A new nonequilibrium radiation model is described that predicts the ultraviolet spectrum of hydroxyl (OH) from the electronic transition A ! X for ow conditions corresponding to the bow-shock ultraviolet 2 ight experiment. Unlike previous studies, the new model includes the direct analysis of the electronically and vibrationally excited states, OH(A; v 0 = 0; 1; 2), of OH in the ow eld simulations. The ow eld is analyzed using the direct simulation Monte Carlo method. Results are presented for the altitude range of 80-100 km, where the Knudsen number varies from 0.036 to 1.3. The computation uses algorithms that improve the numerical resolution of the excited states that typically occur at mole fractions of 10 ¡ 15 . The collisional transfer of rotational and vibrational energies of the electronic state, OH(A), are also studied in detail. It is demonstrated that the usual assumption made in continuum radiation models, that the rotational and vibrational temperatures of the electronically excited state are the same as those of the ground state of the bulk ow, fails. An important improvement is achieved in the spectral prediction using the new nonequilibrium radiation model, and good agreement is obtained between ight data and emission predictions over a range of altitudes. NomenclatureA = excited electronic state designation or reaction rate constant, m 3 molecule ¡1 s ¡1 c = speed of light, m s ¡1 E = activation energy or energy level of state, J F 0 = rotational energy, m ¡1 G 0 = vibrational energy, m ¡1 g = statistical weight H = altitude, km h = Planck's constant, 6:625 £ 10 ¡34 , J s J 0 = rotational quantum number K = rate coef cient, m 3 molecule ¡1 s ¡1 k = Boltzmann constant, 1:38 £ 10 ¡23 , J K ¡1 M = a third body species N = species number of particles or temperature exponent in rate coef cient expression n = species number density, m ¡3 Q = partition function T r = rotational temperature, K T t = translational temperature, K T v = vibrational temperature, K t = time, s V = cell volume, m 3 W = particle weight X = ground electronic state designation or mole fraction µ = vibrational characteristic temperature, K º = frequency, Hz ½ = mass density, kg m ¡3 ¿ = lifetime or vibrational relaxation time, s Subscripts e = electronic index i = particle index r = rotation index s = species index t = translation index u = electronic upper state index and Aerospace Engineering. † Associate Professor, Department of Mechanical and Aerospace Engineering. Member AIAA. v = vibration index v 0 = upper vibrational state x = colliding particle 0; 1; 2 = excited states index 1 = initial condition Superscripts diss = dissociation ex = exchange form = formation qu = quenching [ ] = concentration annotation

Ultraviolet emissions radiated by hydroxyl (OH ) are computed for hypersonic nonequilibrium ow conditions corresponding to the Bow-Shock Ultra-Violet-2 ight experiment. The ow eld is analyzed using the direct simulation Monte Carlo method. These computations include direct analysis of the electronically excited state of hydroxyl. Ultraviolet emission is estimated using a nonequilibrium radiation code. New algorithms are described that improve the numerical resolution of the excited state that occurs at number densities as low as 10 2 3 cm 2 3 . Results are presented for the altitude range from 80 to 100 km. It is shown that the high-altitude emission is sensitive to modeling of the interaction of the gas with the vehicle surface. Sensitivity of emission predictions to freestream concentrations of hydrogen-bearing species is also considered. It is found that the quasi-steady-state assumption often employed in the nonequilibrium radiation code is invalid at high altitude. Comparison of the predicted values for peak OH emission with ight measurements indicates good agreement. Detailed comparisons of the spectra, however, indicate that the simulations fail to include strong nonequilibrium effects observed in the measured data. Nomenclature A= excited electronic state designation or reaction rate constant, m 3 mol 2 1 s 2 1 E = activation energy or electronic energy level, J g = statistical weight h = Planck's constant, 6.625 3 10 2 34 , J s h = altitude, m k = Boltzmann's constant, 1.38 3 10 2 23 , J K 2 1 k = rate coef cient, m 3 mol 2 1 s 2 1 M = third body species N = species number of particles, temperature exponent in rate coef cient expression n = species number density, m 2 3 t = time, s T e = electronic temperature, K T r = rotational temperature, K T t = translational temperature, K T v = vibrational temperature, K V = cell volume, m 3 W = particle weight X = mole fraction Z = distance from the body, m n = frequency, Hz r = mass density, kg m 2 3 t = lifetime, s Subscripts diss = dissociation ex = exchange form = formation qu = quenching [ ] = concentration annotation Superscripts e = electronic index i = particle index r = rotation index s = species index t = translation index v = vibration index 0, 1 = excited states index = initial condition

The rapid growth of video consumption and multimedia applications has increased the interest of the academia and industry in building tools that can evaluate perceptual video quality. Since videos might be distorted when they are captured or transmitted, it is imperative to develop reliable methods for no-reference video quality assessment (NR-VQA). To date, most NR-VQA models in prior art have been proposed for assessing a specific category of distortion, such as authentic distortions or traditional distortions. Moreover, those developed for both authentic and traditional distortions video databases have so far led to poor performances. This resulted in the reluctance of service providers to adopt multiple NR-VQA approaches, as they prefer a single algorithm capable of accurately estimating video quality in all situations. Furthermore, many existing NR-VQA methods are computationally complex and therefore impractical for various real-life applications. In this paper, we propose a novel deep learning method for NR-VQA based on multi-task learning where the distortion of individual frames in a video and the overall quality of the video are predicted by a single neural network. This enables to train the network with a greater amount and variety of data, thereby improving its performance in testing. Additionally, our method leverages temporal attention to select the frames of a video sequence which contribute the most to its perceived quality. The proposed algorithm is evaluated on five publicly-available video quality assessment (VQA) databases containing traditional and authentic distortions. Results show that our method outperforms the state-of-theart on traditional distortion databases such as LIVE VQA and CSIQ video, while also delivering competitive performance on databases containing authentic distortions such as KoNViD-1k, LIVE-Qualcomm and CVD2014.

No abstract