The field of image processing generally deals with signals that are meant for human consumption, such as images or videos over the Internet. An image may go through many stages of processing before being presented to a human observer, and each stage of processing may introduce distortions that could reduce the quality of the final display. For example, images are acquired by camera devices that may introduce distortions due to optics, sensor noise, color calibration, exposure control, camera motion etc. After acquisition, the image may further be processed by a compression algorithm that reduce the bandwidth requirements for storage or transmission.

Image and video Quality Assessment (QA) algorithms may be classified into three broad categories:

->Full-Reference (FR) QA methods, in which the QA algorithm has access to a 'perfect version' of the image or video against which it can compare a 'distorted version'. The 'perfect version' generally comes from a high-quality acquisition device, before it is distorted by, say, compression artifacts and transmission errors. However, the reference image or video generally requires much more resources than the distorted version, and hence FR QA is generally only used as a tool for designing image and video processing algorithms for in-lab testing, and cannot be deployed as an application.

->No-Reference (NR) QA methods, in which the QA algorithm has access only to the distorted signal and must estimate the quality of the signal without any knowledge of the 'perfect version'. Since NR methods do not require any reference information, they can be used in any application where a quality measurement is required. However, the price paid for this flexibility is in terms of the ability of the algorithm to make accurate quality predictions, or a limited scope of the NR QA algorithm (such as NR QA for JPEG images only etc.).

->Reduced-Reference (RR) QA methods, in which partial information regarding the 'perfect version' is available. A side-channel (called an RR channel) exists through which some information regarding the reference can be made available to the QA algorithm. RR QA algorithms use this partial reference information to judge the quality of the distorted signal.

PSNR (Peak Signal to Noise Ratio) formula is available. U can try this quality metric along with the SSIM.

Dear Nayana,

Here are the other parameters ,

MSE

RMSE

SSIM

BC

Contrast

Entropy

Luminance masking

Error Pooling,etc.

However there are tow types of Images ( HSV Models and Theoretical Models ) . You can use any of the above metrics as u wish!!!

Thank you all for the inputs. I think SSIM helps my cause.There is one metric that I am really curious about PSPNR (Peak signal to Perceived Noise ratio), all I know is that to it is similar to PSNR , the only difference being for PSPNR, we need to calculate weighted mean squared error based on JND (Just noticeable differences). However, that is the point where I am struck at. I am not aware of the concept of Just Noticeable Difference and what is it exactly. Anyone with ideas regarding this?

The JND is the difference in stimulus intensity (abscissa) between the points that correspond to levels of 75% and 50% (ordinate, usual values) on a psychometric function. IMO, the article at http://dx.doi.org/10.3758/BF03194544 is a good start on psychometric functions (although it does not deal directly with JNDs).

Hello Nayana Parashar ,

here i have mentioned some Image Quality Metrics

Sampling Rate

Sharpness

Noise

Color Reproduction

Tonescale Reproduction

Exposure

Exposure Uniformity

Target Interval

Overall image quality is dependent upon a number of individual quality attributes:

Is the sampling rate acceptable?

Is the sharpness acceptable?

Is the noise acceptable?

Is the tonescale reproduction acceptable?

Is the color reproduction acceptable?

Is the exposure level acceptable?

Is the exposure uniformity acceptable?

An image that meets all of these criteria is likely to be judged as having good overall quality.

To measure acceptability of each of these quality attributes, it is necessary to have one or more specific quality metrics associated with each attribute. For example, the attribute of noise can be measured using a standard deviation metric or a related metric such as peak signal-to-noise ratio (PSNR).

Please refer the following paper

"A non-linear image denoising method for salt-&-pepper noise removal using Fuzzy-Based approach"

Psnr, ssim and vqm