GR-signature
In Computer Vision applications, objects can be discriminated based on their
shapes. This paper presents a method to discriminate objects based on their
shapes. The approach in this paper exploits Radon Transform and Gradient
to simplify shape descriptors or signatures from an image. The simplied signatures
are called GR-signatures. GR-signatures are used together with the
proposed metric to estimate percentage of rectangularity of a given object.
The percentage of rectangularity helps us to discriminate objects with low
eort. The experiment has shown that GR-signature gives accurate results
to measure rectangularity in comparison to approach proposed by Rosin [5].
The approach has low complexity and provides less sensitiveness to protrusions
and noise.
GR_signature (pdf, 487 KB)
shapes. This paper presents a method to discriminate objects based on their
shapes. The approach in this paper exploits Radon Transform and Gradient
to simplify shape descriptors or signatures from an image. The simplied signatures
are called GR-signatures. GR-signatures are used together with the
proposed metric to estimate percentage of rectangularity of a given object.
The percentage of rectangularity helps us to discriminate objects with low
eort. The experiment has shown that GR-signature gives accurate results
to measure rectangularity in comparison to approach proposed by Rosin [5].
The approach has low complexity and provides less sensitiveness to protrusions
and noise.
GR_signature (pdf, 487 KB)
sitif110 - 11. Feb, 10:19