No abstract

We study the problem of detecting talking activities in collaborative learning videos. Our approach uses head detection and projections of the log-magnitude of optical flow vectors to reduce the problem to a simple classification of small projection images without the need for training complex, 3-D activity classification systems. The small projection images are then easily classified using a simple majority vote of standard classifiers. For talking detection, our proposed approach is shown to significantly outperform single activity systems. We have an overall accuracy of 59% compared to 42% for Temporal Segment Network (TSN) and 45% for Convolutional 3D (C3D). In addition, our method is able to detect multiple talking instances from multiple speakers, while also detecting the speakers themselves.

Dental computed tomography (CT) images and optical surface scan data are widely used in dental computer-aided design systems. Registration is essential if they are used in software systems. Existing automatic registration methods are either time-consuming or rough, and interactive registration methods are experience-dependent and tedious because of a great deal of purely manual interactions. For overcoming these disadvantages, a two-stage registration method is proposed. In the rough registration stage, the rough translation and rotation matrices are obtained by applying unit quaternion based method on the points interactively selected from the two types of data. In the precise registration stage, the stridden sampling is used to reduce computational complexity and the proposed registration algorithm with scale transformation is used for precise registration. The proposed method offers a good trade-off between precision and time cost. The experimental results demonstrate that the proposed method provides faster convergence and smaller registration errors than existing methods.

No abstract

This study aimed to investigate the therapeutic effect and nursing evaluation of patients with cerebral stroke (CS) through intracranial magnetic resonance imaging (MRI) images under the condition of segmentation algorithm. 199 CS patients were selected and divided randomly into a control group (group A) and an experimental group (group B) based on different treatment methods. Patients of group A were given routine antithrombotic therapy, and patients of group B were treated with intravenous thrombolytic therapy under evaluation of segmentation algorithm-based MRI images. Then, there were comparisons on clinical therapeutic effect, neurological damage score, and daily living ability index score of all patients. After treatment, the total effective rate (92.12%) and daily life index (41.45 ± 11.24) of patients in group B were higher than those of group A ( P  < 0.05). However, neurological damage scores (3.36 ± 1.13 points) of patients in group B after treatment were lower than those of group A (5.85 ± 2.31 points) ( P  < 0.05). The routine clinical nursing satisfaction rate (79.8%) was lower than the overall satisfaction rate (97%) of the combination of clinical and imaging nursing ( P  < 0.05). Therefore, there were greatly clinical therapeutic effects of intravenous thrombolysis evaluated by intracranial MRI images under segmentation algorithm for CS patients, and routine nursing could improve patients’ satisfaction, which were worthy of clinical promotion.