Human tracking has attracted much attention from the researchers in the fields of computer vision and pattern
recognition. The problem is generally extremely challenging partly because human bodies are articulated and versatile,
and partly because background clutter, both of which demand a strong human model. However, there is usually a trade-off
between the discriminative power and the complexity of a given model. This paper presents a simple yet distinctive
appearance model for real time human tracking by exploiting the pairwise constraints between parts. The parts in our
model are generated online by sampling the foreground of the scene into overlapping blocks and grouping them into
appearance coherent parts with mean shift algorithm. Constraints between the resulting parts are defined and used to
encode the structure of human body. To tolerate the possible human deformations and occlusions, the model is layered.
With this model, we design an algorithm for human tracking and test its performance on real world image sequences.
Experimental results show that the proposed appearance model although simple, has enough discriminative power to
classify multiple humans even in presence of occlusions and the associated tracking method can run in real time.
This paper presents an adaptive part-based probabilistic model for non-rigid object tracking. Without any assumption on scenes or poses, our model is online generated and updated. The parts in the model are extracted by clustering based on the appearance consistency of local feature descriptors in the object. A probability indicating the possibility of a part belonging to the object is then assigned to each part and adapted during tracking. We also propose a fully automatic algorithm for single object tracking with model matching and adaption. Our approach is evaluated on three different datasets and compared with previous work on visual tracking. The experimental results showed that our approach can track non-rigid object under occlusion and object deformation effectively in real time. Moreover, it works even if the target is partially occluded at initialization step.
Image segmentation is a classical and challenging problem in image processing and computer vision. Most of the segmentation algorithms, however, do not consider overlapped objects. Due to the special characteristics of X-ray imaging, the overlapping of objects is very commonly seen in X-ray images and needs to be carefully dealt with. In this paper, we propose a novel energy functional to solve this problem. The Euler-Lagrange equation is derived and the segmentation is converted to a front propagating problem that can be efficiently solved by level set methods. We noticed that the proposed energy functional has no unique extremum and the solution relies on the initialization. Thus, an initialization method is proposed to get satisfying results. The experiment on real data validated our proposed method.
A new technique for motion blur removal is proposed in this paper. Motion blur occurs when there is relative motion between the camera and the object of interest. To remove the blur, it is first necessary to identify the point spread function (PSF). We study the characteristics of restoration errors due to some elaborately selected PSFs. The direction and extent of the PSF can be determined after some intentional restorations. Wiener filter is then incorporated to restore the blurred image. The algorithm is effective and robust to noise. We apply it to the restoration of motion blurred license plates images. The results are fairly satisfactory.