Related Experiment Video
Updated: Dec 29, 2025
04:48
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
705
Hybrid image segmentation method based on anisotropic Gaussian kernels and adjacent graph region merging.
Zhuo Zhao1, Bing Li1, Xiaoqin Kang1
1State Key Laboratory for Manufacturing System Engineering, Xi'an Jiaotong University, No. 99 Yanxiang Road, Yanta District, Xi'an 710054, Shaanxi, China.
The Review of Scientific Instruments
|February 5, 2020
Summary
This study introduces a hybrid image segmentation method using the Anisotropic Gaussian Kernel (ANGK) and region adjacent graph (RAG) for accurate object identification. The novel approach enhances localization accuracy and noise robustness, effectively addressing oversegmentation issues.
Area of Science:
- Computer Vision
- Image Analysis
- Pattern Recognition
Background:
- Image segmentation is crucial for object identification in image analysis.
- Existing methods often struggle with accuracy and oversegmentation.
Purpose of the Study:
- To propose a hybrid image segmentation method combining edge detection and region merging.
- To improve localization accuracy and noise robustness in image segmentation.
Main Methods:
- Utilized an Anisotropic Gaussian Kernel (ANGK) for edge detection.
- Employed watershed transform for initial coarse segmentation.
- Applied region adjacent graph (RAG) merging with similarity and shape cost functions for fine segmentation.
Main Results:
- Achieved preferable localization accuracy and noise robustness.
- Demonstrated superior segmentation effect compared to conventional methods.
- Effectively solved the problem of oversegmentation.
Conclusions:
- The proposed hybrid method offers a significant advancement in image segmentation.
- It provides a robust and accurate solution for object identification tasks.
- The method's ability to mitigate oversegmentation is a key advantage.
