08-16-2017, 08:46 PM
Visual Tracking for Intelligent Vehicle-Highway Systems
The complexity and congestion of current transportation systems often produce traffic situations
that jeopardize the safety of the people involved. vision sensors provide information that is richer and more complete than other sensors. In this article, detection and tracking techniques for intelligent vehicle-highway applications is proposed where computer vision plays a crucial role. Controlled Active Vision framework is described in detail. It can be used for providing a visual tracking modality to a traffic advisory system.
Detection and Tracking in a Traffic Vision System:
constructing visual tracking modalities for intelligent
vehicle-highway systems requires the consideration of many elements viz:
1)Detection of Traffic Objects of Interest:
Given an arbitrary image, the success of tracking an object such as a pedestrian relies on
the assumption that we somehow are able to detect that the tracked feature windows correspond to
the object in question.an image is considered to be comprised of pixels that are
in one of two categories: figure or ground.ground pixels belong to the objects environment whereas the figure objects belong to the objects of interest. the identification and the analysis of figure pixels in each image of the temporal sequence is done foe the detection.
2)Visual Measurements:
an IVHS sensing modality capable of measuring motion in a temporal sequence
of images is aimed at. The visual measurements are
combined with search-specific optimizations in order to enhance the visual processing from
frame-to-frame and to optimize the performance of the system in our selected applications.
Selection of Features to be Tracked:
we have
reduced the problem of locating trackable features in the entire image to the problem of locating
trackable features in a smaller, rectangular region By computing a bounding box around a traffic object of interest. Here, an assumption is made that most possible feature selections made from within a bounding box either lie entirely within the object or contain at least a portion of the object s pixels.
For more details, read: