Fengxia Li

Research of dynamic terrain in complex battlefield environments

In this paper, we present a novel method for dynamic terrain in battlefield and an efficient plan to simulate crater in the battle. We explore a few methods for dynamic terrain surface, and analyze the different recent algorithms for representation and visualization of multiresolution static terrain, then select ROAM (Real-Time Optimally Adapting Meshes) as our basis to be improved and to be applicable in dynamic terrain projects. We further select half-ellipsoid model in the explosion theory as the physical based model of crater, and provide a procedural texture generation algorithm for crater, then give a hybrid multiresolution algorithm for dynamic terrain. At last the implementation results prove that the method and the algorithm are feasible and efficient.

Dynamic Terrain Visualization Method using Strip Masks and Its Applications

In this paper, we give a view-dependent dynamic terrain visualization method using strip masks, and we also present our implementations results. We partition our terrain into several blocks, and we pre-compute a set of strip masks of blocks. At run time, to render the terrain, we simply choose the mask of blocks that represent each part of the terrain to meet our desired visual fidelity. We also combine our algorithm with the physical based model of craters and ruts. Our implementations are running smooth running smoothly in real time, and our craters and ruts are very realistic. The experiments prove our method is feasible and valid

Simultaneous pose motion recovery and video object cutout

In this paper, we present a novel system for simultaneously performing segmentation and 2D pose motion recovery for the articulated object in a video sequence. The system first preprocesses pixels into superpixels to reduce the number of nodes which largely affects the computational complexity of later optimizations. By starting from true pose estimation obtained with user assistants on each key frame, a parallel pose tracking procedure, whose energy function considers boundary, appearance and pose prior information as well, is conducted forward and backward on in-between frames. With different searching strategies, multiple pose candidates are inferred to help recover missed true poses. Finally, by solving the cost function of the pose motion recovery, which exploits the temporal coherence of object movement, the pose motion and the video object are produced at the mean time. As a parameterized tree-based articulated model drawn by the user is applied to denote the pose, our method is generic and can be used for any articulated object.

Shape Manipulation on GPU

This paper proposes a novel hardware-accelerating deformation algorithm based on curve-skeleton model for 2D shape manipulation. The deformation algorithm can achieve real-time interactive shape manipulation without any pre-computing step. The deforming regions of shapes are demarcated with a simple skeleton frame and are simulated by a curve-skeleton model consisting of triangle-strips. The algorithm obtains two properties that smooth flexion and preservation of area. Smooth flexion is guaranteed due to the continuous derivative of the curve function, and preservation of area is achieved via adjusting the parameters of the control curves. GPGPU technique is adopted to reduce the workload on CPU and to accelerate the rendering. Our algorithm can be used to 2D shape manipulation, character animation and cartoon-like objects deformation. It has been proved feasible and valid by our experiments.

Method of motion data processing based on manifold learning

Due to the high-dimensionality of motion captured data which resulted in the complexity in motion analysis, a method of motion data processing based on manifold learning was proposed. Isomap, a classical manifold learning algorithm, was necessary to be improved and extended in this paper. A framework of motion data processing based on manifold learning was built to embed high-dimensionality data into low-dimensionality space. It simplified the motion analysis, and in the same time preserved the original motion features. In order to solve the inefficiency of processing large-scale motion data, Sample Isomap (S-Isomap) algorithm was proposed. Experiments proved that approximate embeddings of motion data computed by S-Isomap were average 10 times faster than by Isomap, while 10% frame samples were selected.