Predicting the Quality of Fused Long Wave Infrared and Visible Light Images

Abstract

The capability to automatically evaluate the quality of long wave infrared (LWIR) and visible light images has the potential to play an important role in determining and controlling the quality of a resulting fused LWIR-visible light image. Extensive work has been conducted on studying the statistics of natural LWIR and visible images. Nonetheless, there has been little work done on analyzing the statistics of fused LWIR and visible images and associated distortions. In this paper, we analyze five multi-resolution-based image fusion methods in regards to several common distortions, including blur, white noise, JPEG compression, and non-uniformity. We study the natural scene statistics of fused images and how they are affected by these kinds of distortions. Furthermore, we conducted a human study on the subjective quality of pristine and degraded fused LWIR-visible images. We used this new database to create an automatic opinion-distortion-unaware fused image quality model and analyzer algorithm. In the human study, 27 subjects evaluated 750 images over five sessions each. We also propose an opinion-aware fused image quality analyzer, whose relative predictions with respect to other state-of-the-art models correlate better with human perceptual evaluations than competing methods. An implementation of the proposed fused image quality measures can be found at https://github.com/ujemd/NSS-of-LWIR-and-Vissible-Images. Also, the new database can be found at http://bit.ly/2noZlbQ.

Blind Image Quality Assessment for Super Resolution via Optimal Feature Selection

“…Content may change prior to final publication. We conducted our human study following a similar procedure as described in [24], [25] and [26]. The set up was a single stimulus categorical rating [27].…”

“…It has been observed that natural images, under certain transformations, such as bandpass processing, or the removal of the lowest spatial frequency [32] strongly tend towards probability distributions that can be effectively captured using several (but ultimately equivalent) parametric density models. The Generalized Gaussian Distribution (GGD) and the Asymmetric Generalized Gaussian Distribution (AGGD) are examples of such statistical models that have been widely used in previous IQA studies [33], [34], [24], [26]. The GGD is defined as…”

Evaluating virtual image quality using the side-views information fusion and depth maps

“…We test the Gaussianity of the residual differences be tween the DMOS and 3D-IQA predictions after non-linear mapping, and use the F-statistic to compare the variance of the two distributions [76]. In particular the F-test is used to check if, under Gaussian distribution hypothesis, the residuals of two quality metrics being evaluated come from the same distribution and therefore are statistically indistinguishable [45,76,77]. The ratio between variances of the resid Farid et al Information Fusion xxx (2017) xxx-xxx uals of the two quality metrics is computed and compared with the F-ratio to determine their significance.…”

Studying the Statistics of Natural X-ray Pictures

“…NSS models of LWIR images have proven quite useful for a variety of visual tasks, such as IR image quality prediction, VL and IR image discrimination, analysis of IR-specific distortions (such as the halo effect), and thermal nonuniformities. More recently, the statistics of fused LWIR and VL images were analyzed, and a NSS-based algorithm for fused image quality assessment (IQA) was developed [22]. With this motivation, we seek to understand and model the NSS of X-ray images toward improving the solutions to many X-ray imaging problems.…”