Introduction

This is the website for our paper "Learning a Blind Image Quality Index Based on Visual Saliency Guided Sampling and Gabor Filtering", in Proc. ICIP, 2013.

Source Code

Source code can be downloaded here.

Evaluation Results

Experiments were conducted on LIVE IQA image dataset [1]. In order to compare with other methods properly, we followed the experiment strategy in [2]. We randomly selected 23 reference images associated with their distorted copies for training, and 6 reference images with their distorted copies for testing to ensure the content of the training images and the testing images do not overlap. The reported result is the median of 1000 train-test runs. Two metrics, Spearman rank-order correlation coefficients (SROCC) and Pearson linear correlation coefficients (PLCC) were used to evaluate the accuracy of all the algorithms. A value close to 1 for SROCC and PLCC indicates a good performance for quality estimation. The following table shows the experimental results of our proposed method.

Reference                

1. H.R. Sheikh, M.F. Sabir, and A.C. Bovik, "A statistical evaluation of recent full reference image quality assessment algorithms," IEEE Trans. Image Processing, vol. 15, pp. 3440-3451, Nov. 2006.

2. P. Ye and D. Doermann, "No-reference image quality assessment using visual codebooks," IEEE Trans. Image Processing, vol. 21, pp. 3129-3138, 2012.