Surflet-pair-relation histograms: a statistical 3D-shape representation for rapid classification

Abstract

A statistical representation of three-dimensional shapes is introduced, based on a novel four-dimensional feature. The feature parameterizes the intrinsic geometrical relation of an oriented surface-point pair. The set of all such features represents both local and global characteristics of the surface. We compress this set into a histogram. A database of histograms, one per object, is sampled in a training phase. During recognition, sensed surface data, as may be acquired by stereo vision, a laser range-scanner, etc., are processed and compared to the stored histograms. We evaluate the match quality by six different criteria that are commonly used in statistical settings. Experiments with artificial data containing varying levels of noise and occlusion of the objects show that Kullback-Leibler and likelihood matching yield robust recognition rates. The present study proposes histograms of the geometric relation between two oriented surface points (surflets) as a compact yet distinctive representation of arbitrary three-dimensional shapes.

Shape-Based Autotagging of 3D Models for Retrieval

“…As the feature vector, we used the SPRH by Wahl [15] modified to accept polygon soup and polygonal mesh models found in the PSB. Note that the performance of the SPRH is only modest by current state of the art.…”

“…Depending on an application, additional invariance, such as invariance against joint articulation, may be required. For the experiment described in this paper, we used our extension of Wahl's Surflet Pair Relation Histograms (SPRH) [15], a shape feature that has nice invariance properties, readily available, and has reasonable retrieval performance. The Windows 32bit executable of our extension of the SPRH is available at our web site [8].…”

Colour Space Effect on Tracking in Video Surveillance

“…In addition to the combination of utilizing features and parameters, other methods of tracking are utilized for tracking purposes. The different measures used for tracking include properties of the tracked object which cover Manhattan Distance [40], Mahalanobis Distance [31,34], shape tracking [1,32], histogram similarity index [41] and consistent labeling [42]. Other methods of tracking utilize complex statistical and intelligent methods which were Bayesian Network [11,12] and neural network [43].…”

“…Han et al [43] Grayscale Tracking uses neural network to maintain tracking of target object. Wahl et al [41] Grayscale Tracking is based on similarity index calculated from image histogram. Chang et al [6] Color Multiple cameras track based on apparent height and color of tracked target.…”

“…Apart from the tracking methods described, other methods of tracking utilize color information as a means for determining the similarity between objects that is being tracked. The measure of similarity between a reference object and a tracked object can be measured by measures of Mahalanobis distance [34], Bhatacharrya distance [49,50], color distribution [50] and histogram similarity index such as histogram intersection, Squared Euclidian Distance, Chi-Squared and KullbackLeiber [41,[49][50][51]. Based on the different color spaces, different tracking methods were developed based on the different information available on the color spaces.…”