Comparison of stochastic filtering methods for 3D tracking
Pattern Recognition
Hi-index | 0.00 |
Shape matching is one of the most popular methods for recognition and tracking of moving objects in video sequence. When the method is used for recognizing the moving & rotated object, it is limited for practical use due to consuming a lot of calculation time. Aiming at the problem, a real-time tracking method of moving objects based on Kalman filtering and Gabor Decomposition is proposed. First of all, Kalman filter was used to predict possible location of the vehicle in the next frame, and then Gabor wavelet features were used to match points in the predicted region, to accurate location of vehicles. In order to enhance the tracking speed, all of the extracted feature points should be screened in the experiment. Some typical characteristics of selected points were matched with the standard database models. The experimental results show that this method has good tracking results, and vehicles blocked in a short period of time can be tracked effectively too.