An Edge Weighted RS image Quality Evaluation Method

The approaches for remote sensing (RS) image quality evaluation can be categorized by two different criteria. One is the subjective image quality testing based on many observers. The other is the objective image quality testing based on mathematical calculation. Compared with the subjective method,...

Full description

Saved in:
Bibliographic Details
Main Authors: Qin Lu, Liebo Du, Xuemin Xiao
Format: Conference Proceeding
Language:English
Subjects:
Computer science
EWPSNR
Humans
Image coding
Image quality
image quality evaluation
Image reconstruction
Pixel
PSNR
Remote sensing
Software engineering
Testing
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The approaches for remote sensing (RS) image quality evaluation can be categorized by two different criteria. One is the subjective image quality testing based on many observers. The other is the objective image quality testing based on mathematical calculation. Compared with the subjective method, objective image quality testing is easier for implementation. However, the traditional objective image quality evaluation models, such as MSE and PSNR, do not consider the human vision characteristics and hence sometimes have very low consistency with the subjective image quality testing results. To solve this problem, an objective evaluation model with a new evaluation index, named edge weighted peak signal noise ratio (EWPSNR), is proposed in this paper. Taking into account the human vision characteristics, the proposed objective evaluation model pays more attention to the edge distortion of the image. A weighting factor corresponding with the edge zone union of the original RS image and the reconstructed one is introduced into the model. Subjective method, PSNR based objective method and the proposed objective method are used for image quality evaluation with a set of images. The results show that the proposed EWPSNR based objective image quality evaluation method is more consistent with the subjective method than PSNR based one.
DOI:10.1109/CSSE.2008.179