Conference Paper

Image quality metrics: PSNR vs. SSIM

DOI: 10.1109/ICPR.2010.579 Conference: 20th International Conference on Pattern Recognition, ICPR 2010, Istanbul, Turkey, 23-26 August 2010
Source: DBLP


In this paper, we analyse two well-known objective image quality metrics, the peak-signal-to-noise ratio (PSNR) as well as the structural similarity index measure (SSIM), and we derive a simple mathematical relationship between them which works for various kinds of image degradations such as Gaussian blur, additive Gaussian white noise, jpeg and jpeg2000 compression. A series of tests realized on images extracted from the Kodak database gives a better understanding of the similarity and difference between the SSIM and the PSNR.

  • Source
    • "If the success ratio is greater than 0, it implies that song denoising has been successful. A third possibility is to calculate the Peak Signal to Noise Ratio (PSNR), a widely used objective quality metric in image and video processing[38,39]. PSNR looks only at the peak value the signal can reach and the mean-squared error between the reference and the test signals. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Automatic recording of birdsong is becoming the preferred way to monitor and quantify bird populations worldwide. Programmable recorders allow recordings to be obtained at all times of day and year for extended periods of time. Consequently, there is a critical need for robust automated birdsong recognition. One prominent obstacle to achieving this is low signal to noise ratio in unattended recordings. Field recordings are often very noisy: birdsong is only one component in a recording, which also includes noise from the environment (such as wind and rain), other animals (including insects), and human-related activities, as well as noise from the recorder itself. We describe a method of denoising using a combination of the wavelet packet decomposition and band-pass or low-pass filtering, and present experiments that demonstrate an order of magnitude improvement in noise reduction over natural noisy bird recordings.
    Full-text · Article · Jan 2016 · PLoS ONE
  • Source
    • "Objective metrics based on SSIM or MS- SSIM [5] have become popular recently, but the basic SSIM metric does not provide significantly more information than a basic PSNR calculation. For example, the work in [6] derives the relation between PSRN and SSIM. Despite the known limitations, PSNR is commonly used to evaluate system performance and application such as tuning encoding parameters due to the simplicity and application to a narrow range of parameter changes. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The limitations of objective metrics such as PSNR in evaluating video quality are well known to experts but less know to general users. A video tool which exploits perceptual phenomena may report higher subjective quality but lower objective performance than a non-perceptual tool. This presents a problem when describing the performance of a perceptually motivated algorithm to general users. We propose a method for extending existing objective metrics to account for perceptual factors such as viewing distance, ambient contrast, etc. After describing the proposed algorithm extension, we examine the objective results of the proposal and perform subjective viewing tests to confirm the behavior of the extended objective metrics.
    Full-text · Conference Paper · Feb 2015
  • Source
    • "Once, optimum prototypes for each species determined the similarity measures that are simpler therefore faster than image texture methods, were used to score the similarity indexes of query with each prototype. The structural similarity measure (SSIM) [14- 19], feature similarity index (FSIM) [20], peak-signal-tonoise ratio (PSNR) [19], scale invariant feature transform (SIFT) [21], histogram comparison (HISTC) [22] and image spatiogram comparison (SPATC) [23] methods were used as a similarity measure. In this study, employing the SSIM and FSIM similarity measures was assessed in detailed and the others were used for comparison. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Butterflies can be classified by their outer morphological qualities, genital characters that can be obtained using various chemical substances and methods, which are carried out manually by preparing genital slides through some certain processes or molecular techniques. The aim of this study is to evaluate a computer vision and machine learning system that correctly identify butterfly species easier, faster and cheaper than traditional methods. In this study, human vision based image similarity methods were used as feature extractors, which were structural similarity measure (SSIM) method that depends on the combination of luminance, contrast and structural comparisons, and feature similarity index (FSIM) method that depends on combination of phase congruency and image gradient magnitude. First of all a prototype was determined for each species of 19 species, then for each butterfly, the SSIM and FSIM indexes were computed. The machine learning methods had achieved high accuracy rates for identification of butterfly species by these indexes, while it achieved 100% accuracy logistic linear classifier method The accuracy results of using SSIM and FSIM as a feature extraction method were compared with other similarity methods such as peak-signal-to-noise ratio, scale invariant feature transform, histogram comparison, image spatiogram comparison and texture methods.
    Full-text · Article · Jan 2015
Show more