International Journal of Advances in Electrical Engineering
2023, Vol. 4, Issue 2, Part A
Improved image segmentation technique for foreground extraction
Author(s): Devendra Tanaji Rane, Dr. Prashant Kumbharkar and Dr. Archana T Bhise
Abstract: With the new era of Computer Technology, Computer vision has made tremendous progress. It has made a lot of improvements in Digital Image Processing. Digital Image processing techniques are typically categorized into three categories include Image Generation, Image Enhancement, and Image Restoration. Among multiple phases of Image processing, Segmentation Procedure is the area in which Image gets partitioned into its constituent parts or objects [1]. Image Segmentation is a method in which a digital image is broken down into various subgroups/regions called Image Segments which helps in reducing the complexity of the image to make further processing or analysis of the image simpler [3]. Many Image segmentation techniques are available based on two basic approaches Similarity/Region Approach and Discontinuity/Boundary Approach. Image Segmentation techniques are widely used in document processing, object recognition, remote sensing image, biomedicine, and many other aspects like factory production automation, Computed Tomography (CT) Images etc. Among existing foreground extraction techniques, the graph-based method Grabcut can effectively extract the foreground according to edges and appearance models [6]. Gaussian Mixture Model (GMM) is used for modelling the foreground and the background. Though this method is easy to use it has multiple manual interactions and iterations to achieve final segmentation. Accuracy and performance are the main problem areas with current GrabCut method of foreground extraction in image segmentation. Objective here is to develop Image pre-processing and initializing stage using a combination of binarization of mask and then pass it to GrabCut to establish a model of Image Background and Foreground to avoid manual interaction and to achieve more accuracy in segmentation.
DOI: 10.22271/27084574.2023.v4.i2a.41
Pages: 12-29 | Views: 394 | Downloads: 164
Download Full Article: Click Here
How to cite this article:
Devendra Tanaji Rane, Dr. Prashant Kumbharkar, Dr. Archana T Bhise. Improved image segmentation technique for foreground extraction. Int J Adv Electr Eng 2023;4(2):12-29. DOI: 10.22271/27084574.2023.v4.i2a.41
