### why the scale factor is fixed in BinaryDescriptor

I find that the scale factor in `BinaryDescriptor::detect()`

is fixed to sqrt(2), which does not depend on the input parameter `reductionRatio`

. But in `BinaryDescriptor::compute()`

, the scale factor becomes not fixed, which equals to ~~BinaryDescriptor~~reductionRatio

. My question is why they are different?