Summer Paper Reading

1. Visual Tracking

1.1. On-line Random Forrest

1.1.1. IDEA: grow multiple randomized decision trees on-line for classification

1.1.2. KEY

1.1.2.1. On-line Bagging

1.1.2.2. On-line Growing Trees

1.1.2.2.1. choose the best question(test)

1.1.2.2.2. split the leaf only when needed and able to gain best benefit

1.1.2.3. On-line Adaptation

1.1.2.3.1. temporal knowledge weighting

1.1.2.3.2. ability to unlearn useless knowledge

1.1.2.3.3. ability to handle changing data distribution

1.1.3. APPLICATION

1.1.3.1. tracking object in real-time, able to handle

1.1.3.1.1. occlusion

1.1.3.1.2. fast movement

1.1.3.1.3. view point change

1.2. Online Multi-Class LPBoost

1.3. P-N Learning: Bootstrapping Binary Classifiers by Structural Constraints

1.4. New node

2. Structure and Motion Estimation for AR

2.1. Model Based

2.2. Match Moving Based

2.2.1. Line Feature Based

2.2.1.1. Moving in Stereo: Efficient Structure and Motion Using Lines

2.2.2. Point Feature Based

2.3. Homography Matching Based