Weighted variational two-phase image segmentation based on Fuzzy Region Competition

2011 ◽  
Author(s):  
Vinicius R. P. Borges ◽  
Celia A. Zorzo Barcelos ◽  
Denise Guliato ◽  
Marcos Aurelio Batista
Keyword(s):  
Image Segmentation ◽  
Phase Image ◽  
Two Phase ◽  
Fuzzy Region

Two-Phase Image Segmentation with Nonlocal Mean Filter and Kernel Evolutionary Clustering in Local Learning

2014 ◽  
Vol 536-537 ◽  
pp. 172-175
Author(s):  
Bing Chen ◽  
Dong Dong Yang ◽  
Gang Lu ◽  
Hong Xiao Feng
Keyword(s):  
Image Segmentation ◽  
Learning Strategy ◽  
Phase Image ◽  
Local Learning ◽  
Two Phase ◽  
Evolutionary Clustering ◽  
Kernel Clustering ◽  
Mean Filter ◽  
Image Segmentation Algorithm

In this study, a novel two-phase image segmentation algorithm (TPIS) by using nonlocal mean filter and kernel evolutionary clustering in local learning is proposed. Currently, the difficulties for image segmentation lie in its vast pixels with overlapping characteristic and the noise in the different process of imaging. Here, we want to use nonlocal mean filter to remove different types of noise in the image, and then, two kernel clustering indices are designed in evolutionary optimization. Besides, the local learning strategy is designed using local coefficient of variation of each local pixels or image patch is employed to update the quality of the local segments. The new algorithm is used to solve different image segmentation tasks. The experimental results show that TPIS is competent for segmenting majority of the test images with high quality.

scholarly journals Fast Mumford-Shah Two-Phase Image Segmentation Using Proximal Splitting Scheme

2021 ◽  
Vol 2021 ◽  
pp. 1-13
Author(s):  
Nilima Shah ◽  
Dhanesh Patel ◽  
Pasi Fränti
Keyword(s):  
Image Segmentation ◽  
Level Set ◽  
Energy Functional ◽  
Phase Image ◽  
Splitting Scheme ◽  
Two Phase ◽  
Crucial Step ◽  
Error Metrics ◽  
Initial Segmentation

The Mumford-Shah model is extensively used in image segmentation. Its energy functional causes the content of the segments to remain homogeneous and the segment boundaries to become short. However, the problem is that optimization of the functional can be very slow. To attack this problem, we propose a reduced two-phase Mumford-Shah model to segment images having one prominent object. First, initial segmentation is obtained by the k-means clustering technique, further minimizing the Mumford-Shah functional by the Douglas-Rachford algorithm. Evaluation of segmentations with various error metrics shows that 70 percent of the segmentations keep the error values below 50%. Compared to the level set method to solve the Chan-Vese model, our algorithm is significantly faster. At the same time, it gives almost the same or better segmentation results. When compared to the recent k-means variant, it also gives much better segmentation with convex boundaries. The proposed algorithm balances well between time and quality of the segmentation. A crucial step in the design of machine vision systems is the extraction of discriminant features from the images, which is based on low-level segmentation which can be obtained by our approach.

VARIATIONAL TWO-PHASE IMAGE SEGMENTATION BASED ON REGION COMPETITION BY LEVEL SET AND FUZZY APPROACHES

2011 ◽  
Author(s):  
Marcus Aurelio Vinicus R. Pereira Borges ◽  
Marcus Aurelio Batista ◽  
Celia A Zorzo Barcelos
Keyword(s):  
Image Segmentation ◽  
Level Set ◽  
Phase Image ◽  
Two Phase

Phase Image Segmentation and Filtering Algorithm Based on Direction of the Gradient Factor

2014 ◽  
Vol 670-671 ◽  
pp. 1488-1492
Author(s):  
Chi Zhang ◽  
Zhong Wei Li ◽  
Yu Sheng Shi
Keyword(s):  
Image Segmentation ◽  
Median Filter ◽  
Phase Image ◽  
Point Matching ◽  
Performance Specifications ◽  
Noise Data ◽  
Image Edge ◽  
Phase Data ◽  
Effective Phase ◽  
Phase Shift Method

To satisfy rigid performance specifications of structure light measurement by phase shift method, the algorithm of phase image segmentation and filter based on direction of the gradient factor was introduced in this paper. In this method, phase image was divided into three parts by condition of extreme value and direction of the gradient factor, including continuous area, edge area and regional noise. Only phase data in continuous area was processed by median filter. This method can reduce the point matching computation of 3-D reconstruction, and at the same time can protect the image edge details, so as to reduce the noise data, provide accurate and effective phase data for reconstruction. Results of segmentation verify feasibility and effectiveness of presented method.

Sign in / Sign up

Export Citation Format

Share Document