Iris recognition is one of the most representative identification technologies in biometric recognition, which widely used various fields. Recently, many deep learning methods have been owing to their advantages such as automatic learning, high accuracy, and strong generalization ability. The convolutional neural network (CNN) mainstream method image processing domains, but it has poor anti-noise capacity classification easily affected by slight disturbances. CNN also needs a large number...

Graphical User Interface (GUI) is ubiquitous in almost all modern desktop software, mobile applications and online websites. A good GUI design crucial to the success of software market, but designing a which requires much innovation creativity difficult even well-trained designers. In addition, requirement rapid development also aggravates designers' working load. So, availability various automated generated GUIs can help enhance personalization specialization as they cater taste different...

Gabor filters are widely utilized to detect iris texture information in several state-of-the-art recognition systems. However, the proper kernels and generative pattern of features need be predetermined application. The traditional empirical shallow encoding ways incapable dealing with such complex variations imaging including illumination, aging, deformation, device variations. Thereby, an adaptive filter selection strategy deep learning architecture presented. We first employ particle...

Human posture recognition is very critical in human computer interaction studies. With the release of Microsoft Kinect sensor, there has been an increasing interest using for vision based as user's skeleton information can be precisely inferred from depth images generated by Kinect. In this paper we proposed a novel method which automatically identify any user-defined postures. Skeleton image was utilized to generate 9 features representing specific body parts such forearm, thigh, etc. These...

Unlike score level fusion, feature fusion demands all the features extracted from unimodal traits with high distinguishability, as well homogeneity and compatibility, which is difficult to achieve. Therefore, most multimodal biometric research focuses on whereas few investigate fusion. We propose a face–iris recognition method based build special two-dimensional-Gabor filter bank extract local texture face iris images, then transform them by histogram statistics into an energy-orientation...

In this paper, we propose three discriminative feature selection strategies and weighted subregion matching method to improve the performance of iris recognition system. Firstly, introduce process extraction representation based on scale invariant transformation (SIFT) in detail. Secondly, are described, which orientation probability distribution function (OPDF) strategy delete some redundant keypoints, magnitude (MPDF) reduce dimensionality element, compounded combined OPDF MPDF further...

Absfrod -A simple ondproeficol iris acquisition syxtem is established by scporcltimg light unif for rceognitiorr Based on directions of gradient vectors in image. edges ore obfainel Thisprrper inlnodmedNor- mol Line concept, improved operations fhe trudifiond Hough fmnslofion with the deJnition andgiven a fasf locafingproceeding. The U- perimenf results proved validity ond e~riectne~~ tho system and lowtion algorifhm. enhancement algorithm should not suppose that it easy to extract local...

A multimodal biometric system has been considered a promising technique to overcome the defects of unimodal systems. We have introduced fusion scheme gain better understanding and method for face-iris-fingerprint system. In our case, we use particle swarm optimization train set adaptive Gabor filters in order achieve proper basic functions each modality. For closer analysis texture information, two different local features modality are produced by corresponding coefficients. Next, all...

Although there have been many successful multiclassification models in the field of image recognition recent years, adding new classes to trained remains a research focus. In this paper, we present method based on Hadamard error correction output code and convolutional neural network solve problem. Compared with other methods, it not only makes use excellent processing performance but also combines code's characteristics simple construction adaptability arbitrary categories. The are shown...

Design of small interference RNA (siRNA) is one the most important steps in effectively applying (RNAi) technology. The current siRNA design often produces inconsistent results, which fail to reliably select with clear silencing effects. We propose that when designing siRNA, should consider mRNA global features and near siRNA-binding site local features. By a linear regression study, we discovered strong correlations between inhibitory efficacy both neighboring This paper shows that, on...

Current iris recognition technology faces practical difficulties. For example, due to the unsteady morphology of a heterogeneous generated by variety different devices and environments, traditional processing methods statistical learning or cognitive for single source are not effective. The existing data set size situational classification constraints make it difficult meet requirements under deep framework. Therefore, this paper proposes method based on an entropy feature lightweight neural...

Abstract Gabor descriptors have been widely used in iris texture representations. However, fixed basic functions cannot match the changing nature of diverse datasets. Furthermore, a single form feature overcome difficulties recognition, such as illumination variations, environmental conditions, and device variations. This paper provides multiple local representations their fusion scheme based on support vector regression (SVR) model for recognition using optimized filters. In our system,...

As an important part of human computer interaction (HCI) system, posture identification has been extensively studied over last years. Recently, Microsoft Kinect Sensor become a hot spot for because it is efficient in acquiring body joint location information. In this study, based on Kinect, we proposed framework view independent identification. framework, viewpoint rotation transformation was performed original skeleton data and then total 9 features were extracted building SVM classifier....

This paper proposes a secondary iris recognition based on local features. The application of the energy-orientation feature (EOF) by two-dimensional Gabor filter to extraction goes before first threshold similarity, which sets whole database into two categories—a correctly recognized class and be recognized. Therefore, former are accepted latter transformed histogram achieve an (EOHF), is followed second with chi-square distance. experiment has proved that proposed method, because its higher...

For building a new iris template, this paper proposes strategy to fuse different portions of based on machine learning method evaluate local quality iris. There are three novelties compared previous work. Firstly, the normalized segmented is divided into multitracks and then each track estimated individually analyze recognition accuracy rate (RAR). Secondly, six evaluation parameters adopted texture information track. Besides, particle swarm optimization (PSO) employed get weights these...

Current segmentation methods have limitations for multi-source heterogeneous iris since differences of acquisition devices and environment conditions lead to images greatly varying quality from different datasets. Thus, algorithms are generally applied distinct Meanwhile, deep-learning-based models occupy more space take a long time. Therefore, lightweight, precise, fast network model, PFSegIris, aimed at the is proposed by us. First, feature extraction modules designed were used fully...

In this paper, the research focuses on constrained iris under different acquisition states, and trains mixed data without manual label division. A multi-state multi-classification recognition method based convolutional neural network fusion statistical cognitive learning is proposed. The process divided into an image processing module, a classification module result output module. converts images tags through network. By combining characteristics of learning, features same category to...

Iris images captured in less-constrained environments often suffer from adverse noise, challenging many existing segmentation algorithms. In this paper, we propose an efficient Hybrid Transformer U-Net (HTU-Net) to address dilemma. Unlike previous studies that only focus on utilizing popular CNN technology predict iris masks accurately, HTU-Net can simultaneously obtain and parameterized pupillary limbic boundaries by a multi-task network, further enabling CNN-based be applied any regular...