Affiliations: Machine Intelligence Unit, Indian Statistical
Institute B.T. Road Kolkata 700035, Kolkata, India. E-mail: sriparna [email protected];
[email protected]
Note: [] Address for correspondence: Machine Intelligence Unit, Indian
Statistical Institute, B.T. Road Kolkata 700035, Kolkata, India
Abstract: The problem of classifying an image into different homogeneous
regions is viewed as the task of clustering the pixels in the intensity space.
In particular, satellite images contain landcover types some of which cover
significantly large areas, while some (e.g., bridges and roads) occupy
relatively much smaller regions. Automatically detecting regions or clusters of
such widely varying sizes presents a challenging task. In this paper, a newly
developed real-coded variable string length genetic fuzzy clustering technique
with a new point symmetry distance is used for this purpose. The proposed
algorithm is capable of automatically determining the number of segments
present in an image. Here assignment of pixels to different clusters is done
based on the point symmetry based distance rather than the Euclidean distance.
The cluster centers are encoded in the chromosomes, and a newly developed fuzzy
point symmetry distance based cluster validity index, FSym-index, is used as a
measure of the validity of the corresponding partition. This validity index is
able to correctly indicate presence of clusters of different sizes and shapes
as long as they are internally symmetrical. The space and time complexities of
the proposed algorithm are also derived. The effectiveness of the proposed
technique is first demonstrated in identifying two small objects from a large
background from an artificially generated image and then in identifying
different landcover regions in remote sensing imagery. Results are compared
with those obtained using the well known fuzzy C-means algorithm both
qualitatively and quantitatively.
