1 Introduction

1.1 Detection

Find a set of distinctive key points.

1. Given: two images of the same scene with a large scale difference between them
2. Goal: find the same interest points independently in each image
3. Solution: search for maxima of suitable functions in scale and in space (over the image)

1.1.1 Advantages of invariant local features

Locality: features are local, so robust to occlusion and clutter (no prior segmentation)

Distinctiveness: individual features can be matched to a large database of objects

Quantity: many features can be generated for even small objects

Efficiency: close to real-time performance

Extensibility: can easily be extended to wide range of differing feature types, with each adding robustness

1.1.2 Scale Invariant Detection: Summary

• Difference-of-Gaussian (DoG) detector or SIFT [Lowe]: maximize Difference of Gaussians over scale and space

• Harris -Laplacian [Mikolajczyk, Schmid]: maximize Laplacian over scale, Harris ’ measure of corner response over the image

• Maximally Stable Extremal Regions (MSER)

1.2 Local features: main components