Search

Vehicle detection plays a significant role in traffic monitoring. Vehicle detection approaches can be used for vehicle tracking, vehicle classification and traffic analysis. However, numerous attributes like shape, intensity, size, pose, illumination, shadows, occlusion, velocity of vehicles and environmental conditions, provide different challenges for the detection step. With an appropriate vehicle detection technique, we are able to extract valuable knowledge from video sequences, regardless these diverse factors. Since the vehicle detection method choice has a deep impact on this step and the whole traffic monitoring system performances, our objective in this study is to investigate different methods for vehicle detection. Comparison is made on the basis of different metrics such as recall, precision and detection accuracy. These approaches have been tested under different weather conditions (rainy, sunny) and various traffic conditions (light, medium, heavy).

In recent years, cities are experiencing a deep negative impact due to traffic congestion. As the number of vehicles is increasing rapidly, traffic congestion becomes unsustainable in urban domains specially in peak hours leading to time and fuel waste, accidents, additional costs for the economy, environmental problems and consequently impaired quality of life. Tremendous effort is done to provide solutions dealing with traffic congestion detection and monitoring. Based on used technologies, existing solutions can be classified into three categories: sensor technology, Vehicular Adhoc Networks and visual traffic surveillance. In this survey, we review the main recent works proposed in each category and discuss their strengths and weaknesses. We give an insight of future directions for handling congestion using the presented methods.

In the context of platooning, a high degree of cooperation between platooning members is required to perform major maneuvers. One of the most challenging issues is to perform join-maneuver due to the strong interference caused by unintended vehicles entering in the middle during this maneuver. In order to detect and identify vehicles (unintended or not) during a join maneuver, we propose a vision based solution allowing to observe the front area of the platooning member, track and identify newly joined vehicles. In this paper, we focus on delineating this region of interest (ROI) by extracting road lanes in video sequences from on-board cameras in the platooning vehicles. The performance of our method is tested and validated using videos of highway scenes in different weather conditions (sunny, rainy, cloudy).

Segmentation and volume measurement of liver tumor are important tasks for surgical planning and cancer follow-up. In this work, a segmentation method from four-phase computed tomography images is proposed. It is based on the combination of the Expectation-Maximization algorithm and the Hidden Markov Random Fields. The latter considers the spatial information given by voxel neighbors of two contrast phases. The segmentation algorithm is applied on a volume of interest that decreases the number of processed voxels. To accelerate the classification steps within the segmentation process, a Bootstrap resampling scheme is also adopted. It consists in selecting randomly an optimal representative set of voxels. The experimental results carried out on three clinical datasets show the performance of our liver tumor segmentation method. It has been notably observed that the computing time of the classification algorithm is reduced without any significant impact on the segmentation accuracy.

In this paper we propose a new multi-scale fully automatic algorithm based on Graph cuts for vessel extraction. In fact, we combine vesselness, geodesic paths, a multi-scale edgeness map and the directional information for vessel tracking in order to personalize the Graph cuts approach to the segmentation of tubular structures.