Dissertation > Excellent graduate degree dissertation topics show

Abandoned Luggage Detection Algorithm Based on Background Subtraction in Surveillance Videos

Author: NingWenZuo
Tutor: LinXingGang
School: Tsinghua University
Course: Information and Communication Engineering
Keywords: abandoned luggageGaussian Mixed Model Multi-layer CodebookModel Histogram of Oriented Gradientshuman detection
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 53
Quote: 0
Read: Download Dissertation

Abstract


Abandoned luggage is a security risk for public places. Bombs hidden inabandoned luggage have explored in many places. Due to the wide use ofsurveillance cameras in recent years, watching videos and distinguishanomalous events is challenging work for monitoring personnel. Hence,intelligent detection becomes a necessary function for surveillance systems.Abandoned luggage detection is an important research subject.Researchers have proposed many algorithms for luggage detection.However, some of them are not robust to adapt to actual monitoring scenes,while others are not efficient enough for real-time detection. To solve theproblem, this paper proposes two novel frameworks respectively base onGaussian Mixture Model (GMM) and Multi-layer Codebook Model (MCBM),combined with various image features and existing algorithms.The first abandoned luggage detection method uses a GMM whose updatestrategy has been modified to detect the foreground regions of the surveillancescene. Then a time domain filter is applied to obtain the abandoned luggageregions. In addition, a motion region filter is utilized to remove the influenceof moving pedestrians,and a linear illumination compensation model is usedto eliminate the light change in complex scenes, which improve the robustnessof the system.The abandoned luggage detection method based on MCBM uses a fourlayers structure model to detect the stable foreground of the scene assuspicion regions. Then a new mechanism combined with HOG features isdesigned to eliminate the influence of illumination change. At last, a humandetector based on hierarchical HOG matrices is applied to eliminate thefalsealarms caused by pedestrians and eventually to obtain the abandonedluggage regions. The update process of MCBM is optimized and the HOG datais used for eliminating both illumination change and pedestrians. As a result, the algorithm is with low computing complexity.The experiments demonstrate two algorithms both achieve goodperformances on PETS2006and i-LIDS data. They also present favorablerobustness in complicated scenes. Moreover, the stability and computingspeed are tested on a piece of surveillance video with long period. The twomethods are time efficient and reach the speed of19and67frames per secondunder the resolution of320×240.

Related Dissertations

  1. Research on the Technology of Waiting Pressure When Passing Through Abandoned Roadway in Fully-Mechanized Face with Anual Coal Production of 10 Million Tons,TD823
  2. Research of Abandoned Objects and Their Owner Detection Based on Omni-Directional Vision,TP391.41
  3. Research on Techniques in Video Surveillance,TP391.41
  4. Research and Implementation of Human Abnormal Behavior Analysis in Video Surveillance,TP391.41
  5. Pedestrian Tracking and Identifying in Video Surveillance,TP391.41
  6. Exam cheating Intelligent Analysis Method,TP391.41
  7. Analyzing on the Willingness of Farmers in Conversion of Cropland to Forest (Grassland),F224
  8. Research and Implementation on Key Technologies of Intelligent Surveillance System of Object Safty Detection,TP277
  9. Study for Sequencing of Mixed-Model Production Lines in FAW-VW CP2 Body Shop,F426.471
  10. Study on Optimal Sequence of a Mixed-model Assembly Line with Stochastic Operation Times,F273
  11. The goaf 3D laser scanning data analysis and processing,P234.4
  12. Study on Vegetation Restoration Technology of Abandoned Sandy Fields,S651
  13. Research on the Socialization of the Children Being Abandoned by the Trafficking of Women in Countryside,D669
  14. Discussion of the Protection on the Interests of Socially Vulnerable Groups under the System of Comfortable Housing Project,F293.3
  15. Storm’s Dominant Role on the Depostition and Erosion Evolution in the Yellow River Delta,P731.23
  16. Impacts of Different Disturbances on the Vegetation Restoration of Abandonded Farmland,Q948
  17. Risk Assessment on Abandoned Sites,X820
  18. Spatial Heterogeneity of Soil Nutrients and Its Effect on Plants of Old-fields in Loess Region,S158.3
  19. Research into the Crime of Abandonment,D924.3
  20. Discard file Phenomenon,G275.9
  21. The Absorbing Capacity of Liquidambar Formosana and the Soil Remodeled of Heavy Metal Pollution,X53

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile