of selected bands on classification accuracy was obtained thanks to SFFS, while a band importance measure was derived from intermediate sets of bands tested by GA. Such results are a first step toward the identification of the most suitable spectral bands to design superspectral camera systems dedicated to specific applications (e.g. classification of urban land cover and material maps).

Semantic Approach in Image Change Detection

Change detection is a main issue in various domains, and especially for remote sensing purposes. Indeed, plethora of geospatial images are available and can be used to update geographical databases. In this paper, we propose a classification-based method to detect changes between a database and a more recent image. It is based both on an efficient training point selection and a hierarchical decision process. This allows to take into account the intrinsic heterogeneity of the objects and themes composing a database while limiting false detection rates. The reliability of the designed framework method is first assessed on simulated data, and then successfully applied on very high resolution satellite images and two land-cover databases.