I've heard this from a few different researchers but I can't find a definitive value. Given the "scale" extracted by David Lowe's SIFT detector, we need to extract a descriptor at a specific scale.
Is it 6 times the detector "scale"? 3? 1? I've heard conflicting values. What is known about this?
If I understand correctly what you are asking -- In general, the feature is found at the same scale as the SIFT detector says, but in David Lowe's SIFT, the image is pre-smoothed with sigma: 0.5, so, you need to "subtract" this amount of smoothing from the sigma, so the "real" scale could be:
sqrt(sigma^2 - 0.5^2)
where sigma is the scale the feature was found at. you have an excellent implementation of SIFT detector and descriptor at: www.vlfeat.org
