In this paper, we propose an age-group classification algorithm using Histograms of Oriented Gradients (HOG) as the face description. The proposed method classifies subjects into four different age groups. process system is divided three main stages: pre-processing, feature extraction and classification. work, use Iranian Face Database (IFDB) [1] since actual are determined in database. IFDB contains images with span from 1 to 85 years. After HOG features faces extracted then, classified...

In this paper, we use centroid distance and axis of least inertia method for plant leaf classification. For propose the RGB (Red, Green, Blue) image are converted to binary image. Then, Canny operator is applied recognize edges before thinning edges. After that, boundary traced sample shape. Sampling helps us avoid time-consuming computations. We compute these points sampling from line. By selecting a fixed start point normalizing distances, proposed shown be invariant transformations...

Face recognition systems perform accurately in a controlled environment, but an unconstrained environment dramatically degrades their performance. In this study, novel pose‐invariant face system is proposed based on the occlusion free regions. This method utilises gallery set of frontal images and can handle large pose variations. For 2D probe image with arbitrary pose, head first obtained using robust estimation method. Then, normalised by 3D modelling from single input image. consequence,...

This paper proposes a novel human action recognition approach which represents each video sequence by cumulative skeletonized images (called CSI) in one cycle. Normalized-polar histogram corresponding to CSI is computed. That the number of pixels located certain distance and angles normalized circle. Using hierarchical classification two levels, recognized. In first level, course performed with whole bins histogram. second more similar actions are examined again employing special fine...

The aim of this paper is to investigate the performance time delay neural networks (TDNNs) and probabilistic (PNNs) trained with nonlinear features (Lyapunov exponents Entropy) on electroencephalogram signals (EEG) in a specific pathological state.For purpose, two types EEG (normal partial epilepsy) are analyzed.To evaluate classifiers, mean square error (MSE) elapsed each classifier examined.The results show that TDNN 12 neurons hidden layer result lower MSE training about 19.69...

Accurate pupil segmentation is the first and most important step for an iris recognition system. Current methods are based on fitting a model such as circle or ellipse to find detect pupil, while these don't have sufficient accuracy sensitive specular spot reflection. In this paper, we utilize optimized color mapping increase of segmentation, regardless its shape (circular elliptic), removing effects The optimum can be established by iterative minimization algorithm similar...

Abstract Efficient compression techniques are required for animated mesh sequences with fixed connectivity and time‐varying geometry. In this paper, we propose a key‐frame‐based technique three‐dimensional dynamic compression. First, key‐frames extracted from the sequence. Extracted then linearly combined using blending weights to predict vertex locations of other frames. These play key role in proposed algorithm because prediction performance number greatly depend on these weights. We...

Head pose estimation has many applications in the field of computer vision and it is a useful part pose-invariant face recognition. In this paper, we propose novel method to estimate head (yaw pitch rotations) based on fuzzy systems by facial geometric features. Firstly, seven certain points are selected face. These includes some main properties. They all visible even for large variations. Since no point mouth region, obviously insensitive expression. By these points, ratios angles computed...

This paper propose a method to 3D models categorization based on geometric features from face and vertex of any model using probabilistic neural network. For classification, we use histogram two variables, i.e., the angle between normal vector object surface point that connect shape origin point; distance origin. Also, for better separability different models, Euclidean pairs points is used. The most advantage present it leads reduce feature dimension consequently computational cost in...

Integrated region-based segmentation using color components and texture features with prior shape knowledge Segmentation is the art of partitioning an image into different regions where each one has some degree uniformity in its feature space. A number methods have been proposed blind them. It uses intrinsic features, such as pixel intensity, texture. However, virtues, like poor contrast, noise occlusion, can weaken procedure. To overcome them, object interest to be incorporated a top-down...

With a growing emphasis on human identification, iris recognition as biometric identification has recently received increasing attention. Feature vectors are extracted from templates and used for classification purpose. But efficiency of operation depends exclusivity feature vectors. We have improved features by using new filter bank applying locally Principle Independent component analysis features. Simulation results show improvement decreasing false match rate in matching level.

Object removal in image and video sequences has attracted many researchers processing. In this paper, a novel method is proposed to separate moving objects from background based on combination of Kalman filtering subtraction. Also, completion algorithm remove the extracted objects. This uses temporal information frames complete holes appeared after object removal. The associated fixed model are removed filled by exemplar texture synthesis. Experimental results demonstrate privilege...