Research On Object Detection And Tracking Based On Double-Density Dual-Tree Complex Wavelet Transform

The intelligent video surveillance system can detect, track, analysis and understand the object automatically, and then the system can provide the key information which is useful to surveillance and warning. As an important aspect of intelligent video surveillance, the object detection and tracking have become the important subject in the field of pattern recognition, image processing and computer vision. In this dissertation the key technology of object detection and tracking is discussed and the algorithm of object detection and tracking based on double-density dual-tree complex wavelet transform is proposed. The proposed algorithm is of academic value and practical value. The main research works in this dissertation are as follows:①The research progress, present situation at home and abroad of object detection and tracking technology are reviewed, the common methods are also summarized. The fundamental principle, filter bank structure and implementation of the discrete wavelet transform(DWT), the dual-tree complex wavelet transform(DT CWT), the double density discrete wavelet transform(DD DWT) and the double-density dual-tree complex wavelet transform(DD-DT CWT) are studied. Furthermore, the characteristics of DD-DT CWT, such as approximately shift invariance and good directional selectivity, are summarized.②The application of wavelet transform in the field of image de-noising and the quality evaluation index are introduced. A local adaptive image de-noising algorithm based on DD-DT CWT is proposed. Firstly, the degraded images are decomposed using DD-DT CWT. Secondly, according to the statistical properties of wavelet coefficients and the dependency of inter-level and intra-level coefficients, the bivariate shrinkage function with local variance estimation is adopted to processing wavelet coefficients. Finally, the de-noised images are synthesized using the wavelet coefficients. The experimental results indicate that the proposed algorithm is more effective in noise removal and edge reservation than others, meanwhile, the visual quality of the de-noised images is improved.③The characteristics of infrared images are described. The ideas and common algorithms of human detection in infrared images are analyzed. A novel object detection method in infrared images based on DD-DT CWT and wavelet entropy is studied in this dissertation. The candidate object regions are located by using a two-level directional projection method based on the brightness. The feature extraction process of wavelet entropy using DD-DT CWT is studied. The basic principle of support vector machine is described, taking the pattern recognition of two classes object for example. The steps of human detection in infrared images are discussed in detail. The experimental results show that the proposed algorithm can detect human accurately, meanwhile, the detection rate is improved while the false alarm rate is decreased significantly.④The development and application of particle filter are simply introduced. The fundamental principle and the derivation of particle filter are analyzed. After discussing the object tracking algorithm based on particle filter, an object tracking method in infrared images is studied, which takes feature of wavelet entropy using DD-DT CWT as object representation model. The results indicate that the robustness and feasibility of the tracking system using the proposed algorithm are improved. The algorithm can solve human tracking problems in different backgrounds and different states.