Scale Space Based Grammar for Hand Detection

* Final gross prices may vary according to local VAT.

Get Access

Abstract

For detecting difficult objects, like hands, we present an algorithm that uses tokens and a grammar. Tokens are found by employing a new scale space edge detector that finds scale invariant features at object boundaries. We begin by constructing the scale space. Then we find edges at each scale and flatten the scale space to one edge image. To detect a hand we define a hand pattern grammar using curve tokens for finger tips and wedges, and line tokens. We identify a hand pattern by parsing these tokens using a graph based algorithm. We show and discuss the results of this algorithm on a database of hand images.