Dissertation > Excellent graduate degree dissertation topics show

Research and implementation of real-time monitoring of population density estimates in complex scenes

Author: XiePengCheng
Tutor: WuYue
School: University of Electronic Science and Technology
Course: Computer Software and Theory
Keywords: Crowd density estimation Non-parametric background modeling Edge detection Texture analysis SVM
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 105
Quote: 0
Read: Download Dissertation

Abstract


With the rapid development of urbanization and economic, activities of large populations such as entertainment activities, exhibitions, sports events and celebrations are held increasingly frequent. As a result, the safety of these public events draws a high attention of the security department and the crowd density estimation in the security monitoring has become a research focus in the field of computer vision and digital image processing.The focus of this thesis is to research on the methods of crowd density estimation in a real-time monitoring of the outdoor. After analysis of the present estimation methods at home and abroad, we propose a scheme combined pixel-based method with texture-based method to estimate crowd density respectively in low and high-density scenes. In this scheme, we get the moving objects by means of image segmentation, with a background obtained by an adaptive background model. After removing shadows and noises in the moving objects, the scheme calculates the area of objects. If the area is less than a threshold, it’s taken for a low-density scene. Otherwise, it’s a high-density scene.In low-density scenes, we apply edge detections to the extracted objects and calculate the number of edge pixels. Based on the theory of least-square curve fitting, we construct a linear equation to present the relationship between the crowd number and the edge pixel number. Thus, with this linear equation we can estimation the number of crowd after calculating the number of edge pixels of objects.In high-density scenes, we classify the density to three levels of high, very high and extremely high. GLCM is used to analyze the texture and extract texture features from images. Then use PCA to analyze and choose the four most important features as the final texture features. Finally with these feature data we train a SVM used to classify the density level.The result of experiments has shown that it demands a robust background modeling and real-time algorithms to achieve real-time intelligent crowd density estimation of the outdoor. In this paper, we adopt a robust background modeling which is based on ECD non-parametric background model in the outdoor complex scenes and obtain a clear background image. And binary operation is also used to extract moving objects. In addition, morphological processing of the detected edge and PCA-based feature extraction contribute a lot to the performance of real time.

Related Dissertations

  1. Soft Sensor of Naphtha Dry Point on Support Vector Machines Regression,TE622.1
  2. The Research of the Fault Diagnoses Algorithm for the Liquid Rocket Engine Testing Bed Based on PCA-SVM,V433.9
  3. ISAR Imaging Simulation of Space Targets and Target Recognition Based on ISAR Images,TN957.52
  4. Research on Autamatic Music Structrue Analysis,TN912.3
  5. Research on Feature Extraction and Classification of Pulse Waveform for Cholecystitis and Nephrotic Syndrome Diagnosis,TP391.41
  6. Research on Classification Method of Tongue Substance Color and Tongue Coating Color Based on SVM,TP391.41
  7. The Research on Paper Currency Classification Method Based on Harr-Like Feature and Minimal Ball Including Samples,TP391.41
  8. Research on Detecting Optical Fiber Geometric Parameter Based on Machine Vision,TN253
  9. Research on Predicting Intrinsic Disorder Protein Structure Based on Supervision Manifold Learning Algorithm,Q51
  10. Study on the Road Condition Monitoring Based on Vehicular 3D Acceleration Sensor,TP274
  11. Research of Orange Quality Classification Technology Based on Computer Vision,TP391.41
  12. Analysis on Synoptic Climatology Characteristics and Forecast Methods of Fog in Hainan,P457
  13. Study on Visual Target Detection Based on SVM,TP391.41
  14. Research on Relevance Feedback Techniques Based on Long-term Log Learning and RW-Soft SVM in CBIR,TP391.3
  15. Study on Automatic Classification Method for Remotely Sensed Imagery by Incorporating Spatial-Spectral Features,TP751
  16. Multi-step-ahead Stock Price Index Forecasting Based on Hybrid Models,F224
  17. Research on ECG Feature Extraction and Classification Method,TN911.7
  18. Electric vehicle drive control system research,U469.72
  19. Image intensifier and a screen resolution of defect detection technology,TN144
  20. Analog Circuit Fault Diagnosis Method and Its Application,TN710
  21. Research on Feature Selection for Gene Expression Data,Q78

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