14-04-2014, 02:51 PM
Fuzzy C-Means Clustering With Local Information and Kernel Metric for Image Segmentation
Fuzzy C-Means Clustering.pdf (Size: 1.63 MB / Downloads: 54)
Abstract
In this paper, we present an improved fuzzy
C-means (FCM) algorithm for image segmentation by introduc-
ing a tradeoff weighted fuzzy factor and a kernel metric. The
tradeoff weighted fuzzy factor depends on the space distance of
all neighboring pixels and their gray-level difference simultane-
ously. By using this factor, the new algorithm can accurately
estimate the damping extent of neighboring pixels. In order to
further enhance its robustness to noise and outliers, we introduce
a kernel distance measure to its objective function. The new
algorithm adaptively determines the kernel parameter by using
a fast bandwidth selection rule based on the distance variance
of all data points in the collection. Furthermore, the tradeoff
weighted fuzzy factor and the kernel distance measure are both
parameter free. Experimental results on synthetic and real images
show that the new algorithm is effective and efficient, and is
relatively independent of this type of noise.
INTRODUCTION
IMAGE segmentation is one of the key techniques in image
understanding and computer vision. The task of image
segmentation is to divide an image into a number of non-
overlapping regions, which have same characteristics such
as gray level, color, tone, texture, etc. A lot of clustering-
based methods have been proposed for image segmentation
[1]–[5]. Among the clustering methods, one of the most
popular methods for image segmentation is fuzzy clustering,
which can retain more image information than hard clustering
in some cases.
Fuzzy c-means (FCM) algorithm is one of the most widely
used fuzzy clustering algorithms in image segmentation. FCM
algorithm was first introduced by Dunn [6] and later extended
by Bezdek [7]. Although the conventional FCM algorithm
works well on most noise-free images, it fails to segment
images corrupted by noise, outliers and other imaging artifacts.
Its non-robust results are mainly because of ignoring spatial
contextual information in image and the use of non-robust
Euclidean distance.
EXPERIMENTAL STUDY
In this section, we describe the experimental results on
three synthetic images, four medical images and three nat-
ural images with different types of noises. In addition, we
test and compare the efficiency and the robustness of the
proposed method (KWFLICM) with spectral graph grouping
using N ystr om
̈ method (NNcut) [25] and its predecessors,
FLICM [1], RFLICM [12] and WFLICM (only introduce the
trade-off weighted fuzzy factor into FLICM algorithm, here
termed as WFLICM for short). The effectiveness of the trade-
off weighted fuzzy factor can be validated by comparisons
between FLICM and WFLICM. And the effectiveness of
the kernel metric can be validated by comparisons between
WFLICM and KWFLICM.