Classes - Beard

You are here: start » ee490-11reports


|

Meta

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Last revision Both sides next revision
ee490-11reports [2011/04/13 14:59]
ben
ee490-11reports [2011/04/15 12:59]
ben
Line 12: Line 12:
  
 The loop closure program works by first extracting features from an image. There are several different types of features that can be extracted, but the method that we used is called SURF, which extracts features that are scale and rotation invariant. These features are often abstract or very small so don't expect to understand what features the computer is seeing. For information on how SURF works we recommend you go to [http://​www.aishack.in/​2010/​05/​sift-scale-invariant-feature-transform/​]. It explains the SIFT algorithm, which is the algorithm which SURF is based on. Our code is written in OpenCV, which already has a SURF implementation. The loop closure program works by first extracting features from an image. There are several different types of features that can be extracted, but the method that we used is called SURF, which extracts features that are scale and rotation invariant. These features are often abstract or very small so don't expect to understand what features the computer is seeing. For information on how SURF works we recommend you go to [http://​www.aishack.in/​2010/​05/​sift-scale-invariant-feature-transform/​]. It explains the SIFT algorithm, which is the algorithm which SURF is based on. Our code is written in OpenCV, which already has a SURF implementation.
 +
 {{:​campus_challenge:​matching.png|}} {{:​campus_challenge:​matching.png|}}
 +
 The SURF function outputs two vectors. The first vector is called '​imagedescriptors'​. It contains a 64 bit description of each feature found in the image(making it 64 by N). The other output is a vector called imagekeypoints. This vector stores the location of each feature found on the image. To make it so the computer does not have to process every single feature (There will be potentially hundreds of thousands) The features in the images are grouped into clusters using k-means. The SURF function outputs two vectors. The first vector is called '​imagedescriptors'​. It contains a 64 bit description of each feature found in the image(making it 64 by N). The other output is a vector called imagekeypoints. This vector stores the location of each feature found on the image. To make it so the computer does not have to process every single feature (There will be potentially hundreds of thousands) The features in the images are grouped into clusters using k-means.