60 intellectual system basis that lets to make managerial decisions during the reorganization of scientific and educational work in the transition to the competence

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.