Neural Networks

• Decision-based neural networks for object recognition

• Images decomposed into high and low frequency components.

• Image hierarchy:

  • low frequency components reflects overall structure
  • high frequency components delineate details

Previous slide

Next slide

Back to first slide

View graphic version

Using color, edge and texture information help to capture the multi-resolution characteristics of the image.

Image Classification--2 phases

1-color is used to classify each object into one of the defined-color classes (neural network algorithm)

2-texture features are then applied to refine the classification (decision-based neural networks)