UDC 681.7.014.3:
U.Khamdamov, M.Mukhiddinov, O.Djuraev, A.Abdusalomov
IMAGE SEGMENTATION BASED ON GLOBAL CONTRAST FOR SALIENT OBJECT
EXTRACTION
The technology of image segmentation is widely used in pattern recognition, object extraction, etc. It divides a
digital image into multiple regions in order to analyze and distinguish different objects in the image. In this paper, we
propose a novel method for image segmentation based on global contrast in order to extract multiple objects in the
complex natural image. To accomplish this, we apply an image enhancement technique to natural scene images, and a
segmented image is obtained to measure the color contrast of homogeneous regions against all other regions in the
image. The proposed method intends to incorporate both color and neighborhood information. Experimental results
show that the proposed method effectively segment foreground objects and achieves exceptional performance by
comparing with existing methods.
Keywords: image segmentation, object extraction, global color contrast, histogram equalization.
Introduction
Image segmentation is one of the essential yet most
difficult tasks in computer vision. An efficient image
segmentation is one of the most critical tasks in
automatic image processing. Analyzing an image and
extracting useful information from the image to
accomplish some works is an important area of
application in digital image processing and the first step
in analyzing the image is the image segmentation. An
attribute of a pixel in an image and information of pixels
near to that pixel are two fundamental parameters for any
image segmentation techniques. Since the images are
divided into two kinds on the base of their color, i.e. gray
scale and true color images [8]. Therefore, image
segmentation for color images is totally different from
gray scale images [1]. Most gray scale image
segmentation methods such as region growing, fuzzy
methods, clustering, neural networks, edge detection, and
histogram thresholding can be extended to true color
images. Gray level segmentation methods can be applied
directly to each component of a color space, and then the
results can be combined in some way to obtain a final
segmentation result [2]. Furthermore, which technique is
robust and works effectively is depends on the type of
image. It can also describe as similarity of pixels in any
region and discontinuity of edges in image. Edge
detection based image segmentation methods note to the
use of different regions of the pixel gray or color
discontinuity detection area of the edge in order to
accomplish image segmentation. The result taken from
image segmentation process is the main parameter for
further image processing research; this result will also
determine the quality of further image processing process
[1]. Although great progress has been made in the
research area of visual attention, it still remains one of
the most important and challenging issues in the fields of
image analysis, pattern recognition and computer vision.
The proposed method can be divided into two main
stages including 1) pre-processing step, 2) Image
segmentation using global contrast of the pixels for
generation saliency map. The rest of this paper is
organized as follows: Section 2 surveys conventional
image segmentation methods. Then, we illustrate our
proposed method’s framework overview and all steps of
our proposed approach together with a complete
algorithm summary in Section 3. Finally, the conclusion
is given in Section 4.
Do'stlaringiz bilan baham: |