Documentation for Video/RegionShape

Technique Region shape descriptor
Document MPEG-7 Visual (ISO/IEC 15938-3)
Name Whoi-Yul Kim (Hanyang University)
EMail wykim@email.hanyang.ac.kr
Contact wykim@email.hanyang.ac.kr
Type Application
External Libraries none
Related Ds/DSs Still region DS
Used Ds/DSs none
Input Binary/Gray shape masks
Extraction Yes
Client Appl Search & Retrieval
Summary Shape of an object may consist of either single region or a set of regions
as well as some holes in the object in a frame.
The region-based shape descriptor describes shape features of the object
as long as it can be defined by a shape mask. The mask may be binary to contain
holes or multiple disjoint regions within. The mask may also be in gray shade
to distinguish different regions within the object. The descriptor consists
of a set of 35 quantized coefficients of Angular-Radial Transform of an object.
Strong Points The descriptor is rotational, scale and position invariant. It can deal with
objects with multiple regions including holes in the object.
It is robust to noise along the contour of the object that is common during
the segmentation process of an image, or to shape deformations
due to affine transform.
The descriptor is also characterized by its small size, fast extraction time
and matching. The data size for this representation is fixed to 17.5byte.
The feature extraction and matching processes is straightforward to have
low order of computational complexities, and is suitable for tracking shapes
in the video data processing.
Limitations None
Known Problems None
Parameters Uses the general XM-SW parameters.