14-02-2013, 12:44 PM
Image Segmentation Based on Active Contours without Edges
Image Segmentation Based on Active.pdf (Size: 376.96 KB / Downloads: 40)
Abstract—
There are a lot of image segmentation techniques that
try to differentiate between background and object pixels, but
many of them fail to discriminate between different objects that
are close to each other. Some image characteristics like low
contrast between background and foreground or inhomogeneity
within the objects increase the difficulty of correctly segmenting
images. We designed a new segmentation algorithm based on
active contours without edges. We also used other image
processing techniques such as nonlinear anisotropic diffusion and
adaptive thresholding in order to overcome the images’ problems
stated above. Our algorithm was tested on very noisy images, and
the results were compared to those obtained with known
methods, like segmentation using active contours without edges
and graph cuts. The new technique led to very good results, but
the time complexity was a drawback. However, this drawback
was significantly reduced with the use of graphical programming.
Our segmentation method has been successfully integrated in a
software application whose aim is to segment the bones from CT
datasets, extract the femur and produce personalized prostheses
in hip arthroplasty.
Active contours without edges, image segmentation, nonlinear
anisotropic diffusion, parallel image processing
INTRODUCTION
Image processing is a popular technique in many domains,
like medicine, security and surveillance, traffic control and
image editing. The human eye can easily distinguish important
characteristics even in very poor quality images. The aim of
image-processing software-applications is to interpret images
in a similar (or even better) way as compared to a human
being, but only faster. The results obtained by computers can
be very impressive, for example, the 3D visualization of a CT
dataset, with the highlighting of some important characteristics,
like different human tissues. Often these automatic processing
algorithms may lead to errors. The challenge in this domain is
to discover fast and fully automatic image processing
algorithms, or algorithms that require only little human
intervention.
The research presented in this paper concentrates on the
segmentation of poor quality images, like CT images, that have
a granular aspect, small contrast between foreground and
background and inhomogeneities regarding the intensity within
the foreground. The second chapter discusses the state of the
art in image segmentation. The third chapter describes a series
of image processing techniques that were used in our algorithm
and in the segmentation methods that were compared to our
algorithm. Next, we state our motivation for designing a new
segmentation technique. The fifth chapter presents the steps of
our new algorithm. In the sixth chapter we focus on
implementation details both on the CPU and on the GPU with
CUDA. In the last chapter we present a comparison between
the results obtained with our algorithm and the results
produced by other segmentation algorithms.
RELATED WORK
There are a lot of image segmentation techniques, some
based on intensity or texture, others on gradient or shape
characteristics. Some of the methods that have proven to lead
to good results in the segmentation of poor quality images are
briefly presented in this section.
Kass et al. [1] introduce the concept of snakes, or active
contours. Snakes are energy-minimizing splines guided by
external constraint forces and influenced by image forces that
pull them toward features such as lines and edges. Chan and
Vese [2] propose active contours without edges. It is a new
model for active contours, which is based on techniques of
curve evolution, the Mumford-Shah functional for
segmentation, and level sets. This method will be detailed in
the next chapter.
Boykov and Veksler [3] describe the use of graph cuts in
computer vision and graphics through theories and
applications. In image segmentation, a graph is created from
the image or the set of images. The graph construction and the
characteristics that divide the pixels into two disjoint parts, i.e.,
the background and the foreground, will be detailed in the next
section.
In grey scale mathematical morphology, the watershed
transform, originally proposed by Digabel and Lantuejoul [4]
and later improved by Buecher and Lantuejoul [5], is
considered to lead to very good results in image segmentation.
Roerdink and Meijster [6] wrote a review of several definitions
and algorithms of the watershed transform. They describe in
this review the geographic idea behind the watershed transform
as that of a landscape or topographic relief which is flooded by
water. Watersheds are the dividing lines of the domains of
attraction of rain falling over the region. When the water level
has reached the highest peak in the landscape, the process is
stopped, and the result is a landscape partitioned into basins
separated by dams, called watershed lines.
Porwik and Lisowska [7] present the use of the Haarwavelet
transform in digital image processing. Their paper
describes a method of image analysis by means of the wavelet-
978-1-4673-2952-1/12/$31.00 ©2012 IEEE 213
Haar spectrum. Glavasova et al. [8] discuss the wavelet
transform for feature extraction, based on texture analysis, for
the final goal of image segmentation.
There are a lot of other segmentation algorithms, but most
of them are based to some extent on one of the techniques
mentioned above. The challenges in image segmentation come
from the problems previously stated. The small contrast
between foreground and background makes it difficult to
extract all the edges or lines that are used for example in active
contours, graph cuts and watersheds. The inhomogeneities
within the objects prove to be a drawback for active contours
without edges, because this method tries to minimize the
differences within the foreground and the background. The
lack of a texture pattern could be a problem for the wavelet
transform based segmentation. Our algorithm takes into
account the imperfections of poor quality images (like CTs),
not only differentiating between objects and background, but
also between different objects.
IMAGE PROCESSING TECHNIQUES
For the understanding and motivation of our algorithm, we
shortly describe in this section the existing techniques that we
used in our image segmentation. We also present some details
regarding the other segmentation methods that were compared
to our algorithm.