We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Threshold Selection with Relative J-Divergence for Image Segmentation.
- Authors
Fangyan Nie; Pingfeng Zhang
- Abstract
As an image segmentation technique with high real-time performance, thresholding has been widely used in machine vision-based task processing. However, due to the complexity of application scenarios, different methods need to be designed for different task requirements. As an effective tool to measure the information distance between different information systems, relative J-divergence makes up for the deficiency of traditional information divergence. In this paper, a new image thresholding segmentation method is designed and implemented based on relative J-divergence. In the process of image segmentation, the optimal threshold is found based on the minimum relative J-divergence criterion. The optimal threshold is applied to separate the image pixels into two parts, namely target and background, so as to achieve the purpose of image segmentation. In the comparison with some classical algorithms, the proposed method is applied to the segmentation of nondestructive testing and other images. The experimental results verify the effectiveness, application prospects and popularize value of the proposed method.
- Subjects
THRESHOLDING algorithms; IMAGE segmentation; NONDESTRUCTIVE testing; INFORMATION measurement; INFORMATION storage &; retrieval systems
- Publication
IAENG International Journal of Applied Mathematics, 2023, Vol 53, Issue 3, p899
- ISSN
1992-9978
- Publication type
Article