원문정보
초록
영어
Moving object detection in a video sequence is one of the most challenging research topics due to its various applications and complexity. Sometimes, it is also required to identify a moving object without any previous information about its shape or motion. In this paper, we introduce a global mesh generation technique which can construct a hierarchical mesh on moving object(s) inside a video frame. This technique is efficient enough to solve the problem to a great extent and can identify a moving object in a video frame without any previous knowledge. The generated mesh is triangular and probabilistic in nature. The granularity of the mesh depends upon the level of hierarchy, neighborhood triangle decomposition, gradient and temporal information of the video sequence. Experiments show that the proposed method can identify the moving object(s) with good accuracy without any pre or post-processing steps.
목차
I. INTRODUCTION
II. PROPOSED WORK
A. Short description of the defined features
B. Combined expression for probability
C. Fine tuning of parameters α, β and γ
D. Overall description of the algorithm
III. EXPERIMENTAL RESULTS
A. Results on other different video sequences
B. Performance comparison
IV. CONCLUSIONS
REFERENCES