Human behavior analysis is an important challenge in many domains, such as surveillance systems, video content retrieval, human interactive systems, medical diagnosis, etc. With the increasing needs of public safety, intelligent surveillance system becomes an activating issue in computer vision and related research fields. In this thesis we present a method to analyze human behavior in a video sequence with depth information obtained from the depth camera. When interested actions are detected in the scene, the system will trigger alarm information. Contour line and Delaunay triangulation are used to establish human posture model. By traversing the triangulation meshes with the depth first search, we obtain the spanning tree with the depth information, and then construct human posture model with this spanning tree. Posture sequence from video sequence with corresponding posture models can be obtained, and then the posture sequences is clustered into key posture sequence. By querying the key posture sequence, the system can recognize human behavior in real-time and inform users immediately when interested actions detected. Experimental results show that the system is accurate and robust for human behavior recognition.

A real-time Vehicle Make and Model Recognition (VMMR) system is a significant component of security applications in Intelligent Transportation Systems (ITS). A highly accurate and real-time VMMR system significantly reduces the overhead cost of resources otherwise required. In this thesis, we present a VMMR system that provides very high classification rates and is robust to challenges like low illumination, occlusions, partial and non-frontal views. These challenges are encountered in realistic environments and high security areas like parking lots and public spaces (e.g., malls, stadiums, and airports). The VMMR problem is a multi-class classification problem with a peculiar set of issues and challenges like multiplicity, inter- and intra-make ambiguity among various vehicles makes and models, which need to be solved in an efficient and reliable manner to achieve a highly robust VMMR system. To reliably overcome the ambiguity challenges, a global features representation approach based on the Bag-of-Features paradigm is proposed. We extract key features from different make-model classes in an optimized dictionary, through two different dictionary building strategies. We represent different samples from each class with respect to the learned dictionary. We also present two classification schemes based on multi-class Support Vector Machines (SVMs): (1) Single multi-class SVM and (2) Attribute Bagging-based Ensemble of multi-class SVMs. These classification schemes allow simultaneous learning of the differences between global representations of different classes and the similarities between different shapes or generations within a same make-model class, to further overcome the multiplicity challenges for real-time application. Extensive experiments conducted using our approaches yield superior results for images that were occluded, under low illumination, partial camera views, or even non-frontal views, available in a recently published real-world VMMR dataset. The approaches presented herewith provide a highly accurate VMMR system for real-time applications in realistic environments.