Spectrum decomposition in Gaussian scale space for uneven illumination image binarization.

Jianwu Long,Hongfa Chen,Xinlei Song,Zeran Yan,Saeed Mian Qaisar

doi:10.1371/journal.pone.0251014

Abstract

Although most images in industrial applications have fewer targets and simple image backgrounds, binarization is still a challenging task, and the corresponding results are usually unsatisfactory because of uneven illumination interference. In order to efficiently threshold images with nonuniform illumination, this paper proposes an efficient global binarization algorithm that estimates the inhomogeneous background surface of the original image constructed from the first k leading principal components in the Gaussian scale space (GSS). Then, we use the difference operator to extract the distinct foreground of the original image in which the interference of uneven illumination is effectively eliminated. Finally, the image can be effortlessly binarized by an existing global thresholding algorithm such as the Otsu method. In order to qualitatively and quantitatively verify the segmentation performance of the presented scheme, experiments were performed on a dataset collected from a nonuniform illumination environment. Compared with classical binarization methods, in some metrics, the experimental results demonstrate the effectiveness of the introduced algorithm in providing promising binarization outcomes and low computational costs.

Highlights

Image segmentation is a fundamental but challenging research subject in computer vision and has been widely applied in several fields such as military, agriculture, medicine, etc
The reason is that an object illuminated by strong light tends to cause an uneven distribution of image pixels, which deviates from the original pixel value of the object; and the binarization algorithm is difficult to implement on such images
The method introduced in this paper, which combines the Gaussian scale space (GSS) with spectrum decomposition to estimate background, is mainly used for the background estimation of unevenly illuminated images

Summary

Introduction

Image segmentation is a fundamental but challenging research subject in computer vision and has been widely applied in several fields such as military, agriculture, medicine, etc. The text image binarization algorithm based on the precipitation model was proposed by Oh et al [8]. This method needs to extract edge pixels as seed points, and it has poor processing of weak edges and low contrast images. The method based on fuzzy sets that introduced different membership functions to divide pixel categories was proposed by Aja-Fernandez et al [11] They believed that the use of the multiregion thresholding method can reduce the possibility of pixels being misclassified and overcome the situation of images being affected by artifacts and noise.

Materials and methods

XWÀ 1 XHÀ 1

Experiments and analysis

Method Water flow

Conclusion