Image Quality Assessment through Contour Detection
Shuo Liu, Xiang Peng, Zheng Liu · 2019
This paper presents a study on objective image quality assessment from the segmentation perspective. Instead of modeling the image distortion or degradation process, the assessment is achieved from the end user's point of view, i.e. how the image is interpreted by segmentation operation. The accessibility of informative features in an image subject to certain degradation is evaluated by comparing the contour of input and reference images. A new quality index is defined with the F-measure in the contour mapping measure framework. The demonstrative experiments illustrate how the index works for the prepared images and the preliminary results prove the effectiveness of the proposed approach.