Quality Assessment of Enhanced Images

Abstract

AbstractIn the field of image processing, image enhancement and quality assessment (QA) techniques have received much attention due to their great potential in practical applications, such as object detection, recognition, and so on. Image enhancement technique has the ability to alter the visual perceived quality of images, and it is generally accepted that the enhanced images have better visual quality than the original images. However, enhancement techniques usually introduce noise and artifacts into the visual attention regions of images, resulting in over-enhancement. Therefore, it has been a hot issue in recent years to optimize the model structures and parameters to realize appropriate enhancement on the basis of enhanced image QA. For the QA of enhanced images, this chapter first constructs the contrast-changed image QA database and introduces the reduced-reference QA methods that are based on phase congruency and histogram statistics. Second, it shows the no-reference QA methods based on the fusion of non-structural information, sharpness and naturalness, and feature extraction and regression. Third, it presents the automatic contrast enhancement technique based on evaluation criteria guidance. In the end, the QA methods introduced in this chapter are validated on relevant databases, and the necessity of constructing efficient enhanced image QA methods is stated.