Welcome to Our Website Join Free | Sign In
View Story
Title :
Jhon Caceres
Posted On :
Tuesday November 4th,2008 13:02
Name :
Posted by :
Canon [ Bronze ]
Jhon Caceres PhD Student, Department of Civil and Coastal Engineering Geosensing Systems Engineering Weil Hall Suite 500C My current research is focused on 3D urban modeling and classification of urban infrastructure and natural objects by using lidar point data. 3D modeling has become an important application in urban planning, environmental monitoring and civil engineering. In modeling, classification, segmentation and recognition using laser point clouds, automation is the main point of research. Classification of lidar point data is a process that usually is done as a preprocessing step for posterior object modeling over the classified data. As such, this classification is a critical step in the modeling process. Several approaches for segmenting and classifying ALSM data have been proposed. Among them, region growing, clustering, vertical profile based segmentation and slope-based segmentation are the most common. My classification approach is based on a supervised learning scheme that acquires knowledge from the training data statistics. When a statistical supervised learning approach is used, the selection of features and estimation of the class-dependent probability density functions (pdf), i.e. likelihoods, are key aspects for the algorithm performance. In our case, spin images [1] are used for object feature extraction. For pdf estimation, non-parametric methods are used. Weil Hall area at the University of Florida
Sample Banner