Modeling Image Quality

Modeling Image Quality

Gianluigi Ciocca, Silvia Corchs, Francesca Gasparini, Raimondo Schettini
Copyright: © 2015 |Pages: 11
DOI: 10.4018/978-1-4666-5888-2.ch590
(Individual Chapters)
No Current Special Offers

Chapter Preview



Different definitions of quality are found in the literature. It has been defined as the “totality of characteristics of a product that bear on its ability to satisfy stated or implied needs” by the International Organization of Standardization; “fitness for (intended) use” Juran (1988); “conformance to requirement” Crosby (1979); “user satisfaction” Wayne (1983). These definitions and their numerous variants could fit digital IQ as suggested by the Technical Advisory Service for Images: “The quality of an image can only be considered in terms of the proposed use. An image that is perfect for one use may well be inappropriate for another.” According to the International Imaging Industry Association white paper, IQ is the “perceptually weighted combination of all visually significant attributes of an image when considered in its marketplace or application.” We must, in fact, consider the application domain and expected use of the image data (thumbnailing, study, preservation, recognition task, etc…). Different properties contribute to define image quality and different models have been proposed in the literature. De Ridder and Endrikhovski (2002) proposed the Fidelity-Usefulness-Naturalness (FUN) IQ model that assumes the existence of three major dimensions: Fidelity, Usefulness and Naturalness.

  • Fidelity is the degree of apparent match of the image with the original (see Figure 1). Ideally, an image having the maximum degree of Fidelity should give the same impression to the viewer as the original. Genuineness and faithfulness are sometimes used as synonyms of Fidelity.

  • Usefulness is the degree of apparent suitability of the image with respect to a specific task. In many application domains, such as medical or astronomical imaging, image processing procedures can be applied to increase the image usefulness. An example of image usefulness is shown in Figure 2. The image to the left may be accurate with respect to the original object but the image to the right is more usable in an OCR (Optical Character Recognition) application. The enhancement processing steps have an obvious impact on Fidelity.

  • Naturalness is the degree of apparent match of the image with the viewer's internal references (see Figure 3). This attribute plays a fundamental role when we have to evaluate the quality of an image without having access to the corresponding original. Naturalness also plays a fundamental role when the image to be evaluated does not exist in reality, such as in virtual reality domains.

Figure 1.

Image exhibiting low fidelity: a) Very saturated colors; b) Green/bluish color cast


Key Terms in this Chapter

Full-Reference (FR) Metric: Objective quality assessment that performs a direct comparison between the image under test and a reference or “original” in a properly defined image space.

No-Reference (NR) Metric: Objective quality assessment that assumes that quality can be determined without a direct comparison between the original and the processed images.

Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR): Pixel-based FR metrics.

Mean Opinion Score (MOS): Subjective quality assessment score obtained during psycho-visual experiments.

Fidelity, Usefulness and Naturalness, Aesthetic and Content: Dimensions of image quality space.

Artifacts: Defects of digital images such as noise, blurriness, blockiness, etc.

Reduced-Reference (RR) Metric: Objective quality assessment that lies between FR and NR metrics and designed to predict perceptual IQ with only partial information about the reference image.

Image Quality Assessment (IQA): The process of estimating the adequacy of an image with respect to a given task.

Complete Chapter List

Search this Book: