PROJECT TITLE :

Backward registration based Aspect ratio similarity(ARS) For image retargeting quality assessment - 2016

ABSTRACT:

Throughout the past few years, there are numerous kinds of content-aware image retargeting operators proposed for image resizing. But, the lack of effective objective retargeting quality assessment metrics limits the any development of image retargeting techniques. Different from traditional image quality assessment (IQA) metrics, the standard degradation during image retargeting is caused by artificial retargeting modifications, and the problem for image retargeting quality assessment (IRQA) lies in the alternation of the image resolution and content, which makes it not possible to directly evaluate the quality degradation like traditional IQA. In this paper, we interpret the image retargeting during a unified framework of resampling grid generation and forward resampling. We have a tendency to show that the geometric change estimation is an economical manner to clarify the connection between the images. We tend to formulate the geometric amendment estimation as a backward registration problem with Markov random field and give a good resolution. The geometric amendment aims to supply the proof regarding how the initial image is resized into the target image. Under the guidance of the geometric amendment, we develop a novel facet ratio similarity (ARS) metric to guage the visual quality of retargeted pictures by exploiting the local block changes with a visible importance pooling strategy. Experimental results on the publicly on the market MIT RetargetMe and CUHK data sets demonstrate that the proposed ARS can predict a lot of accurate visual quality of retargeted pictures compared with the state-of-the-art IRQA metrics.