Non-Verbal (Sign Language) To Verbal Language Translator Using Convolutional Neural Network

Communication is very imperative for daily life. Normal people use verbal language for communication while people with disabilities use sign language for communication. Sign language is a way of communicating by using the hand gestures and parts of the body instead of speaking and listening. As not all people are familiar with sign language, there lies a language barrier. There has been much research in this field to remove this barrier. There are mainly 2 ways in which we can convert the sign language into speech or text to close the gap, i.e. , Sensor based technique,and Image processing. In this paper we will have a look at the Image processing technique, for which we will be using the Convolutional Neural Network (CNN). So, we have built a sign detector, which will recognise the sign numbers from 1 to 10. It can be easily extended to recognise other hand gestures including alphabets (A- Z) and expressions. We are creating this model based on Indian Sign Language(ISL). Keywords: Multi Level Perceptron (MLP), Convolutional Neural Network (CNN), Indian Sign Language(ISL), Region of interest(ROI), Artificial Neural Network(ANN), VGG 16(CNN vision architecture model), SGD(Stochastic Gradient Descent).

Stress concentration effect on deflection and stress fields of a master leaf spring through domain decomposition and geometry updation technique

Theoretical and experimental large deflection and stress analysis of a master leaf spring considering stress concentration effect of clamping is reported. The non-uniformly curved master leaf spring under three point bending subjected to moving boundaries is modeled. Geometrically nonlinear strain-displacement relations, as necessary for the theoretical analysis, are derived through visualization of physics behind the large deformation problem. An embedded curvilinear coordinate system is considered, to study the combined effects of non-uniform curvature, bending, stretching and shear deformation including cross-sectional warping. Governing equation of the non-uniformly curved beam system is derived in variational form using energy method, based on linear material constitutive relations and the derived nonlinear kinematic relations. An iterative solution scheme through successive geometry updation is developed and executed in MATLAB® software to solve the governing equation involving strong geometric nonlinearity together with complicating moving boundary effect. Experimental deflection profiles under static loading are obtained through manual image processing technique using AutoCAD® software. Whereas, strain measurements are performed using strain gauges with data acquisition system (HBM-MX840B). Comparison between the theoretical and experimental results lead towards observation on stress concentration effect due to presence of geometric discontinuity in form of a small hole in the physical system. A modified formulation is proposed using domain decomposition method incorporating effect of geometric discontinuity through an equivalent curved beam geometry of variable cross-section. The modified theoretical model is validated successfully with the experimental results, and observations on stress characteristics and effect of hole diameter to beam width ratio are made.

Evaluation of optical properties of fluorescent nanofiber using image-processing technique

Purpose The purpose of this study is evaluate the optical properties of polyacrylonitrile (PAN) nanofibers containing fluorescent agents such as fluorescent dye and carbon quantum dots (CQDs) by using image-processing technique of Fluorescence microscope image. Design/methodology/approach The fluorescence microscope image of the pure PAN, PAN/CQDs and PAN/fluorescent dye nanofibers composite was analyzed using several image-processing techniques such as color histogram, lookup table (LUT), Fourier transform, RGB profile and surface plot analysis. Findings The fluorescence microscope image indicates that the fluorescence emission of nanocomposites depends on the type of fluorescent agent. The fluorescence intensity of nanofiber containing CQDs is more than nanofiber containing fluorescent dye. Various image-processing methods provide similar results for optical property of nanocomposites. Analyzing the LUT, the blue value of CQDs/PAN nanocomposite image was significantly higher than other nanocomposites. This was confirmed by other methods such as Fourier transform, color histogram and 3D topography of the electrospun nanofibers. According to analysis of colorimetric parameters, higher negative value of b* indicates bluer color for CQDs/PAN nanofibers than other nanocomposites. The obtained results indicate that the image-processing technique can be used to evaluate the optical property of fluorescent nanocomposite. Originality/value This study evaluates the optical properties of fluorescent nanocomposites by using image-processing techniques such as Fourier transform, color histogram, RGB profiles, LUT, surface plot and histogram analysis.

Flexural Behavior of a Novel Textile-Reinforced Polymer Concrete

Textile reinforced concrete (TRC) has gained attention from the construction industry due to its light weight, high tensile strength, design flexibility, corrosion resistance, and remarkably long service life. Some structural applications that utilize TRC components include precast panels, structural repair, waterproofing elements, and façades. TRC is produced by incorporating textile fabrics into thin cementitious concrete panels. Premature debonding between the textile fabric and concrete due to improper cementitious matrix impregnation of the fibers was identified as a failure-governing mechanism. To overcome this performance limitation, in this study, a novel type of TRC is proposed by replacing the cement binder with a polymer resin to produce textile reinforced polymer concrete (TRPC). The new TRPC is created using a fine-graded aggregate, methyl methacrylate polymer resin, and basalt fiber textile fabric. Four different specimen configurations were manufactured by embedding 0, 1, 2, and 3 textile layers in concrete. Flexural performance was analyzed and compared with reference TRC specimens with similar compressive strength and reinforcement configurations. Furthermore, the crack pattern intensity was determined using an image processing technique to quantify the ductility of TRPC compared with conventional TRC. The new TRPC improved the moment capacity compared with TRC by 51%, 58%, 59%, and 158%, the deflection at peak load by 858%, 857%, 3264%, and 3803%, and the toughness by 1909%, 3844%, 2781%, and 4355% for 0, 1, 2, and 3 textile layers, respectively. TRPC showed significantly improved flexural capacity, superior ductility, and substantial plasticity compared with TRC.

Image Segmentation Using PSO-Enhanced K-Means Clustering and Region Growing Algorithms

Image segmentation is a basic image processing technique that is primarily used for finding segments that form the entire image. These segments can be then utilized in discriminative feature extraction, image retrieval, and pattern recognition. Clustering and region growing techniques are the commonly used image segmentation methods. K-Means is a heavily used clustering technique due to its simplicity and low computational cost. However, K-Means results depend on the initial centres’ values which are selected randomly, which leads to inconsistency in the image segmentation results. In addition, the quality of the isolated regions depends on the homogeneity of the resulted segments. In this paper, an improved K-Means clustering algorithm is proposed for image segmentation. The presented method uses Particle Swarm Intelligence (PSO) for determining the initial centres based on Li’s method. These initial centroids are then fed to the K-Means algorithm to assign each pixel into the appropriate cluster. The segmented image is then given to a region growing algorithm for regions isolation and edge map generation. The experimental results show that the proposed method gives high quality segments in a short processing time.

Speedy Detection Module for Abandoned Belongings in Airport Using Improved Image Processing Technique

Recently, in computer vision and video surveillance applications, moving object recognition and tracking have become more popular and are hard research issues. When an item is left unattended in a video surveillance system for an extended period of time, it is considered abandoned. Detecting abandoned or removed things from complex surveillance recordings is challenging owing to various variables, including occlusion, rapid illumination changes, and so forth. Background subtraction used in conjunction with object tracking are often used in an automated abandoned item identification system, to check for certain pre-set patterns of activity that occur when an item is abandoned. An upgraded form of image processing is used in the preprocessing stage to remove foreground items. In subsequent frames with extended duration periods, static items are recognized by utilizing the contour characteristics of foreground objects. The edge-based object identification approach is used to classify the identified static items into human and nonhuman things. An alert is activated at a specific distance from the item, depending on the analysis of the stationary object. There is evidence that the suggested system has a fast reaction time and is useful for monitoring in real time. The aim of this study is to discover abandoned items in public settings in a timely manner.