PCA-Based Detection of Occluded Object

Problem: One of the drawbacks associated with using feature extraction approaches based on Principle Component Analysis (PCA) is that they are very sensitive to occlusion.

Solution Approach:

• Explore the feasibility of applying PCA to a quadtree representation of correlated images in order to accommodate the presence of occlusion.

• Decisions concerning an object are based on only those portions of the image in which the algorithm has determined that the object is not occluded.

• The algorithm will improve the ability to detect objects in the presence of occlusion.

Previous slide

Next slide

Back to first slide

View graphic version