Image Retargeting Quality Assessment

Matlab implementation of ARS and MLF image retargeting quality assessment measures based on the following three papers.

• Aspect Ratio Similarity (ARS) for Image Retargeting Quality Assessment. ICASSP 2016
• Backward Registration-Based Aspect Ratio Similarity for Image Retargeting Quality Assessment. TIP 2016
• Multiple-Level Feature-Based Measure for Retargeted Image Quality. TIP 2018

The code has been tested on the Windows 10 64-bit OS. To run the code, you need to prepare the MIT RetargetMe dataset first.

• ARS_code is the implementation of ARS measure. You can run MIT_ARS_main.m to obtain the results. If the mex files are incompatible, run the COMPUTE_MEX.m to update the existing mex files. It may take around 1.2 hours on Win 10 (i7-6700 @3.4GHz and 16GB ram). On Xeon processors, BWRegistration may output slightly different matching results and lead to inconsistent prediction performance compared with that reported in the papers. In this case, you can use the calculated All_XX and All_YY to replace the backward registration results.
• MLF_code is the implementation of MLF measure. You can run MIT_MLF_main.m to obtain the results. The MLF_code is dependent on the ARS_code, and you need to be able to run ARS_code at first. It may take around 2.1 hours on Win 10 (i7-6700 @3.4GHz and 16GB ram).