GMM based language identification system using robust features

In this work, we have proposed new feature vectors for spoken language identification (LID) system. The Mel frequency cepstral coefficients (MFCC) and formant frequencies derived using short-time window speech signal. Formant frequencies are extracted from linear prediction (LP) analysis of speech signal. Using these two kind of features of speech signal, new feature vectors are derived using cluster based computation. A GMM based classifier has been designed using these new feature vectors. The language specific apriori knowledge is applied on the recognition output. The experiments are carried out on OGI database and LID recognition performance is improved.     

Model for drying during fluidized-bed combustion of wet low-rank coals

An unsteady state heat conduction model with a convective boundary condition is proposed for the drying of low-rank, high-porosity coals, such as lignites, during fluidized-bed combustion. The drying front is assumed to be the receding surface of a wet core. The solution technique for this moving boundary problem is based on the heat balance integral approach with immobilization of the moving boundary by a change in space variable. The governing cubic equation describing the drying curve in dimensionless form may be solved easily by the Newton—Raphson method. The model predictions are compared with experimental data for Mississippi lignite with excellent agreement. A correlation for estimation of total drying time is proposed. The temperature profiles obtained may be used for the study of the coupled drying and devolatilization in fluidized-bed combustors. The profiles could also be of importance in the study of formation of fissures/cracks in lignites subjected to intense heating conditions encountered during fluidized-bed combustion.     

THERMAL REGENERATION OF ACTIVATED CARBONS

The thermal regeneration of activated carbons loaded with p-nitrophenol (PNP) and other aromatic compounds was studied using a thermal balance. After pyrolysis of the adsorbates in nitrogen at 700°C, the residues were gasified with oxygen at 415 to 500°C or with steam at 840 to 920°C. Residues from PNP were several times more reactive to oxygen than the base carbons and also showed greater chemisorption of oxygen. For steam gasification, only small differences between spent and fresh carbons were found.     

Characteristics and compositions ofCarissa spinarum,Leucaena leucocephala andPhysalis minima seeds and oils

The seeds and extracted oils ofCarissa spinarum (Apocynaceae), (I),Leucaena leucocephala (Leguminosae) (II) andPhysalis minima (Solanaceae) (III) were analyzed for characteristics and compositions. The seeds of I, II and III contained 22.4, 6.4 and 40.0% oil and 10.1, 27.6 and 17.9% protein, respectively. The oils of I, II and III had, respectively, iodine values 70.1, 113.5 and 122.5; saponification values 186, 188 and 189; unsaponifiable matter 5.2, 2.5 and 0.8%, and the following fatty acid compositions (area %): palmitic 12.6, 14.2, 10.5; stearic 7.6, 6.1, 8.6; oleic 72.7, 20.1, 17.3; linoleic 5.2, 53.8, 61.4; linolenic 0.9, 1.8, 0.0, and arachidic 1.0, 2.3, 0.0. II contained 1.7% lignoceric acid. III contained small amounts of hexadecenoic (0.1%), epoxy (0.6%) and hydroxy (1.5%) fatty acids.     

DPV: a taxonomy for utilizing deep learning as a prediction technique for various types of cancers detection

Multimedia Tools and Applications - Deep learning (DL) is a type of machine learning capable of processing large quantities of data to provide analytic results based on a particular...     

Low intensity visual data improvement using DWT and LWT based regularizer

Image Completion plays a vital role in compressed sensing, machine learning, and computer vision applications. The Rank Minimization algorithms are used to perform the image completion. The major problem with rank minimization algorithms is the loss of information in the recovered image at high corruption ratios. To overcome this problem Lifting wavelet transform based Rank Minimization (LwRM), and Discrete wavelet transform based Rank Minimization (DwRM) methods are proposed, which can recover the image, if the corrupted observations are more than 80%. The evaluation of the proposed methods are accomplished by Full Reference Image Quality Assessment (FRIQA) and No Reference Image Quality Assessment (NR-IQA) metrics. The simulation results of proposed methods are superior to state-of-the-art methods.

     

A novel secure solution of using mixed reality in data transmission for bowel and jaw surgical telepresence: enhanced rivest cipher RC6 block cipher

Security threats are crucial challenges that deter Mixed reality (MR) communication in medical telepresence. This research aims to improve the security by reducing the chances of types of various attacks occurring during the real-time data transmission in surgical telepresence as well as reduce the time of the cryptographic algorithm and keep the quality of the media used. The proposed model consists of an enhanced RC6 algorithm in combination. Dynamic keys are generated from the RC6 algorithm mixed with RC4 to create dynamic S-box and permutation table, preventing various known attacks during the real-time data transmission. For every next session, a new key is created, avoiding possible reuse of the same key from the attacker. The results obtained from our proposed system are showing better performance compared to the state of art. The resistance to the tested attacks is measured throughout the entropy, Pick to Signal Noise Ratio (PSNR) is decreased for the encrypted image than the state of art, structural similarity index (SSIM) closer to zero. The execution time of the algorithm is decreased for an average of 20%. The proposed system is focusing on preventing the brute force attack occurred during the surgical telepresence data transmission. The paper proposes a framework that enhances the security related to data transmission during surgeries with acceptable performance.

     

Localized swelling of Fe-Cr-Ni alloy rollers in high-temperature applications

Localized swelling has been observed in 24Cr-24Ni-Nb steel transportation rollers used in the normalizing furnace of a plate mill after prolonged service at high temperature. Due to high localized thermal and mechanical stresses, the chromia layer formed on the roller surface ruptures, exposing the roller substrate to furnace oxygen. Oxidation of second-phase carbides results in the formation of carbon monoxide at very high partial pressure. This leads to formation of voids, leading in turn to localized swelling of the roller material.     

Updating a 1950s-generation aluminum smelter

Hindalco’s aluminum electrolysis cells were initially installed in 1962, and the technology was based on 1950s-generation pots. Although Hindalco expanded its aluminums melting capacity from 20,000 tonnes per year to 175,000 tonnes per year, the basic design of the pots remained unchanged. In view of energy price increases, and to keep pace with the latest developments in aluminum smelting technology, Hindalco undertook efforts to modernize its facilities. In spite of numerous constraints, the Hindalco smelter has been able to achieve performance nearly equivalent to that of 1980s-generation pots by retrofitting new technologies. This has resulted in considerable savings in electrical energy consumption and raw materials usage.     

Pedestrian Detection and Tracking Using Deformable Part Models and Kalman Filtering

Abstract： Pedestrian detection techniques are important and challenging especially for complex real world scenes. They can be used for ensuring pedestrian safety, ADASs （advance driver assistance systems） and safety surveillance systems. In this paper, we propose a novel approach for multi-person tracking-by-detection using deformable part models in Kalman filtering framework. The Kalman filter is used to keep track of each person and a unique label is assigned to each tracked individual. Based on this approach, people can enter and leave the scene randomly. We test and demonstrate our results on Caltech Pedestrian benchmark, which is two orders of magnitude larger than any other existing datasets and consists of pedestrians varying widely in appearance, pose and scale. Complex situations such as people occluded by each other are handled gracefully and individual persons can be tracked correctly after a group of people split. Experiments confirm the real-time performance and robustness of our system, working in complex scenes. Our tracking model gives a tracking accuracy of 72.8% and a tracking precision of 82.3%. We can further reduce false positives by 2.8%, using Kalman filtering.