Study of Clustering In Moving Objects by dandanhuanghuang


We are surveying on

 Clustering in Moving Objects
 Discovering clusters in spatio-temporal data
 Tracking Objects based on color cluster flow

 Study of the various clustering algorithms

 Real-life Applications of clustering
 Video Surveillance
 Soccer Game Records
 Mobile Objects
 Hospital Laboratory Examination
  Motivation increasingly attractive to the data
 Moving objects are becoming
  mining community due to continuous advances in technologies.

 clustering is the assignment of a set of observations into
  subsets (called clusters) so that observations in the same
  cluster are similar in some sense

 cluster analysis is a good way for quick review of data

 Clustering is used in many fields like Data Mining, Pattern
  recognition, Image analysis, information retrieval and

 As clustering ahs vast uses in our daily life's We feel this is an area
  that needs to be explored.
  Types of Clustering
 Hierarchical algorithms find successive clusters using previously established
 Agglomerative ("bottom-up") - begin with each element as a separate cluster
  and merge them into successively larger clusters.
 Divisive ("top-down") - begin with the whole set and proceed to divide it into
  successively smaller clusters.

 Partitional algorithms typically determine all clusters at once

 Density-based clustering algorithms are devised to discover arbitrary-
  shaped clusters. A cluster is regarded as a region in which the density of data
  objects exceeds a threshold. Two typical algorithms of this kind.

 Subspace clustering methods look for clusters that can only be seen in a
  particular projection of the data. These methods thus can ignore irrelevant
Retrieval of moving Clusters
 MC1, is a straight forward implementation of the
 problem definition.

 MC2, improves the efficiency by avoiding redundant

 MC3 is an approximate algorithm which trades
 accuracy for speed.
Analysis of Clustering Algorithms
 There is a need to evaluate various clustering algorithms and
  identify the ones which are expected to give good results for

 A criterion that can be used for evaluation is stability.

 A clustering algorithm is said to be stable if its output i.e. the
  clusters it produces, do not change drastically when small
  changes are made to the input data.

 In this paper, we compare the stability of six hierarchical
  clustering algorithms by carrying out experiments on three
  open-source software systems.
Time-Focused Density-based Clustering
 In this paper, authors consider the clustering problem applied to the
  trajectory data domain.

 A density-based clustering method for moving objects trajectories was
  proposed in this paper

 A set of experiments on synthesized data is performed in order to test the
  algorithm and to compare it with other standard clustering approaches

 a new approach to the trajectory clustering problem, called temporal
  focusing, is sketched, having the aim of exploiting the intrinsic semantics
  of the temporal dimension to improve the quality of trajectory clustering.
A Clustering Based Approach For Discovering
Interesting Places in Single Trajectory
 Algorithm CB-SMoT (Clustering-Based Stops and Moves of Trajectories)
  is based on a traditional algorithm DBSCAN which is a classical density-
  based clustering approach.

 There is an important parameter Eps in the algorithm CB-SMoT, whose
  value can dramatically affect the quality of clustering.

 In this article author propose a new method to calculate the Eps value.

 The experiment proves that using this method to calculate the parameter
  values can significantly improve the quality of clustering.
  Clustering Of mobile Objects
 In this paper, new ways for summarizing spatio-temporal data as trajectories
  were presented, including
 a new similarity measure between trajectories
 an algorithm for recognizing clusters of similar moving objects by
  incremental clustering methods.

 The algorithm is evaluated empirically by the
 quality of object clusters (using Dunn and Rand indexes)
 memory
 space efficiency
 execution times
 Scalability

 The proposed incremental algorithm for clustering moving objects is time
  efficient and quality improving technique for recognizing groups of
  similar moving objects.
   Tracking Moving Objects Based on
   Color Cluster Flow
 Author proposed a new algorithm for tracking non-rigid objects, which is based on
  clustering in the color/position feature space.

 Determine a set of clusters in the first image. These clusters are used for a parallel k-
  means clustering, which performs the tracking of clusters over a sequence of

 The tracking algorithm proved to be robust with respect to shape variations and
  partial occlusions of the objects.

 The trajectories of the cluster centroids, even for objects with large shape variations,
  suggests that object detection based on segmentation of the CCF will be possible.

 The algorithm can also be used in stereoscopy, where matching of image features in
  pairs of images is a similar problem to tracking features over two consecutive images
  in a temporal sequence.
Clustering Analysis of Data on Hospital
Laboratory Examination
 Laboratory tests is a challenging task because of the temporal irregularity
  of data, coexistence of various events and multidimensionalty of

 In this paper, cluster analysis method for multidimensional,
  inhomogeneous time series based on the trajectory comparison

 Application to the chronic hepatitis dataset delivered some interesting
  findings, for example, there existed some patterns in ALB-PLT
  trajectories that took similar temporal courses, and the clusters well
  corresponded to the fibrotic stages.
Application to Soccer Game
 Authors developed a cluster analysis method based on multiscale
  matching and rough, clustering, which may build a new scheme of
  sports data mining.

 Experimental results on real soccer game records demonstrated that
  the method could discover some interesting pass patterns that may be
  associated with successful goals.

 Although the experiments are in the preliminary stage and subject to
  further quantitative evaluation, the proposed method demonstrated
  its potential for finding interesting patterns in real soccer data.

 The future work will include the use of ball speed,use of other feature
  points than inflection points, and optimization of segment difference
Application to Video
 In this paper Author present a trajectory clustering method suited for
  video surveillance and monitoring systems.

 The clusters are dynamic and built in real-time as the trajectory data is
  acquired, without the need of an off-line processing step.

 The trajectories are represented as trees of clusters

 Authors show how the obtained clusters can be successfully used both to
  give proper feedback to the low-level tracking system and to collect
  valuable information for the high-level event analysis modules.
Spatio-temporal data mining is still in its infancy, and even the most basic
  questions in this field are still largely unanswered

 what kinds of patterns can be extracted from trajectories?

 Which methods and algorithms should be applied to extract them?

 Trajectory clustering is only a little attempt.

 How to design the metric space of the similarity

 what methods can be used in clustering trajectory
 Wikipedia

 Trajectory clustering with mixtures of regression models Scott Gaffney and
  Padhraic Smyth Technical Report University of california

 D. Chudova, S. Gaffney, E. Mjolsness, and P. Smyth,
  "Translation-invariant Mixture Models for Curve Clustering,“ Proc. The 9th ACM
  SIGKDD International Conference on Knowledge Discovery and Data Mining,
  2003, pp. 79-88.
 S. Hirano and S. Tsumoto, "A Clustering Method for Spatio-temporal Data and Its
  Application to Soccer Game Records," Proc. the 5th International Conference on
  Hybrid Intelligent Systems, 2005, pp. 612-621

 P. Kalnis, N. Mamoulis, and S. Bakiras, "On discovering moving clusters in spatio-
  temporal data," Proc. the 9th International Symposium on Spatial and Temporal
  Databases, 2005, pp.
 Y. Li, J. Han, and J. Yang, "Clustering Moving Objects," Proc.
  the 10th ACM SIGKDD International Conference on Konwledge DIscovery and Data
  Mining, 2004, pp. 617-622.


 Cluster Analysis of Trajectory Data on Hospital Laboratory Examinations Shoji
  Hirano, PhD, Shusaku Tsumoto, MD, PhD

 Incremental Clustering of Mobile Objects Sigal Elnekave, Mark Last, Oded Maimon

 Time-focused density-based clustering of trajectories of moving objects Margherita
  D’Auria1, Mirco Nanni2, and Dino Pedreschi1

 Trajectory Clustering and its Applications for Video Surveillance C. Piciarelli G. L.
  Foresti L. Snidaro {piccia,foresti,snidaro}

To top