Search

RGBD-based real-time dynamic 3D reconstruction suffers from inaccurate inter-frame motion estimation as errors may accumulate with online tracking. This problem is even more severe for single-view-based systems due to strong occlusions. Based on these observations, we propose OcclusionFusion, a novel method to calculate occlusion-aware 3D motion to guide the reconstruction. In our technique, the motion of visible regions is first estimated and combined with temporal information to infer the motion of the occluded regions through an LSTM-involved graph neural network. Furthermore, our method computes the confidence of the estimated motion by modeling the network output with a probabilistic model, which alleviates untrustworthy motions and enables robust tracking. Experimental results on public datasets and our own recorded data show that our technique outperforms existing single-view-based real-time methods by a large margin. With the reduction of the motion errors, the proposed technique can handle long and challenging motion sequences. Please check out the project page for sequence results: https://wenbin-lin.github.io/OcclusionFusion., Comment: Accepted by CVPR 2022. Project page: https://wenbin-lin.github.io/OcclusionFusion

The polygonal approximation problem is a primary problem in computer graphics, pattern recognition, CAD/CAM, etc. InR 2, the cone intersection method (CIM) is one of the most efficient algorithms for approximating polygonal curves. With CIM Eu and Toussaint, by imposing an additional constraint and changing the given error criteria, resolve the three-dimensional weighted minimum number polygonal approximation problem with the parallel-strip error criterion (PS-WMN) underL 2 norm. In this paper, without any additional constraint and change of the error criteria, a CIM solution to the same problem with the line segment error criterion (LS-WMN) is presented, which is more frequently encountered than the PS-WMN is. Its time complexity isO(n 3), and the space complexity isO(n 2). An approximation algorithm is also presented, which takesO(n 2) time andO(n) space. Results of some examples are given to illustrate the efficiency of these algorithms.

Practical Intrusion Detection Systems (IDSs) based on data mining are facing two key problems, discovering intrusion knowledge from real-time network data, and automatically updating them when new intrusions appear. Most data mining algorithms work on labeled data. In order to set up basic data set for mining, huge volumes of network data need to be collected and labeled manually. In fact, it is rather difficult and impractical to label intrusions, which has been a big restrict for current IDSs and has led to limited ability of identifying all kinds of intrusion types. An improved unsupervised clustering-based intrusion model working on unlabeled training data is introduced. In this model, center of a cluster is defined and used as substitution of this cluster. Then all cluster centers are adopted to detect intrusions. Testing on data sets of KDDCUP’99, experimental results demonstrate that our method has good performance in detection rate. Furthermore, the incremental-learning method is adopted to detect those unknown-type intrusions and it decreases false positive rate.

The highlight line model is a powerful tool in assessing the quality of a surface. It increases the flexibility of an interactive design environment. In this paper, a method to generate a highlight line model on an arbitrary triangular mesh is presented. Based on the highlight line model, a technique to remove local shape irregularities of a triangular mesh is then presented. The shape modification is done by solving a minimization problem and performing an iterative procedure. The new technique improves not only the shape quality of the mesh surface, but also the shape of the highlight line model. It provides an intuitive and yet suitable method for locally optimizing the shape of a triangular mesh.