Object Tracking Using New Hybrid Evolutionary Algorithms

Description

Object tracking is a fundamental problem in computer vision and image processing with numerous applications including visual surveillance, video communication and compression, navigation, display technology, high-level video analysis, traffic control, metrology, video editing, augmented reality and human-computer interfaces to medical imaging, and so on. Despite significant progress in past decades, it remains a challenging task due to large appearance variations caused by Scale Variation, Out-of-View, Fast Motion, Occlusion, Motion Blur, In-Plane Rotation, Low Resolution, Deformation, Out-of-Plane Rotation and Background Clutters. Object tracking in the image can be regarded as a process of searching for the most similar candidate region of the target by an efficient target representation. Therefore, a robust appearance model and an efficient search strategy are of crucial importance for a tracker. With the fast improvement of optimization algorithms, specifically, the metaheuristic algorithms, more intelligent searching techniques are used to resolve object tracking problems based on optimization algorithms by many researchers. Optimization based tracking approaches make no presumptions about the noise or the type of the distribution in the tracking system. Therefore, optimization based tracking approaches enable potential tracking approaches in challenging environments. The development of new hybrid optimization algorithms that robustly track an arbitrary target in various challenging conditions is the aim of this thesis, which consists of five parts. Due to its simplicity and efficiency, a recently proposed optimization algorithm, Sine Cosine Algorithm (SCA), has gained the interest of researchers from various fields for solving optimization problems. However, it is prone to premature convergence at local minima. To overcome this drawback, a novel Hybrid SCA-DE, Hybrid SCA-TLBO and Hybrid SCA-PSO algorithms for solving optimization problems and object tracking are proposed. To validate the tracking capability of proposed trackers, the tracking acts of Particle filter, Mean-shift, Scaleinvariant feature transform, Particle swarm optimization (PSO), Bat algorithm (BA), Sine Cosine Algorithm (SCA) and Hybrid Gravitational Search Algorithm (HGSA) are studied comparatively. To validate the tracking capability of proposed trackers, the tracking acts of Particle filter, Mean-shift, Scale-invariant feature transform, Particle swarm optimization