Vehicle-mounted Camera Research Articles

Robust and Real-Time Traffic Lights Recognition in Complex Urban Environments

The traffic lights play an indispensable role in urban road safety and researches on intelligent vehicles become more popular recently. In this paper an automatic system for robust and real-time detection and recognition of traffic lights for intelligent vehicles based on vehicle-mounted camera is proposed. The method applying image processing and pattern recognition theory mainly works in three stages. First, the candidate regions of traffic lights are extracted using the color threshold segmentation method. Secondly, noise removal and two types of filtering which take account of shape information are applied to the candidate regions. Thirdly, template matching using normalized cross correlation techniques is adopted to validate the traffic lights candidate. Experimental results show that the proposed algorithm works effectively and robustly for traffic lights recognition in complex urban environments.

Multiband Image Segmentation and Object Recognition for Understanding Road Scenes

This paper presents a novel method for semantic segmentation and object recognition in a road scene using a hierarchical bag-of-textons method. Current driving-assistance systems rely on multiple vehicle-mounted cameras to perceive the road environment. The proposed method relies on integrated color and near-infrared images and uses the hierarchical bag-of-textons method to recognize the spatial configuration of objects and extract contextual information from the background. The histogram of the hierarchical bag-of-textons is concatenated to textons extracted from a multiscale grid window to automatically learn the spatial context for semantic segmentation. Experimental results show that the proposed method has better segmentation accuracy than the conventional bag-of-textons method. By integrating it with other scene interpretation systems, the proposed system can be used to understand road scenes for vehicle environment perception.

1-Point-RANSAC Structure from Motion for Vehicle-Mounted Cameras by Exploiting Non-holonomic Constraints

This paper presents a new method to estimate the relative motion of a vehicle from images of a single camera. The computational cost of the algorithm is limited only by the feature extraction and matching process, as the outlier removal and the motion estimation steps take less than a fraction of millisecond with a normal laptop computer. The biggest problem in visual motion estimation is data association; matched points contain many outliers that must be detected and removed for the motion to be accurately estimated. In the last few years, a very established method for removing outliers has been the “5-point RANSAC” algorithm which needs a minimum of 5 point correspondences to estimate the model hypotheses. Because of this, however, it can require up to several hundreds of iterations to find a set of points free of outliers. In this paper, we show that by exploiting the nonholonomic constraints of wheeled vehicles it is possible to use a restrictive motion model which allows us to parameterize the motion with only 1 point correspondence. Using a single feature correspondence for motion estimation is the lowest model parameterization possible and results in the two most efficient algorithms for removing outliers: 1-point RANSAC and histogram voting. To support our method we run many experiments on both synthetic and real data and compare the performance with a state-of-the-art approach. Finally, we show an application of our method to visual odometry by recovering a 3 Km trajectory in a cluttered urban environment and in real-time.

A210 PTAMによる小型ラジコンヘリコプタのマーカレス自律飛行制御(OS3 自律知能無人ビークルの運動と制御2)

Recently, it is examined to use small RC helicopters for patrol or information gathering in buildings. In buildings, GPS cannot be used and an alternative non-GPS-based positioning system is required. In this paper, we develop an autonomous control system based on the vision from a vehicle-mounted camera by using a PTAM that has been developed as a tool for augmented reality. The experimental results show the effectiveness of the proposed method.

Camera motion estimation by tracking contour deformation: Precision analysis

An algorithm to estimate camera motion from the progressive deformation of a tracked contour in the acquired video stream has been previously proposed. It relies on the fact that two views of a plane are related by an affinity, whose six parameters can be used to derive the six degrees-of-freedom of camera motion between the two views. In this paper we evaluate the accuracy of the algorithm. Monte Carlo simulations show that translations parallel to the image plane and rotations about the optical axis are better recovered than translations along this axis, which in turn are more accurate than rotations out of the plane. Concerning covariances, only the three less precise degrees-of-freedom appear to be correlated. In order to obtain means and covariances of 3D motions quickly on a working robot system, we resort to the Unscented Transformation (UT) requiring only 13 samples per view, after validating its usage through the previous Monte Carlo simulations. Two sets of experiments have been performed: short-range motion recovery has been tested using a Staübli robot arm in a controlled lab setting, while the precision of the algorithm when facing long translations has been assessed by means of a vehicle-mounted camera in a factory floor. In the latter more unfavourable case, the obtained errors are around 3%, which seems accurate enough for transferring operations.

A Study on Predicting Hazard Factors for Safe Driving

This paper proposes an algorithm for detecting objects representing potential hazards to drivers based on the combination of local information derived from optical flows and global information obtained from the host vehicle's status. The algorithm uses artificial neural networks to infer the degree of danger posed by moving objects in dynamic images taken with a vehicle-mounted camera. This approach allows more flexible adaptation of the algorithm to many drivers with dissimilar characteristics. Experiments were conducted with both miniature vehicles in a virtual environment and real vehicles in a real driving situation using video images of multiple moving objects. The results show that the algorithm can infer hazardous situations similar to the judgments made by human drivers. The proposed algorithm provides the foundation for constructing a practical driving assistance system

A Survey of Computer Vision Based Pedestrian Detection for Driver Assistance Systems

Computer vision based pedestrian detection has become one of the hottest topics in the domain of computer vision and intelligent vehicle because of its potential applications in driver assistance systems. It aims at detecting pedestrians appearing ahead of the vehicle using a vehicle-mounted camera, so as to assess the danger and take actions to protect pedestrians in case of danger. In this paper, we give detailed analysis of the diffculties lying in the problem and review most of the literature. A typical pedestrian detection system includes two modules: regions of interest (ROIs) segmentation and object recognition. This paper introduces the principle of typical methods of the two modules and analyzes their respective pros and cons. Finally, we give detailed analysis of performance evaluation and propose some research directions.

Detection of Wet-Road Conditions from Images Captured by a Vehicle-Mounted Camera

This paper discusses the detection of wet-road conditions based on images captured by cameras on the rearview mirror of a vehicle. Based on properties associated with water on a road, detection was stable for daytime and nighttime conditions. Water is recognized on the road based on polarization properties from images. Field tests verified detectability on an expressway at an average 100km/h, with favorable results.

A Study on Attention Guide Assistant System to Driver (Prediction Model for Driving Hazards)

This paper proposes an algorithm for detecting objects representing potential hazards to drivers, based on the combination of local information derived from optical flows and global information obtained from the host vehicle's status. The algorithm uses artificial neural networks to infer the degree of danger posed by moving objects in dynamic images taken with a vehicle-mounted camera. This approach allows more flexible adaptation of the algorithm to many drivers with dissimilar characteristics. Experiments were conducted with a model and real vehicles using video images of multiple moving objects. The results show that the algorithm can infer hazardous situations similar to the judgments made by human drivers. The proposed algorithm provides the foundation for constructing a practical driving assistance system.

Vehicle Lane-Tracking Control with Steering Torque Input

SUMMARYAlternative methods of automatic steering system are investigated by using a vehicle-mounted camera that detects lane markers ahead of the vehicle. The control input of such a steering system can be classified into steering angle and steering torque. This paper investigates the physical characteristics of automatic lane-tracking controller using steering angle and steering torque as the control input, and compares their differences. Generally, most of researches about lane-tracking control manoeuvre deal with steering angle control method due to its excellent robustness. In contrast, the steering torque control provides some degree of freedom in permitting the driver to steer the vehicle, thus it has potential for development as a steering assist mechanism. In this paper, the controller design is Linear Quadratic (LQ) controller based on the simplified dynamics of a 2 degree-of-freedom bicycle model and simplified steering model.

Dynamical Estimation of Camera Motion Parameters Using Focus of Expansion

A dynamical vehicle-mounted camera motion parameters' estimation method with the introduction of camera's Focus of Expansion (FOE) is presented here. Compared with other approaches like using line matching of road lane-marks or tracking the horizon, proposed method purely uses points matching between the adjacent frames without need of any previous knowledge or recognition results. This method is able to determine camera's rotation and translation parameters by using only 3 pairs of matching points and thus more suitable for real-time applications. Experiments on image sequences of real road scenes show the effectiveness and precision of proposed method.

LANA: a lane extraction algorithm that uses frequency domain features

This paper introduces a new algorithm, called lane-finding in another domain (LANA), for detecting lane markers in images acquired from a forward-looking vehicle-mounted camera. The method is based on a novel set of frequency domain features that capture relevant information concerning the strength and orientation of spatial edges. The frequency domain features are combined with a deformable template prior, in order to detect the lane markers of interest. Experimental results that illustrate the performance of this algorithm on images with varying lighting and environmental conditions, shadowing, lane occlusion(s), solid and dashed lines, etc. are presented. LANA detects lane markers well under a very large and varied collection of roadway images. A comparison is drawn between this frequency feature-based LANA algorithm and the spatial feature-based LOIS lane detection algorithm. This comparison is made from experimental, computational and methodological standpoints.

Distance and Clearance Perception Using Forward-Looking, Vehicular Television Systems

During off-road navigation, drivers often must make size, distance, and clearance judgments of terrain features and obstacles in order to choose safe navigational routes. These same judgments must be made in the teleoperation of land vehicles using forward-looking television systems. This study evaluated how well subjects perceived size, distance, and clearance using monochrome and color television systems. Thirty-eight subjects (Ss) estimated the size, distance, and separation of two obstacles using video imagery produced by a forward-looking, vehicle-mounted camera. Results indicate that Ss typically overestimated distances, and when in error judging clearance, tended to overestimate the gap between the objects. These biases were expected due to the well documented minification effect of television using standard lenses. More surprising was the somewhat larger bias evidenced by Ss using color versus those using monochrome imagery. This paper will describe the research methodology, the results obtained, potential reasons for the results, discuss the plans for continuing research in this area, and assess the implications these findings may have on teleoperated vehicle design.

Shelter, infrastructure and services in third world cities

Accurate, current and complete information is indispensable for any effective intervention to upgrade a slum or informal settlement. Our involvement in supporting informal settlement upgrades with geospatial information prompted us to investigate what we could learn from literature to improve the collection and representation of geospatial information for such projects. For this paper, we systematically reviewed literature about informal settlement upgrading interventions for which geospatial information was used. We classified the geospatial information into three categories – physical, socio-economic, and boundaries – based on the phenomena they described, and categorized the methods of data acquisition. The results show that most studies collect geospatial information that enumerates and measures empirically observable characteristics (e.g, structures, infrastructure, utility services, mobility network, land description) and characteristics of occupants of informal settlements (e.g., socio-economic status). Fewer studies collected geospatial information about social networks and social ties. Data collection methods ranged from primary sources such as paper-based surveys, handheld/mobile GPS devices, vehicle-mounted cameras, etc., to secondary sources such as spaceborne, airborne and web-based platforms. Few studies made use of unmanned aerial vehicles (UAVs), despite their recent popularity as source of base maps, but we expect this to change in the near future. The results can be used to inform data collection strategies for informal settlement upgrades. Policy makers and other stakeholders involved in informal settlement upgrades can benefit from a single source of knowledge about such information.