The Design of High-Level Features for Photo Quality Assessment
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1
A no-reference objective image sharpness metric based on the notion of just noticeable blur (JNB)
IEEE Transactions on Image Processing
No-reference quality assessment using natural scene statistics: JPEG2000
IEEE Transactions on Image Processing
| Hi-index | 0.00 |
In this paper, we propose a novel framework for blind image quality evaluation. Unlike the common image quality measures evaluating compression or transmission artifacts this approach analyzes the image properties common to non-ideal image acquisition such as blur, under or over exposure, saturation, and lack of meaningful information. In contrast to methods used for adjusting imaging parameters such as focus and gain this approach does not require any reference image. The proposed method uses seven image degradation features that are extracted and fed to a classifier that decides whether the image has good or bad quality. Most of the features are based on simple image statistics, but we also propose a new feature that proved to be reliable in scene invariant detection of strong blur. For the overall two-class image quality grading, we achieved ≈ 90% accuracy by using the selected features and the classifier. The method was designed to be computationally efficient in order to enable real-time performance in embedded devices.