원문정보
초록
영어
The paper presents a method which bases on support vector machine(SVM) and particle filtering for detecting and tracking player in the broadcast sports video. Firstly, through the combination of Support Vector Classification and CourtSegmentationmethod, it proposes the algorithm for examining automatically members in those videos, which is used to initialize the trace of subsequent visual objects; secondly, by combing support vector regression frame and the one of sequential Monte Carlo, it brings forth the improved particle filtering algorithm which is applied to follow visual objects, enabling the traditional particle filtering method to achieve robust trail of such targets even when the particle set is small, together with effective enhancement of the running efficiency of tracking system.
목차
1. Introduction
2. Player Detection based on Court Segmentation
3. Player Track Based on SVR and Particle Filtering
3.1. Introduction of Particle Filtering
3.2. Problems with Traditional Particle Filtering Method
3.3 Improved Particle Filtering Algorithm based on Support Vector Regression
4. Moving Object Detection and Tracking Framework Based on SVM and Particle Filtering
5. Experiment Design and Results
5.1 Testing Results of Player Inspection
5.2. Comparison Results Support Vector Regression Particle Filter with the Traditional Particle Filter
5.3. Experimental Results of Tracking Player
6. Conclusion
References