I Gede Pasek Suta Wijaya

Multi-pose Face Recognition Using Fusion of Scale Invariant Features

This paper presents a new multi-pose face recognition approach using fusion of scale invariant features (FSIF). The FSIF is a face descriptor representing 3D face images features which is created by fusing some scale invariant features extracted by scale invariant features transforms (SIFT) from several different poses of 2D face images. The main aim of this method is to avoid using 3D scanner for estimating any pose variations of a face image but it still have reasonable achievement compare to 3D-based face recognition method for multi-pose face recognition. The experimental results show the proposed method is sufficiently to overcame large face variability due to face pose variations.

Phonographic image recognition using fusion of scale invariant descriptor

The paper presents a pornographic image recognition using fusion of scale invariant descriptor. The pornographic image means the image contains and shows genital elements of human body having large variability due to poses, lighting, and backgrounds variations. The fusion of scale invariant descriptor that is holistic feature is employed to handle those variability problems. This holistic feature that is pose and scale invariant information of pornographic images is extracted by fusing the scale invariant descriptor of skin region of interests (ROIs) of pornographic images. The skin ROI is used to handle the large variability of pornographic images due to background variations. The main aim of this research finds a good solution for pornographic recognition system, which can be developed to limit the accessing pornographic images by teenagers and children. The experimental results show that the proposed method tends to provide high enough accuracy more than 80%, small enough FNR and FPR bout 2.77% and 28.79%, respectively. It means the proposed method is suitable to develop rejection system of pornographic images. Furthermore, these achievements are much better than the achievements of established methods. This results can be achieved because the fusion of scale invariant descriptor consists rich pornographic information representing holistic feature of pornographic images.

Fast and robust face recognition for incremental data

This paper proposes fast and robust face recognition system for incremental data, which come continuously into the system. Fast and robust mean that the face recognition performs rapidly both of training and querying process and steadily recognize face images, which have large lighting variations. The fast training and querying can be performed by implementing compact face features as dimensional reduction of face image and predictive LDA (PDLDA) as face classifier. The PDLDA performs rapidly the features cluster process because the PDLDA does not require to recalculate the between class scatter, Sb, when a new class data is registered into the training data set. In order to get the robust face recognition achievement, we develop the lighting compensation, which works based on neighbor analysis and is integrated to the PDLDA based face recognition.

Pose Invariant Color Face Recognition Based on Frequency Analysis and DLDA with Weight Score Classification

The face images mostly cover with skin color, which exists in chrominance component. That component was discharged in almost all of the previous works. In this paper, we present a method for pose invariant color face recognition, which is based on frequency analysis and DLDA with weight-score classification. The function of frequency analysis (i.e. wavelet and DCT transforms) is to extract the global facial features by selecting the dominant coefficients existing in low frequency components. In this case, the global facial features are created not only in the luminance but also in the chrominance for covering the skin color information. The weight-score is introduced in DLDA in order to reduce the overlap projected facial features. Where, the weight score, which is defined as a whole distance among the considered class and some closely classes to it, is determined by Mahalanobis distance.

Pose Invariant Face Recognition Based on Hybrid Dominant Frequency Features

Face recognition is one of the most active research areas in pattern recognition, not only because the face is a human biometric characteristics of human being but also because there are many potential applications of the face recognition which range from human-computer interactions to authentication, security, and surveillance. This paper presents an approach to pose invariant human face image recognition. The proposed scheme is based on the analysis of discrete cosine transforms (DCT) and discrete wavelet transforms (DWT) of face images. From both the DCT and DWT domain coefficients, which describe the facial information, we build compact and meaningful features vector, using simple statistical measures and quantization. This feature vector is called as the hybrid dominant frequency features. Then, we apply a combination of the L2 and Lq metric to classify the hybrid dominant frequency features to a persons class. The aim of the proposed system is to overcome the high memory space requirement, the high computational load, and the retraining problems of previous methods. The proposed system is tested using several face databases and the experimental results are compared to a well-known Eigenface method. The proposed method shows good performance, robustness, stability, and accuracy without requiring geometrical normalization. Furthermore, the purposed method has low computational cost, requires little memory space, and can overcome retraining problem.

Traffic light signal parameters optimization using particle swarm optimization

This paper proposes a traffic light signal parameters optimization using particle swarm optimization (PSO) for real road network called as Ooe Toroku road network. The main aim of this method is to find out the best traffic light signal parameters, which can solve the traffic congestion on the real road network. The traffic light signal parameters that are optimized are offset, cycles, and splits time of each node of the considered road networks. The considered real road network consists of four junctions/nodes having different time signaling models. In this research, the PSO is attached in Aimsun 6.1 simulator via application interface (API) that is provided by Aimsun 6.1 simulator. The PSO algorithm creates n-particles of traffic light signal parameters and sends them to the Aimsun 6.1 simulator to perform the simulation. The output of simulation will be used to perform the particles evaluation and updating. The experimental results show that the proposed method provides better performance than base-line method (multi-element Genetics Algorithms (ME-GA) based optimization method) which can increase the real and base-line percentage of vehicle flow by about 15.76% and 4.13% of that of real and MEGA, respectively. In addition, the PSO is faster to achieve convergence than base-line method for considered network.

Human face security system using alternative linear discriminant analysis based classifier

Like fingerprint, human face can be applied as a security system because it has almost the same characteristics as that of fingerprint, in terms of the uniqueness and non transferable. Therefore, in this paper, we design and simulate fast human face recognition for the security system. It is realized by implementing the compact features of face image as data dimensional reduction and the shifting-mean LDA as data classifier. The compact features is set of dominant frequency contents and statistical moment of face image and the shifting-mean LDA is an alternative LDA-based classifier which can avoid retraining problem of incremental data. From both off-line and real-time experimental results, the proposed method provides good enough achievements in terms of recognition rate, false rejection rate (FRR), and false acceptance rate (FAR) with requiring short time processing.

Real time face recognition engine using compact features for electronics key

In this research, a design and implementation of real-time face recognition engine using compact features vector extracted by linear binary pattern (LBP) and zoned discrete cosine transforms (DCT) analysis is proposed for electronics key. The function of LBP is to normalize the lighting variations of the input face image and the function of zoned DCT is to define the local descriptors of the face image. In order to get more compact features vector, the predictive LDA is employed for dimensional reduction. The aims of this research is to develop fast and strong face recognition that can be implemented for electronic key which will be implemented for substituting current security system (PIN and password). In addition, the recognition engine is also designed for real time face recognition which can work on hardware having limited resources such as Intel atom computer, raspberry pi, and android smart phone. The experimental results show that the proposed engine provide high enough recognition rate, and small false rejection rate (FRR) and false acceptance rate (FAR). In addition, the engine needs short processing time.

Why the alternative PCA provides better performance for face recognition

This paper presents an alternative to PCA technique, called as APCA, which uses within class scatter rather than global covariance matrix. The APCA technique produces better features cluster than does common PCA (CPCA) because it keep the null spaces which contain good discriminant information. The proposed technique achieves better performance for both recognition rate and accuracy parameters than those of CPCA when it was tested using several databases (ITS-LAB., INDIA, ORL, and FERET).

Real time face recognition using DCT coefficients based face descriptor

This paper presents an alternative real time face recognition using DCT coefficients based face descriptor. The face descriptor consists of dominant frequency content extracted by discrete cosine transforms (DCT), local features extracted by zone DCT, and shape information extracted by hu-moment. The aim of DCT coefficients based face descriptor is to obtain rich information of face descriptor which can provide good performance on real-time face recognition. In this research, dimensional size of face descriptor is decreased by using predictive linear discriminant analysis (PDLDA) and the kNN is implemented for verification. From accuracy, false negative and positive data, the proposed real time face recognition seems to provide good performances. In addition, it also needs short computational time.