Paper Info
Title
An Image Segmentation Method Based on Luminance Distribution and Its Application to Image Enhancement.
Abstract
This paper proposes a novel image segmentation method based on luminance distribution and its application to image enhancement. Many existing image segmentation methods focus on semantic segmentation which separates an image into some meaningful areas. However, those segmentation methods are not effective for image enhancement. The proposed segmentation method separates an image into some areas according to luminance values of pixels. To obtain those areas, the proposed method utilizes a clustering algorithm based on a Gaussian mixture model which is fit by using a variational Bayesian approach. By using the proposed segmentation method, an automatic exposure compensation method is also proposed. The proposed exposure compensation method enables to automatically produce pseudo multi-exposure images from a single image and to improve the image quality by fusing them. Experimental results show that the proposed segmentation method is effective for image enhancement. In addition, the image enhancement method using the proposed segmentation method outperforms state-of-the-art contrast enhancement methods, in terms of the entropy and statistical naturalness.
Year
DOI
Venue
2018
10.23919/APSIPA.2018.8659745
Asia-Pacific Signal and Information Processing Association Annual Summit and Conference
Field
DocType
ISSN
Pattern recognition,Segmentation,Computer science,Exposure compensation,Image quality,Image segmentation,Pixel,Artificial intelligence,Luminance,Cluster analysis,Mixture model
Conference
2309-9402
Citations 
PageRank 
References 
0
0.34
0
Authors
2
Name
Order
Citations
PageRank
Yuma Kinoshita11813.72
Hitoshi Kiya2616113.80