Tracking Algorithms For High Dimensional Image Texture

Image tracking is a major problem in image processing and also a hard problem to solve. Lots of efforts have been made and plenty of algorithms have been developed according to different kinds of research background. However, there is no general solution and universal theory till now. As the bio-medical image acquisition techniques continue to improve, medical image tracking has drawn more and more attentions from researchers. Due to the high volume of generated microscopic image data, it is critical to develop an automated technique for robustly and rapidly processing and analyzing these problems. Different from natural image and video data, there is always certain prior information and content constraints in specific medical image. How to utilize there important information is very important in solving these problems.In this paper, we firstly review some mainstream centerline extraction and multi objects tracking algorithms in medical imaging area. Then two scenarios under the high-dimensional image of the tracking algorithm have been studied.Axon centerline extraction in 3D confocal microscopic image stack is firstly talked about. Neuron axon analysis through confocal microscopic image stack is dedicated in visualizing the geometrical features and topological characteristics of the 3D tubular biological objects, to ascertain the morphological properties and reconstruct the connectivity of neurons. This paper proposes a new curvilinear tracking algorithm which initializes a superellipsoid kernel into the tube by fitting the intensity energy distribution with multiple scales of steps and radii other than Hessian kernel. It is herein solved as an energy optimization in a graphical model with maximum likelihood, which preserves an equilibrium distribution across all the nodes with an attenuation penalty of orientation transition. Local potential energy diffusion of different axons is tracked by dynamic priority and pruning inference, to solve the cross-over problem. The centerline could be semi-automatically extracted following the selected initial point. Experimental results on 3D axon volumes verify that the proposed approach can handle complicated axon structures without elaborate segmentation.Image-based high-throughput vertebrate system experiments are increasingly carried out to facilitate disease modeling and drug discovery. Zebra fish is by far the best genetic system of choice for developmental analysis of vertebrates. However, tremendous amount of images generated from large number of zebra fish have become a bottleneck. This paper presents a fully automated image analysis for cell segmentation and tracking algorithm. Initially, cells are denoised and segmented according to a newly formulated energy function based on newly introduced surface area energy. Our proposed method performs better than classical 3D level set curve evolution methods in segmenting cells with extremely low intensity value and touching cells with significant intensity variance. Hereafter, a three-dimension graph-based multi-objects tracking technique is imposed on identification of zebra fish cells in time-lapse image sequences. Our proposed method takes full advantage of neighboring relationships compared with those previous approaches in literature. Graph regulation measure is taken to alleviate the high complexity in graph construction caused by 3D space. New vertex match criterions are also proposed. The 3D image dataset is synthesized to validate our proposed method by experimental results.