| 类 | 说明 |
|---|---|
| BitSampling |
Provides a simple way to hashing.
|
| LocalitySensitiveHashing |
Each feature vector v with dimension d gets k hashes from a hash bundle h(v) = (h^1(v), h^2(v), ..., h^k(v)) with
h^i(v) = (a^i*v + b^i)/w (rounded down), with a^i from R^d and b^i in [0,w)
If m of the k hashes match, then we assume that the feature vectors belong to similar images. |
| MetricSpaces |
This hashing approach implements the proximity approach "metric spaces" based on the work of Giuseppe Amato.
|
| MetricSpaces.Parameters |
For storing the parameters per feature.
|
| MetricSpaces.Result |