A Two-Stage Evolutionary Fuzzy Clustering Framework for Noisy Image Segmentation

Licheng Jiao,Lingling Li,Wenping Ma,Mengxuan Zhang,Fang Liu

doi:10.1109/access.2020.3029773

Licheng Jiao, Lingling Li + Show 3 more

Open Access

https://doi.org/10.1109/access.2020.3029773

Copy DOI

Abstract

This article presents a two-stage evolutionary fuzzy clustering framework for noisy image segmentation. It is a bi-stage system comprising a multi-objective optimization stage and a fuzzy clustering segmentation stage. In the multi-objective optimization stage, the fuzzy clustering on pixels in inhomogeneous regions is converted into a multi-objective problem, which can preserve image details while restraining noise. The multi-objective problem is decomposed into several sub-problems by the Tchebycheff approach. A trade-off can be obtained by optimizing these sub-problems simultaneously. In the fuzzy clustering segmentation stage, fuzzy clustering with the trade-off between preserving image details and restraining noise is performed on the whole observed image. To deal with this fuzzy clustering problem, an adaptive evolutionary fuzzy clustering algorithm with spatial information is proposed. Experiment results on synthetic and real images illustrate the effectiveness of the proposed framework for noisy image segmentation.

Highlights

Image segmentation is the most critical step in many image applications
We proposed two evolutionary fuzzy clustering algorithms, MOEFC [41] and EAFC [42], for image segmentation
To evaluate the performances of the algorithms on natural images selected from Berkeley segmentation dataset (BSD) [47] and Visual Object Classes Challenge 2012 (VOC2012) quantitatively, the probabilistic rand index (PR) [48], the variation of information (VI) [49] and the pixel intersection-over-union (IOU) across classes are utilized

Summary

INTRODUCTION

Image segmentation is the most critical step in many image applications. It aims to divide an image into several partitions with similar characteristics. The trade-off between preserving image details and restraining noise for pixels in inhomogeneous regions can be obtained by optimizing these fuzzy clustering sub-problems simultaneously. INHOMOGENEITY MEASURE BASED SAMPLING In the multi-objective optimization stage of TEFC, it concerns the trade-off between preserving image details and restraining noise for the pixels in inhomogeneous regions. The multi-objective optimization stage of TEFC searches a trade-off between preserving image details and removing noise for the pixels in inhomogeneous regions. Zc)T represent the number of clusters and a set of cluster centers, the multi-objective fuzzy clustering problem for the sampled pixels Y can be defined as follows: min F(z) = [f1(z), f2(z)]T (3). With each weight vector λi = (λi, 1 − λi)T , different balances of preserving image details and removing noise can be achieved by optimizing the fuzzy clustering sub-problems. The neighbor individuals of the trade-off are output to the fuzzy clustering segmentation stage

FUZZY CLUSTERING SEGMENTATION ON WHOLE OBSERVED IMAGE

CONCLUSION