Dissertation > Excellent graduate degree dissertation topics show

Research and Implementation of HOG Based Human Detection in Image

Author: ZhouKe
Tutor: WangTianJiang
School: Huazhong University of Science and Technology
Course: Applied Computer Technology
Keywords: Human Detection Histogram of orientation gradient Support Vector Machine Cascade
CLC: TP391.41
Type: Master's thesis
Year: 2008
Downloads: 1331
Quote: 14
Read: Download Dissertation

Abstract


Human detection is to study how to get the computer to identify the area of ??human subjects from the image or video technology to people's way of thinking. It can be widely used in national defense, public security, and electronic games, etc.. To go through a series of steps based on the classification of the human body detection. First, the detection window of the different sizes of the input image is a progressive scan; classify the second step, the detection window; Finally, in general, the detection of a plurality of detection results corresponding to a human body objects, so they must be detected The results for fusion to obtain the final detection result. The classifier is good or bad is closely related with the selected feature classifier training. Optional feature in the image is very rich, including color, brightness, as well as some local image characteristics, etc.. Due to the ever-changing body image brightness and color of the object, therefore the most suitable characteristics in the human body detection is able to reflect the direction of the body contour gradient histogram. Its extraction process includes a color space standardized, gradient calculation, the projection, on the direction and spatial as well as the overlapping blocks of the high-quality standardization. Another factor affecting the classifier effect classification algorithm, support vector machine as a binary classification, easy to use, so widely. Both of these factors can be achieved based on the direction of the gradient histogram and support vector machine human detection algorithm. Feature extraction, when there is a lot of duplication of the region need to calculate Integral strike direction of the gradient histogram features. Integral way to greatly shorten the time of feature extraction, and to avoid the disadvantage of feature space can not obtain the global image information. Classification algorithm can be used to cascade several layers of support vector machine series, detection can greatly speed up the the feature matching rate will also increase, but the time and calculate the amount of training the classifier. Fusion results of a variety of ways to maximize convergence algorithm through the detection results on the map to the three-dimensional space, and the assessment of the results of density optimal solution can be found in these results, test results and clarity is an effective fusion algorithm. Accelerate feature based on human detection algorithm based on the direction of the gradient histogram features can detect the different posture upright in the complex context of the human body; human detection algorithm based on integral direction gradient histogram features and cascading framework while maintaining the detection accuracy extraction and classification speed, real-time detection was more.

Related Dissertations

  1. Research on Automatic Detection Algorithm for Substructure Distress of Highway Pavement Based on SVM,U418.6
  2. The Flow Theroy Investigation of Compressor Cascade with Split,TK474.8
  3. Research on Autamatic Music Structrue Analysis,TN912.3
  4. Research on Transductive Support Vector Machine and Its Application in Image Retrieval,TP391.41
  5. Fault Diagnosis Method Based on Support Vector Machine,TP18
  6. Process Support Vector Machine and Its Application to Satellite Thermal Equilibrium Temperature Prediction,TP183
  7. Research on Temperature Control System for High-Performance Oven-Control Crystal Oscillator (OCXO),TN752
  8. Research for Infrared Image Target Identification and Tracking Technology,TP391.41
  9. Study on the Road Condition Monitoring Based on Vehicular 3D Acceleration Sensor,TP274
  10. Research of Diagnosing Cucumber Diseases Based on Hyperspectral Imaging,S436.421
  11. The Research on Intrusion Detection System Based on Machine Learning,TP393.08
  12. Research on Improved K Neighbor Support Vector Machine Algorithm Faced Text Classification,TP391.1
  13. Research on Face Recognition Based on AdaBoost Algorithm,TP391.41
  14. Intelligent Control System of Chlorine Dosing in Waterworks Based on Embedded System,TU991.2
  15. Research on Feature Extraction, Selection and Classification Algorithms for Pulmonary CAD,TP391.41
  16. Research on Subimage Selection and Mathching Method for Synthetic Aperture Radar(SAR) Target Recognition,TN957.52
  17. Research of Facial Expression Recognition Algorithm,TP391.41
  18. Fundus Image Segmentation Based on SVM and Template Matching,TP391.41
  19. Research and Realization of License Plate Character Recognition Algorithm Based on SVM,TP391.41
  20. Cloud Classification Based on Geostationary Meteorological Satellite Imagery,TP391.41
  21. Optical matrix embedded controller software design and development,TP311.52

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