Pseudo no reference image quality metric using perceptual data hiding

Abstract

Image quality assessment have been extensively studied during this past few decades. It is obviously very important to provide a mean to judge an image's quality without having to ask to human observers for a subjective image quality evaluation. Many computer softwares have been build in this aim. This is called objective quality assessment. Such metrics are usually of three kinds, they may be Full Reference (FR), Reduced Reference (RR) or No Reference (NR) metrics. We focus here on a new technique which recently appeared in quality assessment metrics: data-hiding-based image quality metric. Regarding the amount of data to be transmitted for quality assessment purpose, this latter is placed in between RR and NR metrics. A little overhead due to the embedded watermark is added to the image. A perceptually weighted watermark is embedded into the host image, and an evaluation of this watermark leads to assess the host image's quality. In such context, the watermark robustness is crucial. The watermark must resist to most attacks, but it must also be degraded along with the image distortion. Our work is compared to existing metrics in terms of the correlation (et de RMSE ?) with subjective assessment and in terms of data overhead induced by the mark. * http://www.its.bldrdoc.gov/vqeg/projects/rrnr-tv/RRNR-tv _draft_v1_7g.doc † The combinaison is : Q(α) = α * M hf + (1 − α) * M mf , with α lies between 0 and 1.

Evaluation of Image Quality Assessment Metrics: Color Quantization Noise

“…The entire database A57 from Cornell University [5] consists of three reference images distorted by compression distortion, Gaussian blur, and Gaussian white noise. IVC database [3,22] contains 235 distorted images generated from four distortion types JPEG, JPEG2000, LAR coding, and Blurring. In LIVE database [31,30] there are twenty nine reference images distorted with compression distortions, Gaussian blur, White noise, and fast-fading to produce 779 test images.…”

No-Reference Quality Metric of Blocking Artifacts Based on Color Discontinuity Analysis

“…In this section, four image quality databases are employed to evaluate the performance of the proposed method, including LIVE [8], [9], MICT [10], IVC [11], [12] and CSIQ [13], [14]. The LIVE database contains 29 original images and 779 distorted images.…”

No-reference blur image quality assessment based on Simulated Annealing and General Regression Neural Network

“…Then, train the SA -GRNN model. We use the CSIQ [13] (Categorical Image Quality Database), TID2008 [14,15] (Tampere Image database 2008) and IVC [16] (Image and Video Communications) for testing. Test results are shown in table 3 below.…”